Public Service Alliance of Canada v. Treasury Board

T.D. 7/98

Decision rendered on July 29, 1998

THE CANADIAN HUMAN RIGHTS ACT

(R.S.C., 1985, c. H-6 (as amended))

HUMAN RIGHTS TRIBUNAL

BETWEEN:

PUBLIC SERVICE ALLIANCE OF CANADA

Complainant

Commission

Respondent

- and -

CANADIAN HUMAN RIGHTS COMMISSION

- and -

TREASURY BOARD

______________________________________________________________________________

DECISION OF THE TRIBUNAL - PHASE II

Tribunal: Donna Gillis, Chairperson

Norman Fetterly, Member

Joanne Cowan-McGuigan, Member

Appearances: Andrew Raven

Counsel for the Public Service Alliance of Canada Rosemary Morgan and René Duval

Counsel for the Canadian Human Rights Commission

Duff Friesen, Lubomyr Chabursky and Deborah Smith

Counsel for Treasury Board

Location of Hearing: Ottawa, Ontario

Reference: T.D. 2/96

February 15, 1996

TABLE OF CONTENTS

I. INTRODUCTION 1

A. The JUMI Study 1

B. History of Section 11 Complaints 3

C. The Commission's Investigation Process for Complaints Under Section 11 of the Canadian

Human Rights Act 5

D. Expert Testimony 11

II. ISSUES 12

III. SUBMISSIONS OF THE PARTIES 14

A. The Respondent 14

B. The Commission 18

C. The Alliance 22

IV. WAGE ADJUSTMENT METHODOLOGY 26

A. The Commissions's Methodology - Level-To-Segment 27

B. The Alliance's Methodology - Level-to-Composite 40

C. The Respondent's Methodology 46

(i). The Whole Group Methodology 46

(ii). Adverse Inference 58

V. THE EMPLOYER'S CLASSIFICATION SYSTEM 63

VI. SECTION 11 OF THE CANADIAN HUMAN RIGHTS ACT 70

A. Systemic Discrimination 70

(i). Concept of Causality 73

History of Section 11 of the Canadian Human Rights Act 78

B. Prima Facie Case of Discrimination 80

(i). Concept of "Equal Value" 85

(ii). Principle of "Equality" 88

VII. THE EQUAL WAGES GUIDELINES 95

A. History of the Equal Wages Guidelines 95

B. Sections 12 and 13 of the Equal Wages Guidelines 100

(i). Occupational Groups 100

C. Section 14 of the Equal Wages Guidelines 106

D. Validity of Section 14 of the Equal Wages Guidelines 112

E. Section 15 of the Equal Wages Guidelines 121

(i). Direct and Indirect Comparisons 125

F. Interpretation of Section 15 of the Equal Wages Guidelines 132

VIII. SELECTION OF WAGE ADJUSTMENT METHODOLOGY 134

A. Choice of Methodology 134

B. Ratcheting 156

IX. REGIONAL RATES 161

X. REMEDIES 167

A. Retroactivity 167

B. Method and Calculation of Payment 187

C. Interest 189

D. Hurt Feelings/Special Compensation 195

E. Costs 199

XI. ORDERS 202

APPENDIX A - GLOSSARY OF TERMS

I. INTRODUCTION

A. The JUMI Study

1. The Clerical and Regulatory (CR) group of the Federal Public Service filed a complaint with

the Canadian Human Rights Commission (the "Commission") on December 19, 1984, alleging

discrimination under ss. 7 and 11 of the Canadian Human Rights Act, R.S.C., 1985, c. H-6 (the

"Act"). The complaint alleges its members are performing work of equal value to members of

the predominantly male Program Administration (PA) group. The complaint, (Exhibit HR-10),

reads in part:

We allege that members of the predominantly female Clerical and Regulatory

group of the federal public service who are performing work of a value equal to

members of the predominantly male Program Administration group have been,

since the creation of the groups, and are being paid lower wages for that work in

contravention of Sections 7 and 11 of the Canadian Human Rights Act.

2. In March 1985, the Government initiated a program of proactive measures with respect to the

identification and elimination of sex-biased pay in the Federal Public Service. The Government

announced those measures on March 8, 1985 indicating "...[t]his government intends to ensure

that the principle of equal pay for work of equal value is applied in the Federal Public

Service..." The Government's initiative formally commenced three months following the

complaint filed on December 19, 1984.

3. The Federal Public Service unions were invited to participate in the Government's initiative

which was called the Joint Union-Management Initiative (the "JUMI"). Details of the JUMI

have been provided by the Tribunal in two earlier decisions. (see Public Service Alliance of

Canada et al. v. Treasury Board, (1992), (the "Voir Dire") and Public Service Alliance of Canada

et al. v. Treasury Board, (1996), T.D. 2/96 (the "Phase I decision").

4. The JUMI conducted a study, (the "JUMI Study"), to determine the degree of sex

discrimination in pay in the Federal Public Service. The Commission was invited to participate

in the JUMI Study by fulfilling the role of observer. The Commission also provided guidance to

the JUMI Committee which directed the initiative. As a result of the JUMI the Commission

agreed to hold the 1984 s. 11 complaint in abeyance pending the completion of the JUMI Study.

The Commission also agreed to postpone all s. 11 complaints against the Treasury Board filed

after the announcement of the JUMI Study and to await the results of the study before

investigating any outstanding complaints.

5. With the breakdown of the JUMI Study the action plan agreed to by the JUMI Committee

was never completed. That plan called for the JUMI Committee to devise methods for system

wide correction to eliminate sexually based wage disparities. The plan was never

implemented. Rather, in early 1990, the government took the unilateral measure of providing

equalization payments to three occupational groups, the Clerical and Regulatory (CR) group, the

Education Support (EU) group and the Secretarial, Stenographic & Typing (ST) group. The

government used the evaluation results of the JUMI Study to calculate the equalization

payments.

6. After the collapse of the JUMI Study, the Public Service Alliance of Canada (the "Alliance")

filed a separate complaint on February 16, 1990, on behalf of six female-dominated occupational

groups that were surveyed in the JUMI Study. It alleged discrimination in pay contrary to s. 11

of the Act. The complaint identified employees in the CR occupational group and five other

female-dominated occupational groups. That complaint (Exhibit HR-10) reads in part:

It is alleged that the results obtained through the process of the Joint Union-

Management Initiative on Equal Pay for Work of Equal Value have demonstrated

the existence of wage rates which are in contravention of Section 11 of the

Canadian Human Rights Act. Specifically, the wages received by the employees

of female dominated occupational groups of CR, ST, DA, EU, HS and LS are

lower than the wages earned by employees of the 53 male dominated occupational

groups included in the study who perform work of equal value. It is further

alleged that this wage difference is gender based, and that the equalization

adjustments to the CR and ST groups announced by Treasury Board on January

26, 1990 are not sufficient to correct this contravention of Section 11.

7. During the JUMI Study the Commission was provided with the results of the job evaluations

that were conducted. The Commission and the Alliance rely on the JUMI Study results to allege

a contravention of s. 11 of the Act.

B. History of Section 11 Complaints

8. Ms. Elizabeth Millar, Head, Classification and Equal Pay Section, Collective Bargaining

Branch for the Alliance, reviewed the history of s. 11 complaints filed by the Alliance with the

Commission after s.11 of the Act was proclaimed in force on March 1, 1978. In November

1979, the first complaint was filed on behalf of 3,300 employees in three female-dominated sub-

groups of the General Services (GS) group claiming discrimination when compared with the four

male dominated sub-groups in the same occupational group (the "GS complaint"). The then

12,100 member GS group had seven sub-groups, each paid at different rates. The three lowest

paid sub-groups, food, laundry and miscellaneous personal services, were female-dominated

according to the Commission's calculations and the remaining four, messenger, custodial,

building and stores services were male-dominated. This complaint was further complicated by

the fact there were 22 zones or regions paying rates different from the national rates. A Tribunal

was appointed to hear the complaint. Prior to the Tribunal hearing a settlement was reached in

March 1982. The settlement equalized each female sub-group regression pay line to the average

regression pay line for the four national male sub-groups. (see Section IX). According to Ms.

Millar the methodology of settlement of the GS complaint became embodied in the amendments

to the 1986 Equal Wages Guidelines, (the "Guidelines"), passed pursuant to the provisions of the

Act, for adjusting wages in group complaints.

9. Both Ms. Millar and Mr. Paul Durber, Director of Payment Equity with the Commission,

referred to the Library Science (LS) occupational group complaint, also filed in 1979, which

dealt with the issue of indirect comparisons. The LS occupational group is an occupational

group of the Respondent. That was a case filed by the Alliance on behalf of the entire female-

dominated LS occupational group claiming discrimination in comparison with the male-

dominated Historical Research (HR) group of the Respondent. During the investigation the

Commission evaluated positions drawn from each level in each group. The complaint was

settled in 1980. The methodology for adjustment is found in a paper entitled "Equal pay for

Work of Equal Value in the Federal Public Service of Canada" prepared in 1983 by John G.

Campbell, Head, Equal Pay and Classification Research, Personnel Policy Branch, Treasury

Board of Canada Secretariat, (Exhibit PSAC-94). In his paper Mr. Campbell describes the

methodology used to adjust the wages of the LS occupational group at p. 47:

An adjustment was derived for each pay level of the Library Science group from

the plotted pay difference between the Historical Research pay line and the mid-

range salary of the Library Science level at the average point evaluation for the

level. [emphasis added]

10. According to Ms. Millar another complaint grew out of the GS complaint. In 1981 the

Alliance filed the Hospital Services (HS) complaint. The Complainant jobs included cooks,

dietary aides, orderlies, nursing and dental assistants and other members of a predominately

female hospital services occupational group. The comparative jobs were the predominantly male

jobs in the GS group which included cooks, messengers, laundry, stores and ground staff. HS

jobs were compared to GS jobs using a classification standard (job evaluation plan for the GS

jobs). A settlement was reached in July 1987 for adjusting the rates of pay for each level in the

HS occupational group to the corresponding level in the GS occupational group. It was agreed

that similarly valued points between the two groups were to be paid the same wage. That meant,

for example, a 200 point-rated HS job was to be paid the same as a 200 point-rated GS job. The

settlement was incorporated into a Human Rights Tribunal's Order. In 1987 that Tribunal

ordered the employer to pay 5,000 workers the settlement amount. Both the Alliance and the

Respondent agreed before this Tribunal the Order of 1987 does not preclude further adjustments

if this Tribunal finds a wage gap still exists as a result of the JUMI Study.

11. Mr. Durber provided examples of other s. 11 complaints the Commission has investigated

since 1978. The majority of s.11 complaints were settled with the consent of the

Commission. These include both individual and group complaints. The Commission has

accepted s. 11 complaints lodged by various public sector unions on behalf of sub-groups of

occupational groups. It is noted one complaint compared work from one sub-group to another

sub-group within the same occupational group.

12. Mr. James Sadler, a Senior Consultant in the Equal Pay Section of the

Commission, testified to the effect that not all complaints lodged with the Commission are

against the Treasury Board or the Public Service of Canada. Other employers included a mining

company, a national trucking company, a security services company, an airline, a native band

council and the Territorial Government. Those employers did not have the same formalized

group classification structure which exists in the Federal Public Service. The evidence reveals

settlements were reached using both direct and indirect methods of comparison. (see Section

VII, E(i)).

13. Moreover, the evidence demonstrated not all the groups who complained were occupational

groups of this Respondent.

C. The Commission's Investigation Process for Complaints Under Section 11 of the Canadian

Human Rights Act

14. Mr. Sadler outlined the investigation process followed by the Commission for a s. 11

complaint. He testified it usually originates when a complainant contacts the Commission and

alleges discrimination under the terms of s.11 of the Act. At that stage the Commission elicits

as many facts as are available from the complainant. Commission staff then review and analyse

the intake information to determine whether there is, in fact, a gender problem; whether it is an

individual or group problem; whether the complainant and the group complained about are

employed by the same employer in the same establishment. The Commission then advises the

potential complainants whether or not they should proceed to register a complaint.

15. Following acceptance of the complaint by the Commission an investigator is designated to

investigate the complaint. Although not required to do so under the Act or the Guidelines, the

Commission then notifies the employer of the nature of the complaint and includes some of the

details. The notification includes the investigator's name and requests that the respondent submit

any defences available to it. The respondent is also asked to produce relevant materials required

for a preliminary investigation such as classification plans, job descriptions, a list of employees

and other pertinent information.

16. Upon receiving a response from the employer an analysis of the response is conducted to

determine whether the information received from the employer is contrary to that received from

the complainant. If, in the opinion of the Commission, a valid defence is raised by the employer

it will then determine whether its investigation should proceed or be discontinued at that stage.

17. If the complaint continues beyond this point the investigator begins a job fact gathering

exercise. This may include sending job descriptions to all employees or a statistically selected

sample of employees with questionnaires relating to the four criteria specified in the Act, i.e.,

skill, responsibility, knowledge and working conditions. During this stage the Commission

expects the supervisory and management individuals within the organization to review and

comment on the completed questionnaires. After a review of the questionnaires the Commission

interviews specific incumbents and their supervisors and, in certain cases, reviews manuals and

documentation. The investigator may observe what machinery or equipment may be required to

do the job. Typically the investigator conducts a field audit to observe the employees at work. A

desk audit may also be conducted by the investigator. This involves reviewing the

documentation, job questionnaires and job descriptions.

18. Once all the job fact information has been gathered the Commission normally establishes a

job evaluation committee, usually consisting of three individuals, to perform job evaluations. If

the employer has an existing job evaluation system the Commission uses s. 9 of the Guidelines to

analyse that system and to determine its suitability for a pay inequity complaint. Otherwise, the

Commission uses the Aiken or Hay job evaluation plan. Section 9 of the Guidelines reads:

Method of Assessment of Value

9 Where an employer relies on a system in assessing the value of work

performed by employees employed in the same establishment, that system shall

be used in the investigation of any complaint alleging a difference in wages, if

that system

(a) operates without any sexual bias;

(b) is capable of measuring the relative value of work of all jobs in the

establishment; and

(c) assesses the skill, effort and responsibility and the work conditions

determined in accordance with sections 3 to 8.

19. The next step in the process, according to Mr. Sadler, is the wage gap analysis which

involves comparing the value and wages of the female group against the value and wages of the

male group to determine any short falls. Mr. Sadler explained this will complete the

investigation. Before a formal report is submitted to the Commissioners the investigator

discloses the findings of the investigation to the complainant and the respondent employer. At

this stage the investigator informs the parties as to the Commission's proposed recommendations

and provides "informal advice to the parties."

20. The Commission then prepares a formal investigation report which is disclosed to the

Commissioners and delivered to the complainant and the respondent employer for review. The

parties are given 30 days in which to make any submissions concerning the report and to raise

issues they wish to have considered. The maximum amount of time allotted is 60 days but the

Commission will consider requests for more time. If the parties make submissions the

investigator who is charged with the complaint is required to consider all the facts raised in the

submissions and to determine whether or not there is a basis on which to reopen certain aspects

of the investigation. The investigator may also decide whether the material submitted by the

parties justifies a response by the Commission. Copies of the submissions are delivered to the

opposite party in order to provide it an opportunity to comment. When this stage is complete the

submissions are attached to the final report with the investigator's comments.

21. Consequently either the original report or an amended version of the report will be submitted

to the Commissioners who, in their regularly scheduled meetings, will consider the

complaint. Following some discussion the Commissioners determine whether to dismiss the

case, refer it to a Tribunal or suggest an appropriate resolution of the complaint. Subsequently

the Secretary to the Commission notifies the parties in writing of the Commission's decision.

22. We note that not all of the steps in the investigation process outlined by Mr. Sadler were

followed by the Commission in its investigation of the complaints now before this Tribunal.

23. The Commission's investigation into the two complaints before us began in March 1990

following the collapse of the JUMI Study. Details of the Commission's investigation were

provided in the Tribunal's Phase I decision on the issue of reliability. During the investigative

stage the Commission retained Mr. Sunter, a former director of Statistics Canada, to assist in its

investigation. His work included an analysis of gender bias in the job evaluation results and a

wage gap analysis involving a critique of the Treasury Board's methodology of wage adjustment

used by the Respondent in formulating the unilateral wage adjustments in January 1990.

24. Mr. Sunter became involved in the analysis of the JUMI Study data on April 5, 1990. He

completed five reports on the subject of gender bias and wage gap analysis. His recommended

wage adjustment methodology was eventually adopted by the Commission.

25. We note Mr. Sunter's first report, "Sex-Based Wage Disparity in the Public Service of

Canada - I, (Exhibit HR-156), is dated May 1990. The second report prepared for the

Commission, "Sex-Based Wage disparity in the Public Service of Canada - II, Analysis for

Seven Female-Dominated Groups", (Exhibit HR-199), is not dated but Mr. Sunter testified it was

done sometime between April and June of 1990.

26. Mr. Sunter's next report entitled "Sex-Based Wage Disparity in the Public Service of

Canada", (Exhibit HR-206), is also undated but was completed somewhere between April and

June of 1990. His fourth report on the subject, entitled "Sex-Based Wage Disparity in the Public

Service of Canada", (Exhibit HR-146), is dated October 1991. On p. 2 of this report he refers to

the previous reports. According to this reference the date beside the fourth report is September

1991. Mr. Sunter's last report, (Exhibit HR-200), dated October 1992 is a supplement to his

second report, entitled "Sex-Based Wage Disparity in the Public Service of Canada - Supplement

to Report II".

27. Due to the intervention of the JUMI Study, the Commission's investigation of these

complaints was modified in a number of key areas. The most notable are as follows:

(i). The JUMI Committee used the constituted occupational groups under the

employer's classification system to select jobs to be evaluated in the JUMI

Study. Jobs in the Federal Public Service are classified into occupational

groups. The JUMI Committee agreed only male- and female-dominated

occupational groups, as defined in s. 13 of the Guidelines, would be included in

the JUMI Study. Based on s. 13 of the Guidelines, excepting the Executive

Category, the JUMI Committee agreed as of March 1985 there were nine female-

dominated occupational groups, fifty-three male-dominated occupational groups

and eight gender-neutral occupational groups. Positions from the eight gender-

neutral occupational groups were then excluded from the study. The Alliance, the

Commission and the Respondent have accepted the occupational groups

designated by the JUMI Committee and informed this Tribunal that the sex

predominance of the complainant groups and the comparator groups was not and

is not in issue. Thus the Commission was relieved of its initial task of

determining the sex predominance of complainant and comparator occupational

groups pursuant to s. 13 of the Guidelines; and

(ii). The Commission itself did not embark on a process for evaluating

work. Rather the evaluation results of the JUMI Study were accepted by the

Commission as evidence of the value of work. Because the Commission's

investigation into s. 11 complaints against the Treasury Board were held in

abeyance during the JUMI Study the Commission suspended its involvement to

await the results of the JUMI Study. These results were tendered before this

Tribunal as evidence of the value of work performed by male and female

employees whose work is the subject of these complaints.

28. The Commission's wage gap analysis conducted by Mr. Sunter was disclosed to the parties

in the Commission's Investigation Report of September 28, 1990 (Exhibit HR-43). This report

referred to the revised scores, which included across-the-board adjustments used by the

Respondent to calculate the wage equalization payments made in February 1990. It also

included Mr. Sunter's critique of the Treasury Board's methodology and his proposed

methodology of level-to-segment, which was adopted by the Commission.

29. The Commission's final investigation report went to the Commissioners in late September

1990. On October 16, 1990, the Commission decided to refer the issue of wages to a

Tribunal. (Exhibits HR-10 and PSAC-25). This included the 1984 complaint and the 1990

complaint. On January 23, 1991 this three member Tribunal was appointed to conduct a hearing

into the two complaints affecting members of the Alliance and other s. 11 complaints filed by the

Professional Institute of the Public Service of the Canada (the "Institute"). On May 10, 1991,

pursuant to the Act, the Commission requested the scope of the Tribunal be extended to include

indirect compensation. (Exhibit T-2).

30. On September 9, 1991, the hearing formally commenced. The Institute's complaints are no

longer before us. Those complaints were resolved by a negotiated settlement in 1995. A

Consent Order was issued by this Tribunal dated May 31, 1995, giving effect to that settlement.

D. Expert Testimony

31. The evidence concerning calculations of pay differentials was provided, for the most part, by

three qualified experts. Two of these experts, produced by the Commission included the

statisticians, Mr. Alan Sunter and Dr. Richard Shillington. Mr. Sunter, a former Director at

Statistics Canada was hired by the Commission during its investigation into the complaints. Mr.

Sunter provided the Commission with an analysis of the wage gap using a wage adjustment

methodology described as "level-to-segment". In Mr. Sunter's opinion this methodology

demonstrates a wage gap between the complainant employees and the comparator employees.

32. Dr. Shillington testified as to his involvement in the JUMI Study. In the course of that

testimony he gave his opinions on Sunter's analysis regarding the statistical technique of

regression analysis and on wage adjustment methodology in general.

33. The third expert Dr. Eugene Swimmer, produced by the Alliance, was accepted as an expert

in labour economics and statistics. He testified almost exclusively on the Alliance's preferred

wage adjustment methodology referred to as the "level-to-composite".

34. Aside from the statistical experts the Tribunal had the benefit of the testimony of Dr. Nan

Weiner, a pay equity expert with an international reputation, produced by the Commission in

order to "educate" the Tribunal in the varied aspects of pay equity including wage adjustment

methodology. Mr. Norman Willis, a pay equity expert, whose "Willis Plan" formed the basis of

the JUMI Study provided consultative services to the JUMI Committee. He provided the

Tribunal with his opinions about many aspects of pay equity including appropriate wage

adjustment methodologies.

35. Lastly evidence of a statistical nature was provided by Mr. Terry Ranger on behalf of the

Alliance. Mr. Ranger has been employed with the Alliance since September 1976 and is

currently the Head of the Research Section in the Collective Bargaining Branch. His job requires

an understanding of statistical methodology and principles. He provided the Tribunal with a

historical perspective of wage adjustment methodology. He explained the Alliance's approach in

wage adjustment methodology based on the composite line.

36. Two Commission expert witnesses, Mr. James Sadler and Mr. Paul Durber, testified as to

the Commission's understanding of the application of s. 11 of the Act and the companion

Guidelines when investigating s. 11 complaints. Mr. Durber provided background information

about the work of Mr. Sunter and the Commission's views of Sunter's wage adjustment

methodology.

II. ISSUES

37. The fundamental issue before the Tribunal is whether the Complainants have established a

prima facie case of discrimination contrary to s. 11 of the Act. It calls into question the

interpretation of s. 11 of the Act and the sections of the companion Guidelines pertaining to

group complaints. Those sections read:

The Act:

11. (1) It is a discriminatory practice for an employer to establish or maintain

differences in wages between male and female employees employed in the same

establishment who are performing work of equal value.

(2) In assessing the value of work performed by employees employed in the

same establishment, the criterion to be applied is the composite of the skill, effort

and responsibility required in the performance of the work and the conditions

under which the work is performed.

(3) Separate establishments established or maintained by an employer solely or

principally for the purpose of establishing or maintaining differences in wages

between male and female employees shall be deemed for the purposes of this

section to be the same establishment.

(4) Notwithstanding subsection (1), it is not a discriminatory practice to pay to

male and female employees different wages if the difference is based on a factor

prescribed by guidelines, issued by the Canadian Human Rights Commission

pursuant to subsection 27(2), to be a reasonable factor that justifies the difference.

(5) For greater certainty, sex does not constitute a reasonable factor justifying a

difference in wages.

(6) An employer shall not reduce wages in order to eliminate a discriminatory

practice described in this section.

(7) For the purposes of this section, "wages" means any form of remuneration

payable for work performed by an individual and includes

(a) salaries, commissions, vacation pay, dismissal wages and bonuses;

(b) reasonable value for board, rent, housing and lodging;

(c) payments in kind;

(d) employer contributions to pension funds or plans, long-term disability plans

and all forms of health insurance plans; and

(e) any other advantage received directly or indirectly from the individual's

employer.

1976-77, c. 33, s. 11.

The Guidelines:

Complaints by Groups

12 Where a complaint alleging different wages is filed by or on behalf of an

identifiable occupational group, the group must be predominantly of one sex and

the group to which the comparison is made must be predominantly of the other

sex.

13 For the purpose of section 12, an occupational group is composed

predominantly of one sex where the number of members of that sex constituted,

for the year immediately preceding the day on which the complaint is filed, at

least

(a) 70 per cent of the occupational group, if the group has less than 100

members;

(b) 60 per cent of the occupational group, if the group has from 100 to 500

members; and

(c) 55 per cent of the occupational group, if the group has more than 500

members.

14 Where a comparison is made between the occupational group that filed a

complaint alleging a difference in wages and other occupational groups, those

other groups are deemed to be one group.

15 (1) Where a complaint alleging a difference in wages between an

occupational group and any other occupational group is filed and a direct

comparison of the value of the work performed and the wages received by

employees of the occupational groups cannot be made, for the purposes of section

11 of the Act, the work performed and the wages received by the employees of

each occupational group may be compared indirectly.

(2) For the purposes of comparing wages received by employees of the

occupational groups referred to in subsection (1), the wage curve of the other

occupational group referred to in that subsection shall be used to establish the

difference in wages, if any, between the employees of the occupational group on

behalf of which the complaint is made and the other occupational group.

38. The Tribunal will address the most appropriate wage adjustment methodology for

estimating the existence and extent of the pay equity wage gap as it pertains to the complaints

before us.

39. Also, at issue is the method of comparison and the application of the law in the case of a

group complaint.

III. SUBMISSIONS OF THE PARTIES

A. The Respondent

40. All parties are in agreement that in determining the extent of a wage gap under

s. 11 of the Act it is necessary to employ a wage adjustment methodology. The Respondent

contends the wage adjustment methodologies of the Commission and the Alliance pose two

problems for the Tribunal. They are:

(i). The wage adjustment methodologies proposed by the Commission and the

Alliance assume s. 14 of the Guidelines to be valid. It is the position of the

Respondent that s. 14 of the Guidelines is invalid because it is inconsistent with s.

11 of the Act.

(ii). If, however, the Tribunal finds s. 14 to be valid, it is the Respondent's

position, that the Commission's and the Alliance's wage adjustment

methodologies must be rejected for other reasons. The Respondent contends the

methodology proposed by the Canadian Human Rights Commission does not

comply with s. 14 of the Guidelines because it selects individual male

comparators from a combination of male-dominated occupational groups, and the

resulting segmented comparator is not an "occupational group". In addition, the

Respondent submits the Commission's methodology does not comply with s. 15

of the Guidelines because it uses only parts of male-dominated occupational

groups, the values in the "low" or "high" range of the male-dominated

occupational groups, (referred by the Respondent as "heads" and "tails") as

comparators. The Respondent contends the language of s. 15 of the Guidelines

specifically requires whole male-dominated occupational group comparators, not

heads or tails.

The Respondent believes the Alliance's methodology, although consistent with the wording of

s. 14 of the Guidelines, is contrary to the principle of equal value found in s. 11 of the Act

because it combines male-dominated occupational groups performing work of unequal value.

41. The Respondent advocates that its wage adjustment methodology is the only methodology

consistent with both s. 14 of the Guidelines and s. 11 of the Act. Pivotal to the Respondent's

argument is the concept of causation. The Respondent seeks support for this argument from the

Phase I decision, supra, which this Tribunal rendered on February 15, 1996. The Respondent

contends the wage gap within s. 11 of the Act must be caused by gender-based

discrimination. In its earlier decision the issue before the Tribunal was whether the job

evaluation scores generated from the JUMI Study were reliable for purposes of the s. 11

complaints. The Respondent now submits because the Tribunal held s. 11 of the Act was

designed to redress systemic discrimination caused by "sex" it is incumbent on the Complainants

and on the Commission to prove any pay differences between male and female employees must

be caused by gender-based discrimination and not for any other reason.

42. Together with the causation factor the Respondent submits two other principles of law are

expressed or implied in s. 11 of the Act. These principles are:

(i) it is necessary to compare the wages of female employees only with the wages

of male employees performing work of equal value; and

(ii). the word "employees" must be given the same meaning in relation to both

males and females, so that comparisons will be made on the basis of either

individuals to individuals, or groups to groups.

43. In accordance with the principles found in s. 11 of the Act, the Respondent submits its wage

adjustment methodology, based on the concept of "central tendency" (see Section IV, C(i)), is a

sound approach in determining whether or not two occupational groups of different genders are

performing work of equal value. Respondent counsel further contends ss. 12 to 15 of the

Guidelines are clear and mandatory and that occupational groups must form the basis of

comparison in group complaints.

44. Respondent counsel contends the purpose of the Guidelines is to implement the broad

general principle of equal pay for work of equal value found in s. 11 o f the Act. According to

the Respondent s. 14 of the Guidelines is invalid because it provides for the combination of

comparator male-dominated occupational groups into a "deemed" group. This, they contend, is

inconsistent with the concepts of causation and equal value required by s. 11 of the Act. The

Respondent submits that to eliminate discrimination under s. 11 of the Act the female

complainant occupational group can only be compared to the lowest paid male-dominated

occupational group of equal value. According to the Respondent any differences in wages

resulting from this comparison will be the result of gender-based discrimination and will thus

satisfy the "causation" requirement of s. 11 of the Act.

45. The Respondent argues s. 14 of the Guidelines is inconsistent with the concept of causation

because the difference in wages between the lowest paid "male" occupational group comparator

and higher paid "male" occupational group comparator of equal value cannot be attributed to

gender-based discrimination since s.11 of the Act only applies to occupational groups of

different genders. For this reason the Respondent submits the Tribunal is precluded from

making an inference that a difference in wages between a female-dominated occupational group

and a "deemed" male group pursuant to s. 14 of the Guidelines is attributable to gender-based

discrimination. The Respondent interprets the "deemed"group as the inclusion of all male-

dominated occupational groups in combination to become the comparator group. (see Section

IV, C(i)). This is because the deemed group comprises not only the lowest paid "male"

occupational group but other higher paid "male" occupational groups as well.

46. The Respondent contends, therefore, the effect of s. 14 of the Guidelines allows for an

adjustment of differences in wages which may be attributed to other factors such as market

values or bargaining unit strength and further allows for the inclusion of male-dominated

occupational groups of unequal value, as in the case of the Alliance's composite line.

47. The Respondent states causation is "operationalized" by s. 11 of the Act in two ways. The

Tribunal is not acquainted with the meaning of the word "operationalized" and interprets it to

mean activated. The first concerns the concept of equal value found in s. 11 of the Act. The

Respondent contends equal value is mirrored in the Respondent's whole occupational group to

whole occupational group wage adjustment methodology using the measurement standard of

"central tendency". The second way causation is "operationalized" is that the Respondent's

lowest paid male-dominated occupational group clearly establishes the pay difference with the

female-dominated occupational group are caused by gender-based discrimination. The

Respondent argues to include higher paid male-dominated occupational groups in the comparator

will bring into consideration the "zone of non-discrimination" (a phrase coined by the

Respondent in argument) contrary to s. 11 of the Act. It graphically described all male-

dominated occupational groups of equal value to the lowest paid male-dominated occupational

group.

48. The Respondent submits, if the Tribunal discards the concept of causation under s. 11 of the

Act and finds s. 14 of the Guidelines valid, any adjustment in wages based on either the

Alliance's composite line or the Respondent's whole deemed occupational group methodology

will set into motion an opportunity for future complaints under s. 11 of the Act by any male-

dominated occupational group which has been included in the "deemed" group under s. 14 of the

Guidelines. The Respondent argues if a male-dominated occupational group finds itself below

the average wage of the "deemed" group it can then complain to the Commission, pursuant to s.

11 of the Act, that it is entitled to have its wages adjusted to the average wage of the comparator

group. The resulting phenomenon is described by the Respondent as "ratcheting." Should the

complaint succeed an adjustment to the wages of the lower paid male-dominated occupational

group would cause the average of the deemed group to rise creating a new pay equity gap

between male and female employees thus starting the process of adjustment all over again. If the

process of adjustments and counter-adjustments continues, the Respondent contends the wages

of all groups used in the analysis would reach the wage level of the highest paid male-dominated

occupational group comparator.

B. The Commission

49. The Commission contends the purpose and goal of the Act and s. 11 of the Act is

"equality". It submits s. 11 of the Act is one of the primary provisions for identification and

resolution of systemic discrimination in the Respondent's pay practices. The Commission argues

the concept of "equality" enshrined in s. 11 of the Act is paramount and consistent with the

nature and purpose of the Act as expressed in s. 2. According to the Commission the goal of the

Act is to achieve equality of opportunity, the protection of rights and privileges in employment

and the provision of services to all individuals. Within that framework it argues s. 11 also has as

its purpose the achievement of equality of remuneration in employment regardless of sex.

50. The Commission contends the concept of "equality" embodies a standard of reasonableness

which must not be restricted by a technical or narrow interpretation of the Act and the

Guidelines. According to the Commission fairness implies reasonableness and s. 11 should be

interpreted as requiring "reasonable" or "fair" treatment. Thus, the Commission contends, s. 11

seeks "on-average fairness," not necessarily the best solution nor the least possible solution but

one that is the most reasonable.

51. The Commission contends the words "equal value" found in s. 11 of the Act gives rise to the

principle of "averaging" applicable to group complaints. The Commission submits, within the

context of compensation policies and practices, the Tribunal is entitled to rely on compensation

experts, pay equity experts and statistical experts to determine the means by which pay equity

can be achieved.

52. In order to achieve equality the Commission contends it is necessary to test for patterns of

treatment of male work to obtain equality of result which will result in "on-average

fairness". The identification of patterns is best demonstrated, according to the Commission, by

its methodology of level-to-segment.

53. The Commission relies on the liberal approach adopted by the Supreme Court of Canada in

interpreting the provisions of the Act to give effect and meaning to the rights enshrined in the

legislation relying on Canadian National Railway Co. v. Canada (Human Rights Commission),

[1987] 1 S.C.R. 1114 and Ontario Human Rights Commission and O'Malley v. Simpsons-Sears

Ltd., [1985] 2 S.C.R. 536. It advocates a purposive approach to the interpretation of s. 11, one

that is compatible with the scheme of the Act.

54. The Commission contends it is not the "classification" of the work that s. 11 of the Act is

seeking to remedy. It argues the essential element in considering s. 11(1) of the Act is the

meaning of "work" in the phrase "employees performing work of equal value". It submits the

identification of occupational groups of equal value, which forms the basis of the Respondent's

methodology, derogates from the focus of s. 11 which is the identification of "work" of equal

value not "groups" performing work of equal value. The Commission submits in order to assist

in the identification of work of equal value it promulgated the Guidelines pursuant to s. 27 (2) of

the Act. It submits the applicable guidelines for group complaints includes ss. 12 through 15 of

the Guidelines.

55. The Commission seeks a broad meaning for "work" and contends it is not groups which

define work, rather, that work defines the group. It argues an overly restrictive approach to the

interpretation of s. 11 of the Act and to the meaning of "work" could lead to absurd results. As

an example the Commission argues not every employer's establishment has a classification

system comprised of occupational groups. An employer may have individual male jobs

unrelated by task or salary structure. An interpretation of "work" restricted to "groups" would

render s. 11 inoperative which, the Commission contends, is inconsistent with the purpose and

goals of the legislation. The Commission submits the intent of s. 11 of the Act must be to

provide for comparison of "work" performed by male and female employees regardless of the

occupational group designation of either the Complainant or the comparator.

56. According to the Commission, based on the testimony of Mr. Durber, one of the initial steps

in investigating a group complaint under s. 11 of the Act is to identify the gender predominance

of the complainant and the comparator group. The Commission submits ss. 12 and 13 of the

Guidelines set out the means of identifying whether or not the "work" is male or

female. Identification of occupational groups is made at this stage by the Commission. The

percentage criteria for assessing gender predominance is found in s. 13 of the Guidelines. In its

view groups are used to identify gender predominance but are not included in the Commission's

assessment of a wage gap which is provided for in ss. 14 and 15 of the Guidelines.

57. The Commission contends that with large group complaints, as in the complaint before the

Tribunal, s. 14 of the Guidelines enables the amalgamation of the identified male work and the

combination of male work becomes the "deemed group" for purposes of s. 14 of the

Guidelines. Accordingly a comparison of work of equal value between the female work and the

male work can then be conducted to determine whether the (female) Complainant groups are

receiving equal pay to the equally valued work of the opposite gender (male) comparators. The

Commission contends its methodology achieves this objective. According to the Commission,

its methodology looks for patterns of male wages and selects male data from the "deemed" group

that is "on average" to the point values of the female Complainant level, achieving "on-average

fairness" in pay pursuant to the intention of s. 11 of the Act. It submits the level-to-segment

methodology uses the most relevant data for comparisons.

58. The Commission submits the process for determining work of equal value under s. 11 of the

Act, includes a four-step process. These steps are set out below:

(i) Determine whether the comparator group and the Complainant group are of

opposite genders, by applying s. 13 of the Guidelines.

(ii) Evaluate the work employing a gender neutral job evaluation plan. This step

ignores occupational groups and gender.

(iii) Separate the genders between "male" dominant and "female" dominant

"work", and compare the "work" to determine if it is of equal value.

(iv) Choose a methodology to identify if there is a wage gap.

59. The Commission argues there is no requirement in s. 11 of the Act for a finding of fact that

"sex" is the cause of the difference in pay between male and female employees. Moreover, the

Commission argues, the differences in pay arising from their methodology are based on the

compensable factors of skill, knowledge, responsibility and working conditions, all of which

were evaluated in the JUMI Study. These compensable factors are enunciated in s. 11(2) of the

Act and further clarified in ss. 3 through 8 of the Guidelines. The reliable job evaluation results

produced in the JUMI Study are used by the Commission for estimating the wage gap. The

Commission submits the resulting wage gap demonstrated by the level-to-segment wage

adjustment methodology must then result from sexually based wage disparities.

60. The Commission further contends if causation is a factor it need only prove that "sex" is

only one of several possibilities for the discrimination under s. 11 of the Act relying on Holden

v. Canadian National Railway (1990), 112 N.R. 395 (F.C.A.) and Uzoaba v. Correctional

Services of Canada, T.D. 7/94 at 91 (C.H.R.T.); affirmed (April 21, 1995), 94 F.T.R. 192

(F.C.T.D.)). The Commission submits the wage gap found to exist in the JUMI evaluations

establishes discrimination on the ground of sex.

61. Lastly the Commission rejects the phenomenon of ratcheting both from a legal and an

operational perspective. Testimony provided by Mr. Durber was to the effect the Commission

adjusts both the female and the male values in the same manner in order to obtain a comparison

by using averaging techniques. According to Mr. Durber the process of adjusting female

evaluation results to an average male regression line does not discriminate against the male

evaluation results. Therefore the males are not subjected to reverse discrimination because, as

with the female results, some are above and some are below the average male regression line.

C. The Alliance

62. The Alliance, the bargaining agent representing the Complainants, supports the

Commission's interpretation of s. 11 of the Act and the validity of s. 14 of the Guidelines. The

Alliance stresses that s. 11 of the Act was intended to address one type of systemic

discrimination, that is, the payment of different or unequal wages between groups of

predominantly male and predominantly female employees performing work of equal value. The

Alliance submits s. 11 of the Act is not aimed at the general wage gap between males and

females but is directed at a systemic problem rooted in history and in attitudes about female work

which tended to undervalue work traditionally performed by females.

63. The Alliance contends this Tribunal's Phase I decision, supra, is not to be interpreted as

requiring the Commission and the Alliance to establish the cause of the wage gap arising from

their respective methodologies. The Alliance argues the Tribunal's decisions must be read in the

context of the issue then before the Tribunal which dealt with the meaning of gender bias and

whether it was present in the job evaluation results.

64. The Alliance further submits there is no reason to prove "cause" under s. 11 of the Act

because that section addresses a systemic problem which by its nature is hard to pinpoint. The

Alliance believes, the inclusion by the legislature of subsection 11(4) of the Act read together

with s. 16 of the Guidelines which lists reasonable factors enabling employers to justify the wage

gap in certain circumstances, rebuts the Respondent's assumption that "cause" is a necessary

element in systemic discrimination. The Alliance supports the Commission's position on the

notion of "cause" as not being a requirement of s. 11 of the Act. Further, the Alliance submits,

the Complainant is not required to establish that gender-based discrimination is the main or

major cause of the discrimination in order to find a breach of s. 11 of the Act. Moreover any

question about causation, according to the Alliance, should be directed at determining whether

the wage gap is the result of the application of the employer's pay practices and pay systems,

including the Respondent's distinctive classification systems, which existed at the time of the

JUMI Study and still exist.

65. The Alliance submits in order to implement s. 11 of the Act a prima facie case will be

proven once a difference in wages is demonstrated based on the four criteria identified in s. 11(2)

of the Act and defined in greater detail in ss. 3 through 8 of the Guidelines.

66. The Alliance contends in order to achieve the spirit and intent of s. 11 of the Act the

reference to "occupational group" under the heading "Complaints by Groups" in the Guidelines

should not be given a restrictive interpretation. Based on evidence it led about the function of

work performed by the female-dominated CR occupational group and the wide range of job

functions in that occupational group the Alliance submits the Respondent's whole occupational

group wage adjustment methodology must assume significant common features, such as

similarity of work, within each of the occupational groups in the Employer's classification

system.

67. In her testimony Ms. Millar illustrated the magnitude of the variation in work within the

largest female-dominated group, namely the CR occupational group. Four separate job

questionnaires from the CR occupational group demonstrate the variation of work performed by

employees in that group. According to Ms. Millar the CR occupational group is over 80%

female-dominated. The group consists of some 50,000 employees and is relatively large when

compared to the male-dominated occupational groups that participated in the JUMI Study. The

four job questionnaires which were evaluated during the JUMI Study related to positions of

Detachment Clerk with the RCMP, a Chief of Administrative Services, an OAS Security Analyst

and an Applicant Services Unit Assistant. These four positions selected by Ms. Millar are

illustrative of the two-sided functional distribution of work in the CR occupational group, i.e.,

the clerical functions and those functions that provide advice and guidance to the public with

respect to federal laws and regulations.

68. The Alliance submits no evidence was led by the Respondent to establish a commonality

within the occupational groups to support reliance upon or the use of the Respondent's

classification system.

69. The Alliance further advocates a broad liberal interpretation of "occupational group" due to

the fact the occupational group classification system was established arbitrarily by the

Respondent through its unfettered authority in the area of classification structure. The Alliance

submits the segregation of work in the Employer's classification system has contributed to the

systemic discrimination addressed in the s. 11 complaints which was to be remedied by the JUMI

Study. It contends arbitrarily established groupings which had contributed to the problem should

not now form the basis for comparison under s. 11 of the Act and ss. 14 and 15 of the

Guidelines.

70. The Alliance submits there is no validity to the Respondent's "ratcheting" argument. It

contends the Tribunal is entitled to rely on the testimony of the pay equity experts and the

Commission witnesses who found no basis for "ratcheting" and whose testimony is that the

remedies afforded by s. 11 o f the Act are intended to provide redress for female-dominated

occupations because of discrimination arising from the historic undervaluation of "women's

work".

71. The Alliance advocates the composite male line as the most appropriate adjustment

methodology to implement pay equity. It points out the JUMI Study selected positions to be

evaluated from 9 female-dominated occupational groups and 53 male-dominated occupational

groups. It submits the selection and sample reflected the parties' intention and agreement at the

time that the wage adjustment methodology would be a composite line for the male comparator

group. It claims the legislated standard existing in 1987 with the newly revised Guidelines was

the composite line and that it was the standard adopted by the parties to the JUMI Study. The

Alliance rejects the Respondent's contention that the parties to the JUMI Study did not comply

with the legislated standard.

72. The Alliance submits the "weighted quadratic" male composite line constitutes equal value

which meet the requirement of s. 11 of the Act. It submits equal value is achieved through the

mechanism of equating equal points between the female-dominated occupational group scores

(by sub-group or level) with the same point value on the weighted quadratic male composite

line. The composite line reflects the Respondent's wage/value relationship for male-dominated

occupational groups. (see Section IV, B).

73. The Alliance further submits a number of policy considerations are relevant to the issue of

the most appropriate wage adjustment methodology. The Alliance lists the following policy

considerations in paragraph 158 of its written submissions to support its methodology:

(1) Consistency with the governing legislative scheme;

(2) Fairness to those affected;

(3) Consistency in wage adjustments as between complainant groups;

(4) Reliability;

(5) The methodology to which the parties agreed;

(6) Methodology most consistent with the views of pay equity experts;

(7) The methodology which is acceptable from a statistical perspective;

(8) The methodology which utilizes the most available data;

(9) The methodology which is capable of being explained to, and understood by,

the people affected;

(10) The methodology with the least anomalies;

(11) The methodology which is simplest to implement with the least decision

rules;

(12) The methodology which the parties have utilized successfully in other

cases;

(13) The methodology which may be utilized by the same employer in a variety

of establishments.

IV. WAGE ADJUSTMENT METHODOLOGY

74. A wage adjustment methodology is a statistical method used to implement pay equity. Each

of the methods advanced by the parties used the statistical procedure of regression analysis to

calculate wage regression lines. The regression line used in each method estimates the

relationship between point values of the sample of jobs evaluated in the JUMI Study and hourly

wages paid for each job. The Commission and the Alliance calculate regression lines for the

male comparator only. The Respondent calculates regression lines for both the male comparator

group and the female complainant group. Regression lines are a form of averaging and are used

by the parties in the respective wage adjustment methodologies to calculate whether a difference

exists between the average wage paid to a female complainant group and a male comparator

group.

75. The point values in this case are the numerical scores assigned by the JUMI

Evaluation Committees to the jobs evaluated during the JUMI Study. The fifteen Job

Evaluation Committees evaluated approximately 1,700 jobs from female-dominated

occupational groups and 1,407 jobs from male-dominated occupational groups that were

randomly selected for evaluation.

76. The system of evaluation used by the evaluation committees was the Willis Plan, a point

factor plan, which has a rating scale constituted for the four job factors listed in s. 11(2) of the

Act. These comprise skill, effort, responsibility and working conditions.

77. The hourly wages used in calculating the regression lines reflect the wage rates for the

different jobs evaluated in the study recorded in the relevant collective agreements between the

Respondent and the Alliance for the fiscal year 1987/88. The wage rates cover the period April

1, 1987 to March 31, 1988. (Exhibit HR-256).

A. The Commissions's Methodology - Level-To-Segment

78. Mr. Sunter initially considered all male-dominated occupational groups that were surveyed

in the JUMI Study as one group for purposes of comparison. After receiving a complete set of

the committee evaluations he began his analysis by drawing a regression line for the data set of

the 1,700 female evaluations. He testified the resulting regression line showed a linear

relationship between wages and value for the female scores. He noted that as the value of points

increased the wages increased thus creating a linear relationship. He followed the same

procedure for the male data set with the same linear result.

79. Mr. Sunter used the female and male linear regression lines as a way of exploring the data to

decide which kind of regression line would be appropriate under the circumstances. He

indicated the distance between the lines can be used to calculate wage adjustments. He referred

to this method as the line-to-line approach. However, Mr. Sunter did not use the overall

regressions as an actual basis for adjustment. He is of the view a regression line of the overall

male evaluation scores is too "rough" an instrument to be useful for calculating a wage

gap. (Volume 108, p. 12995). He used them only to demonstrate that a difference in rates of pay

existed between female-dominated occupational groups and male-dominated occupational

groups.

80. Mr. Sunter addressed the question of how much variation was produced by the overall male

and female regressions by constructing a type of sensitivity analysis to determine the stability of

the regression analysis. For this analysis he invented a large population using the JUMI Study

results. This population was used to select repeated examples similar to the sizes of data used in

the actual JUMI Study. He calculated regressions to see what variations existed and found that

he had a very stable estimate of regression differences from each of the simulations. Mr. Sunter

concluded from this analysis his regressions of the male and female data were demonstrative of

the overall difference between the male line and female line.

81. Mr. Sunter then narrowed his scope and, using a male regression line for male values in the

same range of values for the whole female-dominated occupational group, he compared the mean

(average) distance for each value from the female-dominated occupational group, sub-group or

level under consideration to the male regression line. He referred to this methodology as the

level-to-line.

82. Mr. Sunter further narrowed the scope of his comparison by using the level-to-segment

methodology. He calculated the mean (average) distance for each job evaluation score from the

female-dominated occupational level or sub-group to the regression line for the male values in

the same range as the female-dominated occupational level or sub-group being evaluated. All

available male scores between the point parameters of the female group level or sub-group under

consideration were included in drawing each male regression line. For example, the CR

occupational group is comprised of seven separate levels and has no sub-groups. Mr. Sunter

calculated seven separate segmented male regression lines one for each level in the CR

occupational group.

83. Mr. Sunter used "weighted" regression lines throughout. The weighting accounts for the

differences in the sampling probability. Not all male groups had the same probability for

sampling and therefore the evaluation scores were weighted to reflect the sampling ratios.

84. Mr. Sunter's calculations using the level-to-segment methodology are found in Exhibit HR-

219. This exhibit demonstrates a wage gap exists between the female complainant occupational

groups and the comparator groups.

85. Mr. Sunter considered the level-to-segment methodology to be the best method for

calculating a wage gap using the female level or sub-group and the male values in the same

range of points in the female level. The Commission accepted Mr. Sunter's level-to-segment

method as the best estimate of the wage gap and one which provides a measure of fairness to the

Complainants.

86. Using this methodology the differences found for each level or sub-group to segment within

a female occupational group are combined to produce a total payout (the "pot") for each female-

dominated occupational group. According to Mr. Sunter this methodology requires further

delineation. It does not address how the pot should be distributed within a complainant

occupational group. That is to be determined by the Alliance and the Respondent. Mr. Sunter

explored a variety of methods for the distribution of the pot. (Exhibit HR-219).

87. Mr. Sunter testified as to the purpose for aggregating the adjustments into a pot from each of

the level-to-segment comparisons for each female complainant occupational group. That, he

explained, concerns elimination of statistical errors. He says in Volume 118 at p. 14283, line 6

to p. 14285, line 11:

Q. Under what circumstances, from a statistical point of view, might there be

problems, though, with using each individual level? You have been quite careful

throughout to say it's each individual added up all together that is best for the

group.

A. Yes.

Q. From a statistical point of view, can you tell us what problems might be

associated with using each individual level as a measure of the wage gap?

A. The standard errors associated with each of these numbers may be large

enough that there is no real justification between -- even though these are the best

point estimates, so to speak, for each level, nevertheless examination of the

associated errors may indicate there is no statistically significant difference

between one level, let us say, and its successor. So, a statistician might say, "No,

you have to combine those two levels for the purposes of calculating an

adjustment."

Also, when you look down the figure -- and I think you could classify this as a

statistical reason -- it can happen that the standard errors themselves are subject to

errors in their estimates. They are not written in stone. This is an estimated error.

...

Q. If you have problems when you are at the individual level -- and you have

told us that the best estimate of the wage gap is by totting up all of the individuals

-- can you tell me why the problems aren't there by the time you get to the total

payout?

A. That's because of the way in which errors add. By the time you have added

these things up to the level of the whole group or even of the whole subgroup, the

estimates both of the adjustment itself and the estimates of the standard error that

you associate with that adjustment have both become very reliable. So, that's

simply a result of adding up the sample to the point at which you have a

sufficiently large sample.

88. In his view, the level-to-segment calculation is the most favourable method to calculate the

total pay out for the female-dominated occupational groups because the simple linear regression

used for the male data is a very convenient way of summarizing the male data over short

ranges. He summarizes and expresses his preference for this method in Volume 112 at p. 13490,

line 5 to p. 13491, line 13:

Q. Can you summarize for us why that is your preference statistically?

A. Because the simple linear regression is a very convenient way of

summarizing the general appearance of the data. But if you look at that in some

fine detail, you will see that the regression does not fit particularly well in

particular segments of the overall range. So it makes sense to say: Let us, for

each level, take the most appropriate range for that level.

In this case, we are not depending in any way on some overall assumption that a

straight line fits over a broad range. We are saying: No, no, it is good enough to

summarize the data over a short range, which is what the level to segment

does. And any errors of a sampling or a process error type in the calculation of

course tend to get absorbed when you add up the values for different segments.

We see this in the data that is presented that at a particular level the standard

error of the distance may be relatively large. But when I add them all up and

weight them, in virtually all cases that standard error becomes of negligible

size. It simply disappears. So I have a high degree of confidence in the overall

adjustment indicated by the level to segment.

All the other methods, line to line and level to line, involve the assumption that

the regression on the male side and, in some cases, on the female side fits well

over the whole range. Well, it doesn't fit well over the whole range. At least, we

get much better fits if we take in a series of segments.

89. Mr. Sunter believes that the level-to-segment method provides a more viable, robust and

defensible method than any method which depends on assumptions about the nature of wage

progression across-the-board which is required in the method preferred by the Alliance. In this

sense Mr. Sunter distinguishes very carefully between what he refers to as analytical studies in

which statisticians attempt to fit models to things and descriptive studies in which statisticians

are simply trying to make estimates, (Volume 120, p. 14515). It was this latter approach Mr.

Sunter preferred and pursued in his analysis.

90. Although Dr. Shillington did not perform any analysis himself, he preferred a series of

individual shorter regressions to a larger regression covering a wider range of values for a

group. Dr. Shillington preferred the series of level-to-segment regressions because that method

avoids capturing the shape of the wage-to-salary curve, which is a regression through the group

as a whole. He remarks in Volume 135 at p. 16531, lines 16 - 23:

It's not the same situation with a very narrow range because you are not trying to

capture the shape of the wage-to-salary curve. You are simply averaging data. I

would rather see the summary of a series of segments because each of those is

less sensitive to questions around the shape of the wage-to-salary curve than at the

group as a whole level.

91. Mr. Sunter testified the level-to-segment methodology provides the best available estimate

of the wage gap. He testified with the available data generated by the JUMI Study he was unable

to do a "point-to-point" comparison between the female scores and the male scores. He

described that kind of process as a "descriptive application of statistics." He testified in Volume

108 at p. 13012, line 8 to p. 13014, line 7 as follows:

THE WITNESS: Let me step back a little. What I would like to do, if I had the data and if the

data were available, would be to have a set of male wages for each of these female points and

corresponding to each female point I would have a set of male wages. Let's say, for this point

here I would have a set of male observations at the same value, and then I would take the average

of their wages.

That is not possible because we don't have such a set of points. But, if that were

the case, then I would have a very simple estimate of the wage gap. All I would

do would be to take for each point the distance from the female wage to the

average of the corresponding male wages and then I would average those

distances. That would be a simple estimate within the descriptive class of

applications, and there would be no argument about it. No model would be

involved --

THE CHAIRPERSON: You wouldn't need a regression line.

THE WITNESS: You wouldn't need a regression line.

CHAIRPERSON: You would just be going point to point.

THE WITNESS: Just point to point. It would be a straightforward descriptive

application.

Then I would say that is clearly within the class of descriptive application of

statistics, and it is what it is: it is the best available estimate of the wage gap.

Unfortunately, I don't have such data. Although I can get very close to it, I

cannot get all the way to it. The best I can do is to draw this little segment

regression that you see here on HR-204 and calculate average distances to that

regression. That is as close as I can get to the purely descriptive application.

THE CHAIRPERSON: So what you were saying earlier, when you were saying

that from a statistical perspective that is the best model, if you don't have male

points for each one of those female points, the next best thing is to draw a

regression line between the two segments. Is that what you are saying?

THE WITNESS: Yes. I have brought it down to as local a level as the data will

allow me to do.

92. Dr. Shillington supports Mr. Sunter's approach of using regressions in the descriptive sense

rather than the approach that depicts the nature of the relationship between male values and

salary. Dr. Shillington expressed this opinion in Volume 135 at p. 16533, line 5 to p. 16536, line

9:

Q. When you talked earlier in your answer about a reasonable regression, what did you mean by

that? Just what -- well, I won't ask you any more than that. You used the words you want to

make sure you have a "reasonable regression".

A. The reason for preferring a series of individual shorter regressions, if I can

say that, to a larger regression over a wider range for the group is that the wider

regression analysis is going to be more sensitive -- the answer you get will be

more sensitive to whether or not you do a logarithm transformation of the

regression that we have talked about, put in other curvature terms. That larger

regression is going to depend on how you specify your regression model. The

individual shorter ones won't so much.

Q. Is it fair to say that that deals with the form of the regression?

A. Yes.

Q. In terms of the answer that you have just given, when we are looking at the

composite line based on the total male universe of jobs, what you call the male

composite line, what impact, if any, does the statement which you have just made

have on the use of a male composite line?

A. It's the same discussion, but it's more on a broader range. Would I prefer a

series of regressions over a short range and summarize that information to a

regression line through all the male values? I would prefer the shorter

regressions, again subject to sample size. The reason is that if you do a single

regression through all the lines ---

Q. Through all the lines?

A. Excuse me. If you do a single regression through all the male questionnaire

points, then you better be very, very certain that you have that line specified

correctly.

In this situation [level-to-segment] there is no true line. We are summarizing

data again. Nobody is suggesting that there is linear or a logarithmic or real

relationship between Willis scores and salaries that is from nature. It's not like

doing a regression of the relationship between the height of a building that you

drop a ball from and the speed of the ball when it hits the ground.

There, if you are doing a series of experiments, on one axis you have the height

of the building and on the other axis you have the speed of the ball when it hits

the ground. You do a regression through those points. Your physicist will tell

you that it's quadratic, that there's a squared term in the equation, and that's not ---

Q. So, in the real world there is a proper form for an equation for dropping a ball

out of the window of a building.

A. Because that's the way the physics works, within limits. But in this situation

nobody is coming to you and saying that because of the classification system,

because of the physics, there is a linear relationship or a log relationship. We are

just trying to find a curve that fits the data. If you are going to use one overall

curve, then you better be quite certain you got it right.

In the current circumstance, with the number of questionnaires you have and the

sample size you have, you don't need to do that composite analysis because you

have enough observations, enough questionnaires in the individual ranges that

you can say at the value two hundred (200), let's say, or two hundred (200) to two

fifty (250), if we want to know what males in that range are paid, we can simply

use the males in that range and not have to deal with formulating a regression

equation that fits the overall shape of all the data correctly. [emphasis added]

93. In 1990 the Respondent adjusted the job evaluation results (scores) for gender

bias which it claimed entered into the job evaluation process. Mr. Sunter testified

that the effect of the adjustments by the Respondent to the JUMI Study evaluation

results, presented by the Respondent in its methodology paper, (Exhibit HR-185),

shifted the female regression line and the male regression line. The

shift narrowed the wage gap and resulted in a smaller dollar payout in the

equalization payments made by the Respondent following the breakdown of the

JUMI Study. The Respondent also removed or trimmed the data by removing the

top 10% of the scores and the bottom 10% of the scores as part of its methodology

in 1990.

94. Mr. Sunter criticized the Respondent's methodology of 1990 for removing all evaluation

scores below a certain level and above a certain level. He testified in Volume 110 at p. 13254,

line 15 to p. 13255, line 21:

Q. How does this trimming compare to the removal of outliers which you did in your analyses

that you testified to earlier?

A. What I have called outliers are observations which are so extreme, so far

away from the mean for the particular variable that I am looking at -- let us say

point scores or perhaps relation of wage to point score. They are so far away

from the mean from the rest of the observations that I have reason to suspect that

they do not belong to the same population or were created by the same

process. In other words, there has been a blunder of some kind. That is a possible

explanation. I have some reason to throw them away on those grounds or to

discard them for the purpose of the present analysis. Typically, the number is

very small because I am going to be throwing things away, let us say, at the first

and the 99th percentile.

I am only going to do it after I have looked at the data. I may not throw away

anything.

That is quite different from a trimming procedure or a censoring procedure, as it

is sometimes called, in which I decide beforehand that I am going to only look at

observations between the 10th and the 90th percentile.

The first has a justification in statistical procedure. The second, if it has any

justification at all, would have to be a justification with respect to the procedure

that is going to follow that.

95. Mr. Sunter rejected the composite line as a method of comparison. He testified a statistical

problem of "local inadequacies" arises with a composite line. According to Mr. Sunter "local

inadequacies" arise when the data departs quite significantly from the overall regression and is

measured by the distance of the male data from the regression line. If this occurs the male data is

being misrepresented by the regression line relative to the actual location of the data. Mr. Sunter

did not examine the data in detail to assess the question of local inadequacies instead he

dismissed the composite line as a matter of principle.

96. According to Dr. Shillington the benefit of using segmented lines is that this methodology

avoids local inadequacies or situations where the regression line is being forced through data that

really does not fit the regression line very well. The level-to-segment analysis can account for

small variations in the relationship between Willis points and wages that may not be captured

with a composite line.

97. According to Mr. Sunter he utilized regression analysis as a computational device to

summarize the data for the purpose of calculating an average distance from the female points to

the male points in the level-to-segment methodology. In that context he testified statistical

factors such as "goodness of fit" (a regression line is a good fit to a set of points if the average

distance of those points from the line is very small), "distribution of points" (the distribution of

values between the male and female groups), and "local inadequacies" are insignificant. Mr.

Sunter testified if regression analysis is used in the context of model building to describe a

process in the real world which is intended as a basis for action, as in for example the composite

line, then the validity of the regression line becomes significant and "goodness of fit" is quite

meaningful.

98. There is evidence of different distribution of values between the female groups and the male

groups against which they are compared. Mr. Sunter testified the difference in distribution of the

values in the ranges under consideration from the distribution of values in the female-dominated

occupational group did not pose a problem. He testified as follows in Volume 126 at p. 15351,

line 20 to p. 15354, line 3:

THE WITNESS: There is no difference in principle between the comparison of

a line-to-line between the two regressions over the whole range for an occupation,

and comparison of regressions over the range for a particular level within the

occupation. In principle it's the same problem and in both cases one would prefer

that the two data sets have a similar distribution and that distribution be squarely

well over the range of the observations.

Having said that, we simply have to work with the data we have and we may not

have that kind of equal distribution or well-distributed data in either case. That

does not in itself invalidate the regression. If they are particularly different, the

distributions, you may want to have a look at that problem in case it affects the

regression. And indeed you may introduce some weighting to make sure the

distributions are officially the same.

The difference between the line-to-line and the level-to-segment in this context

is not to do with that particular problem. It is when you move down to the level-

to-segment approach you are typically taking much shorter ranges for the

regressions and so the distribution, or the differences in the distribution across

that range become that much less important. That's all.

That's not the main reason, of course, that I went to level-to-segment, that was

mainly for the purposes of dealing with what I have referred to as local

inadequacies in the overall regression. But since this point has been raised that is

one of the considerations for going down to the level-to-segment, not the main

one.

THE CHAIRPERSON: I would like to follow-up, Ms. Morgan.

Why do the differences in distribution become less important than the level-to-

segment?

THE WITNESS: Because you are working over a much smaller range and

because, you will recall, that the actual level-to-segment calculation for a

particular level -- I have said always that you should be fairly cautious about how

you interpret that particular number. It is only when you add the numbers up over

a whole set of levels that you can be confident that the errors at any particular

level are beginning to cancel out.

So if you do it over a series of short ranges rather than over one long range, if I

am making myself clear here, then you are dealing both with the local inadequacy

problem to some extent and you are also dealing with this problem of differences

in distribution of the two data sets, those two things are occurring simultaneously.

99. Mr. Sunter's methodology produced approximately fifty male regressions for the Alliance's

complainant groups. A question of sample inadequacy arose in a few cases. In some instances

Mr. Sunter found the number of "observations", (job evaluation results or scores), in the female

level restricted the breadth of the range thereby limiting a sufficient number of male observations

(data) for inclusion in the comparator segment. He addressed this problem by expanding the

female range to include more male observations to ensure the regression had statistical

validity. He referred to this expansion of the female range as a "decision rule". This problem did

not arise in the calculations for the Alliance's groups. It concerned the complainant groups

represented by the Professional Institute of the Public Service of Canada. This Tribunal is no

longer seized of the complaints by the groups represented by the Institute.

100. Mr. Sunter found a composite line approach to calculate a wage gap unsatisfactory. In his

view the local regressions which correspond to a segment of the female values were more

accurate and therefore more reliable, whereas the composite line is more precise but arguably

less accurate. He did however use a composite line in his initial analysis for purposes of the

overall regression to demonstrate the difference between male and female salaries, (the line-to-

line approach).

101. Mr. Sunter concluded the level-to-segment methodology as the most accurate calculation of

the wage gap. He testified in Volume 112 at p. 13492, line 21 to p. 13499, line 5 as follows:

THE CHAIRPERSON: Before you do that, Ms. Morgan, I just have one more

question.

Mr. Sunter, in your evidence when you were describing method 3, the level to

segment, you were using words like "precise"and "accurate", and one may be

more accurate but not precise or maybe precise but not accurate.

THE WITNESS: Yes.

THE CHAIRPERSON: We have some definitions in your glossary that briefly

describe the difference between what is precise and what is accurate. Could you

review that? What is the difference between something that could be precise but

not accurate, or accurate but not precise, and how that fits into these different

methodologies and the interpretation that was given in your glossary.

If you need time to think about that -- I don't mind if we go on with what Ms.

Morgan wants to do. But that area is a little hazy with me as to the meaning of

those terms.

THE WITNESS: I think I can do it off the cuff.

Precision is a term generally used to denote replicability. If I measured the

length of this pen, for example, ten times with some instrument and those ten

measurements all come out to be very close to each other, I would say that the set

of measurements is very precise.

However, it is not necessarily accurate. If the instrument I used to measure it is

biased in some way, then I would say that the result I get is inaccurate. So it can

be both inaccurate and precise, as those terms are used in statistics.

It could be accurate but imprecise. I could say: Here is an instrument for

measuring the length of pens, a pen measuring instrument, which is very

rough. But on the average it gives exactly the right value. You may have to

measure the length of the pen 100 times to get the right value, or close enough to

the right value, but that measurement will be very accurate. But the measurement

instrument itself and the series of measurements are imprecise.

So, if an instrument or a procedure on the average gives the right result, you say

that it is accurate. If it gives you the right result with very few measurements, you

say that it is precise because every individual measurement comes close to the

average.

In talking about those regressions, I seem to recall saying that as we go from line

to line to level to segment, we are trading off precision for accuracy.

MEMBER FETTERLY: Could you say that again?

THE WITNESS: We are trading off precision for accuracy for a particular

level. What I meant to say by that was that if you simply take the line-to-line

distance and use that for calculating the adjustment for a particular level within an

occupational group, you are using an instrument which may be very precise, but it

is rather inaccurate because the regressions themselves don't fit very well. But the

individual distances calculated may come very close together. On the average,

they happen to give the wrong answer, but they are very precise.

When we go to level to segment, we have traded off some precision because we

are not moving down to very small sample sizes, but we are much more accurate.

In adding up those level-to-segment distances across the thing, I recover, so to

speak, my precision by increasing the number of observations so that I have the

best compromise of accuracy and precision by doing it that way for the total

adjustment.

MEMBER FETTERLY: Thank you.

THE CHAIRPERSON: What is the definition of "accurate"and what's the

definition of "precision"?

THE WITNESS: I am not sure I can give you a precise definition. An

instrument is precise if its standard error is small. I was trying to give you an

English language description of what that term means, but once you asked me for

a definition...An instrument is accurate if the expected of the value of the result

agrees with the true value.

THE CHAIRPERSON: What you have just talked about, this precision and

accuracy, if you look at the line-to-line regression that has been done using the

Treasury Board methodology, does this apply to that type of methodology? Are

you saying that it's very precise, but it's not very accurate?

THE WITNESS: Yes, line-to-line on that scale of trade-off, if you like, on that

spectrum from high accuracy to high precision. Obviously you would like to have

both, but very often there is a trade-off. On that trade-off the inaccurate end of

the spectrum.

This is essentially -- you see, the argument for using it is revolved in Treasury

Board's discussion around questions of sample size and sample design. In effect,

what they are saying is: No, you have to do it that way because the sample design

and the sample size only give you sufficient precision when you do it at that level,

I may [be] placing words in their mouth. They didn't actually say that, but that's

the gist of the argument.

My reaction to that is, yes, it's very precise, but it's also inaccurate. A well

known statistician once said that it's better to have rough answers to the right

questions then very precise answers to the wrong ones and I don't have much

trouble with that statement.

THE CHAIRPERSON: Just on that, I don't want to explore it too much, but

what is a statistician striving for here when you are applying

methodologies? Does one have more significance over the other, precision over

accuracy or accuracy over precision? Is there a standard within your profession

that says this is something that precision is the key or accuracy is the key, what is

important?

THE WITNESS: No, there is not standard. We do invent measures that combine

those things. Something called the mean squared error combines, if you like,

accuracy and precision. So, the object in sampling exercises, for example, is to

minimize the sum. It's not a simple sum, but the kind of geometric sum of bias

and sampling variance. This is called the mean squared error.

So, in that sense, there is a standard. We are trying to get the best of both

worlds. But very often there is a fairly clear trade-off that you have to

make. You have to go for one or the other.

THE CHAIRPERSON: In the exercise here that was undergone, in trying to

determine if there was a wage gap, et cetera, et cetera, what, in your opinion, is

what you should be striving for?

THE WITNESS: I think it's fairly clear -- it's clear at least in my mind -- that

here what you should be going for is accuracy. After all, the differences in -- if

you simply look at the magnitude of the differences that we are talking about here

between male rates of pay and female rates of pay, precision in that calculation

down to the last cent is not of great importance. On the average, we ought to be

getting the right results. So, accuracy is the point and not precision, in my view,

here.

B. The Alliance's Methodology - Level-to-Composite

102. The wage adjustment methodology proposed by the Alliance is to adjust female wages

using values calculated by occupational group and level and where applicable, sub-group and

level for the complainant groups to a male composite line. The male composite line in the

Alliance's methodology is a regression line using all jobs sampled from the male-dominated

groups in the JUMI Study. This methodology is referred to as the level-to-composite

methodology.

103. Dr. Eugene Swimmer, an expert in labour economics and statistics, did the statistical

analysis on behalf of the Alliance that produced the composite line. He testified as to the validity

of the composite male line. Based on his analysis, Dr. Swimmer was of the opinion that a

quadratic equation best reflected the male data generated in the JUMI Study. In contrast to Mr.

Sunter's use of regression lines Dr. Swimmer utilizes the regression line as a model to predict the

wages for all employees in male-dominated occupational groups in the Federal Public Service.

104. Using a scattergram (a graph showing the different evaluation point-scores) Dr. Swimmer

observed the male data traced an upward sloping curve which tended to flatten out at high values

of the job evaluation points. Because of this flattening out at the higher end values Dr. Swimmer

preferred a quadratic equation to a linear equation to predict the male wages. He testified the

more complex quadratic equation produces a curvilinear male regression line.

105. Dr. Swimmer noted the sampling procedure employed in the study was not equally

representative of all occupational groups. In some cases the male data tended to be over

represented in some occupational groups and under represented in others. Dr. Swimmer felt it

appropriate to adjust for this difference. Therefore he weighted individual observations to reflect

their sampling probability. In this process he allowed for each of the male evaluation points to

contribute equally to the formula for drawing the regression line.

106. According to Dr. Swimmer, since the regression line produced is used to predict the

relationship between wages and points for a population of approximately 100,000 employees

working in the Federal Public Service, the "goodness of fit" of the line becomes a statistical

issue. According to Dr. Swimmer an index number has been developed by statisticians, referred

to as r2, which gives an indication of the goodness of fit of the line. In statistical terms the r2 is

the standard error of the regression line as a whole. In every regression line the r2 statistic will

be between 0 and 1. The closer the r2 is to 1 the more reliable the line is as a predictor of

wages. In graphic terms a regression line with much of the data on the line with a little above

and a little below is a better fit than if the data is scattered all over.

107. Dr. Swimmer testified the r2 for the simple linear regression for the male data from the

Study was 0.72. When the regression is weighted it produces an r2 of 0.71. Dr. Swimmer

compared the r2 between the quadratic equation for weighted and unweighted forms with the

respective linear equations and this comparison suggested to him that the quadratic model may

be superior to the linear model on statistical grounds.

108. Dr. Swimmer then formally tested whether the quadratic (curved) model was significantly

better than the linear model using a statistical test called the Wald F Test. Based on these results,

Dr. Swimmer concluded the quadratic equation was a superior way of summarizing the data of

the approximately 1,408 males in the sample.

109. Dr. Swimmer testified there are two attributes about regression estimates that concern

statisticians; one is "unbiasedness" and the other is "efficiency". Dr. Swimmer described an

"unbiased" estimate as one which, if the process is replicated many times, results in the average

of the estimates being virtually identical to the slope of the regression line for the true

population. He testified regression estimates are said to be "efficient" when there is no other

process which is unbiased and where they will generate values close to the entire population

values as is the case with the ordinary least square method (the regression line).

110. Dr. Swimmer assured the Tribunal both of these attributes were present in the Alliance's

composite line methodology. This was accomplished by assessing yet another statistical

consideration called the "error term". He identified the "error term" as the difference between

the actual value of wages and the predicted value of wages from the regression equation at a

given number of job evaluation points. Dr. Swimmer employed algebraic techniques to correct

for a problem he referred to as "heteroscedasticity" which results in increasingly greater errors

when predicting wages which fall at higher job point values. As a result, he produced the

following weighted quadratic equation for the 1987 data: Pay 87 = 4.029 + .060 Point - .000014

Point2. (Exhibit PSAC-164).

111. The results of Dr. Swimmer's analysis are found on p. 22 of his report, (Exhibit PSAC-

164), and are as follows:

For the reasons addressed earlier, based on my analysis of the data and the

application of accepted statistical tests, I believe that:

a. a single composite regression provides an excellent fit for these data

b. the weighted quadratic equations, corrected for heteroscedasticity generate

unbiased and efficient estimates of the population characteristics and are therefore

the best specification of the overall model for the relationship between hourly

wage rates and job evaluation points for employees in male dominated

occupational groups

c. the predicted male hourly wage rates from these weighted and 'corrected'

quadratic regressions would serve well as the bench mark for adjusting hourly

wages of female dominated groups.

112. Dr. Swimmer provided no opinions on either the level-to-segment methodology or on the

whole group methodology, the latter being the Respondent's methodology. Mr. Ranger of the

Alliance provided calculations of the wage gap using the level-to-composite methodology in

Exhibit PSAC-187. The actual method of calculation is similar to Mr. Sunter's level-to-segment

methodology except the Alliance uses the weighted quadratic composite line in calculating the

differences between the female and male groups rather than individual segmented lines. Using

the weighted quadratic equation the Alliance has demonstrated a wage gap exists between the

female complainant occupational groups and the comparator groups.

113. The Alliance submits the parties had agreed in the JUMI Study to utilize the composite line

to calculate the size of the wage gap during the phase to follow the job evaluation phase of the

JUMI Study. The essence of the Alliance's submission on this point is found in paragraphs 173

and 174 of its written submission. They read as follows:

173. In the respectful submission of the Alliance, the evidence, both viva voce

and documentary, confirms, in clear terms, the fact of the parties' agreement that

the composite male line would be utilized as the male comparator to calculate the

size of the wage gap following the job evaluation phase of the Study. While the

Alliance recognizes that no final agreement was reached at the outset of the Study

or in the course of it respecting the manner in which female complainant groups

would be utilized, it seems clear that the only issue on this aspect was whether the

female groups would be compared by entire female group or by level within each

female group or sub-group as the case may be.

174. In the submission of the Alliance, the agreement on the approach to male

comparators has been confirmed by Ms Manseau, Ms Jaekl, Ms Millar, Mr.

Sadler, Mr. Ranger, Ms Brookfield and, notably, Norman Willis. This agreement

is also confirmed in the Gower Exhibit respecting the sample size given that the

manner in which males would be compared was fundamental to determining the

size of the male sample. No party, including the Commission, subsequently

objected to the use of the male composite line during the Study. [emphasis

added]

114. The appropriateness of the sample design used in the JUMI Study is not an issue in these

proceedings. Initially the Respondent wanted a smaller sample than the Alliance to conduct job

evaluations. The sample size was finally agreed upon by the parties and approved by Statistics

Canada. The original sample was reduced with the approval of Statistics Canada. The female

complainant group sample was larger than the male comparator group sample. The female

complainant group sample was represented by group, sub-group and level for the female-

dominated occupational groups and by group for the male-dominated comparator occupational

groups.

115. The Alliance refers to a paper written by Allen R. Gower, Social Survey Methods Division,

Statistics Canada, dated February 12, 1987, entitled "Observations Regarding Sample Size 'Equal

Pay for Work of Equal Value.'" In that paper Mr. Gower described the method of sample

selection in the JUMI Study as an equal probability sample (Exhibit PIPSC-12). Mr. Gower

provided comments about proposed sample sizes for the JUMI Study and methods of adjusting

salaries. We note Mr. Gower's comments were based on an assumption that the salaries of

employees of female-dominated group would be adjusted using the female level to the composite

line. He addresses two preliminary sample size proposals, one comprising 5,200 positions

suggested by Mr. Willis and the other comprising 2,550 positions proposed by Mr. Jean

Bourdeau of the Treasury Board. According to Mr. Gower's report, the reliability of a male

composite line using different samples depends on the size of the sample. He writes on p. 5 of

his report:

The sample sizes for female-dominated groups are 2,422 and 1,950 in the Willis

proposal and the Bourdeau memorandum respectively. However, the respective

sample sizes for the male-dominated groups are 2,436 and 600. For male-

dominated groups the difference lies in the assumptions made about the

adjustment method. A sample size of 2,436 would very likely provide good

reliability for a male composite line (and for separate male lines for many of the

groups). A sample size of 600 would provide less reliable data (assuming that the

same sample selection method). Although the data would be sufficiently reliable

(± $500) for a male composite line (based on previous data relating to CHRC

complaints), there would not be sufficient reliability for separate male lines for

most male-dominated groups.

116. The excerpts from JUMI Committee meetings used by the Alliance in support of its

contention that the parties agreed to utilize a male composite line as the male comparator to

adjust wages are reproduced as follows:

(i) October 9, 1986:

Statistical Methodology

Since the staff side had not had an opportunity to discuss this matter thoroughly,

a committee agreement was not possible. The management side indicated its

preference that the wage of each female dominated group be adjusted on a

percentage basis to the composite male line. Christine Manseau agreed to make

the staff side position known at the next meeting, October 29.

(ii) October 29, 1986

Statistical Methodology

The staff side indicated its preference for the male composite line to female level

approach at a $300 reliability with 3,100 to 5,300 positions. The staff side also

indicated that the input of Willis would be useful at this point.

(iii) January 30, 1987

The Male Composite Line

The salary range for the male composite line should be from about $15,000 to

$75,000. Given that we are only interested in a good salary-point relationship for

the male composite line, a sample of 600 positions would be [reliable]. We could

add the constraint that there should be at least 3 to 5 positions sampled in each

male-dominated occupational groups.

(iv) March 2-6, 1987

Statistical or Adjustment Methodology

Ted Ulch, CHRC representative, stated that 1986 Equal Wages Guidelines

required the use of level ratings.

The Committee agreed to defer decision on the adjustment methodology until

the evaluation of sampled positions had been completed.

(v) October 14, 1988

Letter from the Canadian Human Rights Commission

Mr. Willis stated that Mr. Ulch's recommendation could be carried out on male-

dominated groups given that the male wage line was to be treated as a universe;

this might not be possible with female-dominated positions, for which the

occupational group and level would be used to determine any wage curve. JUMI

had agreed to go to a male composite line but no decision had been made on

female-dominated groups.

117. The Alliance also refers to the testimony of Mr. Willis concerning the sample design and

size to support its contention of an agreement by the JUMI participants to use the male

composite line. Mr. Willis testified it was his understanding that the sample of positions selected

for evaluations was made on the basis of male-dominated positions being treated as "one

universe" and the female-dominated positions being separated by group and level for wage

adjustment.

C. The Respondent's Methodology

(i). The Whole Group Methodology

118. The Respondent's methodology is based on the use of whole occupational groups as a basis

for comparison. The whole occupational groups are those identified by the Employer's

classification system and used in the JUMI Study.

119. The Respondent's methodology consists of two separate approaches. The preferred

methodology is to adjust the pay of a female-dominated occupational group where a wage gap

exists between that female-dominated occupational group and a single, lowest-paid, whole male-

dominated occupational group performing work of equal value to the value of work of the

female-dominated occupational group. Since s. 14 of the Guidelines provides for a comparator

group composed of all male-dominated occupational groups, the Respondent's second approach

is to compare the wages of the complainant group to the "deemed" group. The deemed group is

composed of all the male-dominated occupational groups identified to be of equal value to the

complainant group.

120. The choice of male comparator, either the single male-dominated occupational group or

the deemed group depends on the validity of s. 14 of the Guidelines. According to Respondent

counsel, if s. 14 of the Guidelines is found to be invalid, the deemed group is eliminated and the

Tribunal is driven to the lowest paid male-dominated occupational group as the comparator

group. Respondent counsel provided the Tribunal with the option, absent any conflict between s.

11 of the Act and s. 14 of the Guidelines, that it could accept the deemed group approach

instead. The deemed group approach, which the Respondent says must be adhered to if s. 14 is

applied, permits only the whole group methodology to the exclusion of either the level-to-

segment or the level-to-composite line approach.

121. The Respondent bases its methodology on the concept of "central tendency". Dr.

Shillington described the term "central tendency"in Volume 140 at p. 17281, lines 1-10 as

follows:

"Central tendency"is a term used in statistics to encapsulate various ways of

trying to capture the middle of the data. It would include the average or the mean

as one way of measuring the middle. The median is another. You could come up

with other ways as well. All of those together are trying to encapsulate something

about typical, average, middle. So, central tendency refers to all of those

measurements.

The Respondent's methodology addresses whether the female-dominated occupational group and

the lowest paid male-dominated occupational group have the same "central tendency". The

Respondent contends the central tendencies of two groups can be compared by using the "mean"

or the "median" to see if they are equal.

122. Dr. Shillington testified the average of a set of data is the arithmetic mean, the sum of the

numbers divided by sample size. The limitation of the mean is that it can be influenced

markedly by a few extreme values. On the other hand, he described the median as the middle

number when the values in a set of data are arranged in ascending or descending order. It is

synonymous with the 50th percentile. A significant advantage of the median is that it is not

strongly influenced by individual extreme values. According to Dr. Shillington the median is

affected by the distribution of the values and the order of the values (lowest to highest).

123. According to Dr. Shillington the term central tendency can be used to present single or

summary conclusions about the value of work of a sample for members of a group.

124. The Respondent applied its whole occupational group after the demise of the JUMI

Study. In early 1990 the Respondent used the JUMI Study results to make unilateral pay

adjustments to three of the female-dominated occupational groups that were surveyed in the

JUMI Study. Documentary evidence of these adjustments is contained in a methodology paper

received by the Commission from the Respondent in March of 1990, (Exhibit HR-185). Mr.

Sunter did an analysis of the Respondent's methodology as part of his work for the

Commission. Both Mr. Sunter and Dr. Shillington were critical of the Respondent's 1990

methodology. The Respondent led no evidence to explain the basis of the methodology it used

in 1990. Some aspects of the methodology found in Exhibit HR-185 were retained by the

Respondent in the methodology it advocated before the Tribunal in argument.

125. In 1990 the Respondent used a statistical test as part of its methodology for adjusting

wages. The test, applied in 1990 and based on whole group comparisons, used a standard

statistical test referred to as the Wilcoxon Rank Sum Test. This test was used as a significance

test to determine the difference between two groups, the complainant group and the comparator

group. More specifically the test compared a set of numbers from each group to determine if the

median value, i.e., the 50th percentile of the two groups, was different. If the medians of the

complainant female group and the composite male group were significantly different the

Respondent rejected the male comparator group. The appropriateness of the Wilcoxon Rank

Sum Test for wage adjustment was criticized by Dr. Shillington and he dismissed the notion of

using any or all significance tests for making comparisons. Dr. Shillington's concern with the

Wilcoxon Rank Sum Test is that as the sample size becomes larger the test will demonstrate that

all these groups are different. Dr. Shillington testified the Wilcoxon Rank Sum test asked the

wrong question which is directed at the exact identification of values.

126. The Respondent now argues the concept of "central-tendency" should be applied in

assessing "equality", rather than the statistical test it used in 1990 to determine the size of the

wage gap.

127. Dr. Shillington also disagreed with the Respondent's approach requiring whole male-

dominated occupational group comparators. He testified in the strongest terms that if he had to

use the "mean" or "median" for making group comparisons, the approach should be one that

would retain data rather than exclude it. Dr. Shillington stated the data must inform the analysis

in a group complaint. We refer to his testimony in Volume 140 at p. 17306, line 4 to p. 17307,

line 20:

Q. Let me understand a bit better. Is your contention that one should not use a

test of central tendencies to determine comparators? Is that your basic

contention?

A. No. I think I said it as clear as I could in response to one question from the

Chair about setting criteria.

I would approach the question in the following fashion. If you must use males

via the group process, a group is in or out, you would say that we want to include

the male groups that have enough overlap in value whose values are in some

collective way similar enough to the females that we would be comfortable saying

yes, they are "comparable", that they inform the analysis.

It is not whether or not the medians or the means are identical.

Q. Dr. Shillington, I suggest that when you just made that statement you are

entering into the field of deciding what comparability means. You have told us

before that you are going to leave that to the lawyers to argue.

A. No, no, no. The interpretation of the statute and whether or not that means

that the groups have to be identical the lawyers are -- I wish you well.

The issue about whether or not in choosing comparators -- in identifying the

male groups who are comparable, the issue about whether or not that is a

significance test or a technique which includes groups not using significance tests

but based on some kind of rule of thumb, the average for one group must be

within the two standard deviations, or the median for one group should be within

the quartile range, or -- and I said I don't know what the right answer is. But I am

sure that if you use a significance test, you will get into the logical inconsistency

of developing a technique which, if you had a large enough sample size, there

would be no comparators. [emphasis added]

128. The Respondent's methodology now proposed arose from an "off the cuff" suggestion

made by Dr. Shillington in response to Commission counsel questioning his examination in-

chief. Commission counsel asked for Dr. Shillington's opinion about devising a methodology

that requires only whole male dominated groups as comparators for female-dominated

groups. Dr. Shillington's response is found in Volume 132 at p. 16176, line 18 to p. 16185, line

3:

What if the Tribunal in these proceedings determines that the legislation, the Canadian Human

Rights Act, requires that the analysis of discrimination in pay requires that only whole male-

dominated groups could be used as comparators for female complainant groups? Have you

given any thought to this question and what sort of things you would look at for devising a

possible methodology?

A. Yes, I have given some thought to approaches you might explore.

Stepping back again to the fundamental research question, you want to identify

male groups which would be included in the analysis for comparison

purposes. So they would be groups where there is a goodly number of

observations that have job values comparable to the female job values.

Again, without having pursued this at all with data or tried to operationalize it,

you could try various techniques that at least half or at least three-quarters of the

male values would have to be within the female range. You could ask that the

median of the male values be between the twenty-fifth and seventy-fifth

percentile of the female range. There are various ways of creating an operational

definition that would try to ensure that the male comparator groups had

observations that are roughly comparable in job value to the job values of the

female group.

Q. Before you go into any further suggestions or we get clarification of those

few suggestions you threw out, what about the use of the Wilcoxon Rank-Sum

Test as a significance test to determine comparable male-dominated groups?

A. Using it as a significance test, as I have said, I think is addressing the wrong

question. The question is not whether or not the groups have the same

median. You don't sit there and assess the evidence that these groups have

different medians. If you have enough data, you will have evidence that they

have different medians.

The question is, are they similar enough that they can be used for comparison

purposes, not whether or not we have enough evidence that they are different.

So, no, I would not advise using a significance test in this at all, actually.

Q. You have given, I heard, two possible suggestions. To what degree have you

explored these ideas?

A. These are strictly off-the-cuff or off-the-top-of-my-head ideas about the way

you might approach it. I don't want anybody to think that I would defend those to

any strong degree. You would have to get in and look at the data and get some

sense of what the impact of those decision rules might be before you would feel

confident that you had one that was operationally doing what you wanted it to do,

which is identify male groups which are, in some sense, comparable.

Q. On what basis can you wholly discard the use of the Wilcoxon Rank-Sum

Test as an alternative methodology for identifying comparable male-dominated

whole groups, whereas you are not completely rejecting these other

methodologies?

A. I think it is a matter of identifying whether or not the test that you use or

propose addresses the question, the appropriate question.

I think the question, assuming that we are going to take a group approach,

is: Does this group have enough male questionnaires with job values in the range

of the female group that we want to include it?

That is the non-statistical question, I think. I think the types of decision rules I

suggested as possibilities are ones that could be used to answer that

question: Does the male group have enough values in common with the female

groups to include it?

The Wilcoxon isn't answering that question. The Wilcoxon significance test

is: Do we have evidence that the medians are different, non identical?

With very large sample sizes, it would not be difficult to create a set of male

questionnaires which are very, very close to the set of female questionnaires in

terms of job values but which would have significantly different medians. It

would not be difficult at all.

The reason I am confident in saying I would not use the Wilcoxon is because I

am confident it is answering an inappropriate question.

THE CHAIRPERSON: Could I ask a question, Ms. Morgan?

MS. MORGAN: Please do.

THE CHAIRPERSON: If you are analysing data and you are trying to come up

with criteria to say that these are similar enough or they have values that are very

close, that they are similar or comparable, what kind of criteria when you are

analysing data statistically do you look at to say yes, they meet what is required?

THE WITNESS: I don't think this is a deep statistical question. I think that this

is a question that everybody here could address.

If you look at the CR group, as an example, and the various male occupational

groups, you could look at the males in terms of the median of the male values for

each group, their maximum, minimums, the twenty-fifth or seventy-fifth

percentile. What you want to do is identify the male groups which, if you

included them, would meet whatever your criteria of comparability is.

That might sound like I am ducking the question.

THE CHAIRPERSON: That is the question I asked you, what would your

criteria be, and you are saying it should meet the criteria. I am saying can you

define the criteria.

THE WITNESS: Again, I don't think there is a pre-defined statistical

technique. I have problems with using significance level because, as your sample

sizes get large, you are going to definitely have no comparators. I think that

should tell you that there is a philosophical flaw in the reasoning.

I would start and say -- you might start and say: What would happen if we were

to include the male groups where the median is within ten per cent of the female's

median? You could just see whether or not that seemed to work reasonably.

I don't know what the criteria ---

THE CHAIRPERSON: Is your objective in this exercise to get a reasonable

number of male comparators? Is that one of the -- maybe if you approach it from

what is your objective here, if you are going to answer your question. What are

you trying to achieve?

THE WITNESS: I have a small amount of difficulty answering because I would

start with -- I wouldn't use the groups as a conduit at all; I would just say the

males, as long as your job value is in the range.

But if you had to go through a group approach, you certainly -- if you came up

with the technique that you thought met certain predefined criteria but then had no

sample, I think you probably haven't achieved anything very useful.

So having a goodly number of male observations that pass the criteria would

have to be practically one of your criteria. I say "practically" because you may

start off by saying you want observations which meet some criteria of

comparability and then you may design a technique that gives you that, and then

you have no sample size because when you utilize that criteria ---

THE CHAIRPERSON: So one of your criteria has to be, would it not, that you

have to have a sample size? Wouldn't that be one of your criteria?

THE WITNESS: I am trying to differentiate between a criteria that speaks to

comparability, which is -- maybe I will call it more of a philosophical or

theoretical criteria, and then a criteria of practicality, which is that if you use a

technique which you are totally satisfied with theoretically or philosophically but

leads you to inadequate sample sizes, you are probably going to end up going

back and revisiting the question and saying maybe we have to loosen up.

If you started off and said "We want to have the male group to have the same

distribution"-- which you can characterize the Wilcoxon approach. You could say

philosophically, if we have the same distribution, that you are unassailable, in a

sense, but you get no sample size and it is not a very practical recommendation.

But I don't even think the same distribution is necessary if you go back and say,

"We need, for the comparison, males who come from groups where there is at

least enough overlap in the job values that there is a reasonable comparison, that

there are enough males at 250 and 260 so that we are not extrapolating or

interpolating, we are making a reasonable assessment."

If that is our criteria, then you could assign inclusion criteria that would follow

that type of rule and hopefully give you also a practical application because you

wouldn't have small sample sizes.

So there is a bit, in my mind, a little bit of a two-stage analysis here: Design a

criteria for inclusion that meets some kind of philosophical definition of what you

mean by comparability, and make sure that you haven't defined that so narrowly

that you end up with no observations. [emphasis added]

129. The Respondent contends Dr. Shillington's philosophical definition of comparability,

referred to above, should be rephrased to mean the legal requirements of the

legislation. According to Respondent counsel there are two legal requirements for a

methodology to be consistent with s. 11 of the Act. Firstly, the requirement for comparisons by

occupational groups and secondly, the comparison of wages of only occupational groups

performing work of equal value. Respondent counsel argues the legislation affords the Tribunal

some discretion in applying the second requirement by striking a balance and thus avoiding an

overly technical or restrictive interpretation of s. 11 of the Act.

130. The Respondent's preferred choice in selecting a comparator group is further explained by

its written submissions in paragraph 126, referred to as banding:

126. Another way to assess the proximity of the means or medians of two groups is to prescribe

a band of values around the mean or median of the complainant group, and to identify all male

groups having a mean or median value within the value band of the complainant group. The

band might be based on a specified number of evaluation points, a specified percentage of the

value of the complainant group, or a specified number of percentiles. This method does not rely

on any statistical test.

131. The Respondent summarizes the benefits of the banding method when choosing

comparators in paragraphs 367 and 368 of its written submissions as follows:

367. The Whole Group Comparator methodology, using the banding method to

identify the lowest paid male occupational group as the comparator:

(a) complies with section 11 of the Act in that:

(i) it is designed to identify a group of male employees performing work of

equal value; and

(ii) it provides a means to distinguish between the portion of the wage gap

caused by discrimination based on sex and the portion of the wage gap caused by

other factors;

(b) complies with sections 12, 13 and 15 of the Guidelines because it provides

for comparison between a complainant which is an "occupational group" and a

comparator which is an "occupational group";

(c) is easy to understand;

(d) is neither technical nor restrictive;

(e) does not rely on statistical significance tests;

(f) is not seriously affected by measurement error; and

(g) is versatile in terms of application to groups of different sample size.

368. For these reasons, the Respondent submits that the Tribunal ought to select

the whole group comparator methodology using the banding method to identify

the lowest paid male occupational group as the comparator for estimating the pay

equity wage gap in this case.

132. We note there is only a slight reference in the evidence for the banding method technique

in pay equity adjustments. Dr. Weiner made brief mention during her testimony to the concept

of banding in compensation practices. According to Dr. Weiner bands of points are established

in some compensation systems to respond to situations where jobs have slightly different point

values but are considered as having comparative value. She explained this process in Volume 7

at p. 1004, line 13 to p. 1005, line 16:

Q. Now, I know that you didn't want to get into the business of comparable comparator, but in a

nutshell what is it?

A. A comparator?

Q. A comparable comparator?

A. Okay.

In reality, using -- we've talked about point systems, we've talked about the fact

that they are based on subjective judgments. And while committees make the best

subjective judgements they can, it's pretty hard to say that a job at 100 points is

really that much lower than a job at 101 points. So we might say well those two

are really comparable.

Now, we probably could say that a job at 100 points is different than a job at 200

points; but is a job at 100 points different than one at 110? Now we are getting

into a grey area.

The analogy I use is that in grading in a classroom situation, you can give

students a 70 or 71 or 72 and they all get the same letter grade. So we do some

grouping.

So it is quite common in compensation to say that the jobs from 100, let's say to

150 points, are the same. In some systems it's 100 to 120, sometimes it's 100 to

200. So there is some, if you will, banding of points that are then deemed to be of

comparable value. So that's allowed under the legislation.

133. The Respondent at first suggested the median be used as the central tendency for

comparison, (Respondent's written submissions, paragraph 89). Comparator groups were to be

identified by selecting male groups having the median of the observations of work value within a

+/- 10% band of the median of the observations of work value for the female-dominated

occupational group. The Respondent abandoned that proposal in its oral argument in favour of

the mean. Referring to Exhibit R-176, a compilation of male comparator groups for each female

complainant occupational group using firstly the Wilcoxon test, then a banding mean and lastly a

banding median, Respondent counsel submits in Volume 240 at p. 32023, lines 9-18 as follows:

I can tell the Tribunal right now that the proposal that the employer is making is column

number (2) [banding mean] and the reason is that we have evidence on that in the record,

whereas for the methodology that the employer talked about in its written submissions, which

was plus or minus 10 per cent of the median, there is no evidence on the record for that. So, the

employer is proposing to the Tribunal column number (2), for which we have evidence.

134. In applying the Respondent's last proposed methodology, the lowest paid whole male-

dominated occupational group comparator is identified by selecting a male-dominated

occupational group having the mean of their job evaluation scores (or observations) of work

value within a band of +/- 10% of the mean of the job evaluation scores (or observations) of

work value of the female-dominated occupational group.

135. The Respondent contends the wage gap is then calculated by estimating the differences

between occupational groups with central tendencies which are equal. In applying this

methodology we note there would be no adjustments to the complainant groups with the lowest

paid male comparator group. Using the deemed group approach we note two of the female-

dominated occupational groups would receive adjustments, these are the LS and HS groups. The

Respondent's calculations for the wage gap are provided in Exhibit R-179. According to

Respondent counsel differences between female and male occupational groups were made by

drawing a vertical line from the mean of the regression line of the female occupational group to

the male regression line. The estimated wage rate of the male group is calculated at the point the

vertical line intersects the male regression line. The female wage rate and the male wage rate are

then compared to determine if a difference exists. (Volume 245, p. 32658 ff.).

136. Respondent counsel further submits a selection of comparators based on the mean rather

than the median will yield more comparators and in that sense will, or should, satisfy Dr.

Shillington's concerns about including more observations in making the comparison. However

the Respondent provided no evidence to support its submission.

137. Respondent counsel also submits the Tribunal need not accept the band of +/= 10%

because it has the discretion to define the criteria of comparability and to decide how wide the

band should be if two groups are equal. Other parameters of comparability suggested by

Respondent counsel are found in Volume 240 at p. 32028, lines 2-14 as follows:

It is basically as simple as this. You as a tribunal or parties in another case will

say: We will consider two groups to be equal if their means are within a certain

distance of each other. They could be within a certain percentage distance of each

other, they could be within a certain percentile of each other, within a certain

corridor, they could be within a certain number of fixed points. Let's say if they

are within 40 Willis points of each other we would consider them to be equal.

This is what we are referring to as defining how equal is equal.

138. No witness was produced by the Respondent to testify to the merits of a wage adjustment

methodology based on the concept of central tendencies or the suggested methodologies raised

by Respondent counsel in the foregoing paragraph. The Respondent contends, since the sample

from each group in the JUMI Study was to be representative of the work of the occupational

group from which it was drawn, the Tribunal should therefore be able to rely on the measure of

"central tendency" as an estimate of the average of the value of work of all jobs in the

occupational group.

139. Respondent counsel argues the Tribunal is entitled to rely on the sample sizes used in its

"banding mean methodology." According to Respondent counsel these numbers compare to the

sample sizes used by Mr. Sunter in his level-to-segment methodology. That raises a question as

to whether the sample size of the Respondent's deemed group is representative of the whole

population of the deemed group. We note the sample sizes used by Mr. Sunter in the level-to-

segment methodology were not intended to represent whole occupational groups.

140. No witness came forward to verify the sufficiency of the sample sizes for the Respondent's

wage adjustment methodology. Respondent counsel acknowledged this presents a weakness

in their case.

141. During argument Respondent counsel acknowledged a weakness in its methodology

respecting sample size when making submissions about Mr. Gower's report. (see Section IV,

C(i), Paragraph 115). This is found in Volume 241 at p. 32054, line 23 to p. 32055, line 8:

So, it is clear that in our case there is a sample of 1,400 jobs. It's not 600, its not

2,400, its somewhere in the middle. So, according to Mr. Gower, it is probably

not -- 1,400 is not strong enough to make a group-to-group comparison and that is

a weakness. This is a weakness that we are not trying to cover over, it is a

weakness of the whole group methodology in this case. The question is whether

there are any male groups in this case that do have an adequate sample size in

each of those cases.

(ii). Adverse Inference

142. The Alliance claims the Tribunal ought to draw an adverse inference against the

Respondent because the Respondent led no evidence to support its whole group wage adjustment

methodology. The Respondent argues the circumstances of the case and the questions raised by

s. 11 of the Act and s. 14 of the Guidelines do not relate to situations appropriate for drawing an

adverse inference as, for example, where a party is asserting or denying a controversial fact and

fails to call a witness to present its version of the facts in dispute. The Respondent contends

there is no dispute of any fact in issue which would require witnesses from the Respondent or

that would justify drawing an adverse inference against the Respondent.

143. Moreover the Respondent contends the dispute lies in the correct interpretation of the law,

this being an issue which is wholly within the purview of this Tribunal. The Respondent argues

the Tribunal is not entitled to rely on the opinions of pay equity and statistical experts to assist it

in the interpretation of the law. The Respondent submits the Tribunal may, nonetheless, draw

upon their expertise to find the most appropriate methodology on which to apply the principle of

equal pay for work of equal value enacted by Parliament in s. 11 of the Act and s. 14 of the

Guidelines.

144. We find there exists a major difficulty with the employer's wage adjustment methodology

arising from the lack of reliable evidence before us that the data generated by the JUMI Study

supports a methodology of whole group comparators.

145. The Respondent argues that the wage line for each male-dominated occupational group

used in its wage adjustment methodology is a reliable measurement for calculating the wage

gap. But the Respondent has presented little or no evidentiary basis to support its contention that

its wage adjustment methodology is a reliable measurement for calculating the wage gap. Mr.

Sunter's use of relatively small sample sizes for his segmented regression does not, in our

opinion, validate the Respondent's methodology. The evidence demonstrates the sample sizes

used by Mr. Sunter in the level-to-segment wage adjustment methodology were not intended to

represent the range of values of an occupational group but only to summarize the male data

within the range of the female values of the female level being evaluated.

146. The evidentiary problem became apparent during the Respondent's oral submissions. They

were addressed in Volume 245 at p. 32704, line 10 to p. 32709, line 22:

MR. FRIESEN: All right. I'll refer the Tribunal to the evidence, and then I'll

come back to it. The evidence is referred to in Volume 241 of the transcript, and

that was where Mr. Chabursky was making submissions in May and he was asked

a question by the Tribunal about representativity of the evidence. And he gave

references to, in answer to that question, he provided the references to the

evidence and quoted at some length from the relevant references. I'll refer the

Tribunal to that.

The sources are given in this transcript at page 32045, beginning at line 4. Mr.

Chabursky says:

The Chair will recall, and I will take you to the evidence in a moment that

indicates that if positions are chosen on a random basis -- probability sampling is

what they call it -- then they are representative of the population from which they

are taken, but the question is, were they chosen on a random basis? We have

evidence for that in volumes 201 and 203 from Dr. Swimmer.

Before I take you to that, though, it is true that the Joint Initiative did not design

the sample for purposes of comparing groups to groups, and that is the evidence

of Mr. Ranger. He gave that evidence in Volume 204 at page 26343.

Now, before I read on, I should point out that this whole discussion takes up

about 12 pages. It proceeds up to about -- up to page 32056. I wasn't proposing

to read it this morning, but to refer the Tribunal to it.

So we're being consistent. We're saying there is some evidence of

representativity, there's some evidence on which the Tribunal could rely. We

acknowledge that the parties in designing the Joint Initiative did not intend that

the evidence be used for this purpose. We acknowledge that the evidence of

representativity and of sample size is weak, and that if it's controversial, there

would be a basis for challenge. If the other side say, You can't use it for that, then

the Tribunal may be precluded from using it for that purpose.

MEMBER COWAN-McGUIGAN: Have you just summarized the 12 pages?

MR. FRIESEN: No, I haven't summarized it. I'm giving -- having referred to the

evidence -- if the Tribunal wants, we can read through all that. But there it is --

I'm offering the reference to the evidence --

THE CHAIRPERSON: If you could give us just some points, Mr. Friesen, about

where it is in the evidence that, for example, the 26 ships crews would be an

adequate and reliable sample.

MR. FRIESEN: W ell, there's no evidence to that effect, Madam Chair. There is

no evidence to that effect.

THE CHAIRPERSON: So what are you saying then about your samples for the

occupational groups that you want us to use for the deemed group?

MR. FRIESEN: I'm saying that there is some general evidence about

randomness of the sample, and that a random sample can be used to draw

conclusions and make inferences and generalizations about the population from

which it was chosen. There is evidence to that effect. But that is general

evidence. There is no specific evidence that says you can use these samples to

draw -- to make inferences and generalizations about the SC group. There is no

evidence to that effect, I acknowledge that.

THE CHAIRPERSON: Can you use, more specifically, 26 to make a regression

that you can rely on for purposes of --

MR. FRIESEN: There's no evidence that we can do that.

THE CHAIRPERSON: -- the wage gap.

MR. FRIESEN: There's no evidence that we can do that. We do submit,

however, that the sample size is reliable enough for the deemed group.

THE CHAIRPERSON: On what basis?

MR. FRIESEN: The basis is the one that I gave earlier, that we really apply the

principles that Mr. Sunter gave us and the evidence that he gave us about what he

did when he used a regression to draw a wage curve for the purpose of making

calculations.

THE CHAIRPERSON: Yes, but does it follow that if the 26 isn't reliable for the

ship's crew or the 112 isn't reliable for the GS, if you put all those groups

together, you make them a deemed group, that the 345 would be reliable for all

the occupations in that deemed group? Does that follow? If it's not reliable

individually, how can it be reliable as a composite of all of them?

MR. FRIESEN: Because simply put, now we have 345 observations that we have

selected, and we're using them in the same way, exactly the same way as Mr.

Sunter used his observations, his 76, his 47, that didn't represent anything. They

didn't represent anything, and yet he used them to draw regression and calculate a

wage gap.

We're saying the Tribunal can take that evidence and say, Therefore we will do it

for the deemed group and the Respondent is precluded from challenging that

decision.

THE CHAIRPERSON: See, I don't think Mr. Sunter was taking that as

representative of a group. He was just taking some values and saying, These are

representative of values that fall within the average of the female level. That's all

he was doing. He wasn't saying they were representative of anything.

MR. FRIESEN: All right.

THE CHAIRPERSON: So but what you want us to do is say they are -- I think

your sample is that they're representative of those occupations in the deemed

group.

MR. FRIESEN: Well --

THE CHAIRPERSON: And I think that's a little different.

MR. FRIESEN: Well, Madam Chair, what we've done is to meet the

requirements of the Act which says, Identify occupational groups, choose the ones

that are performing work of equal value, and if Guideline 14 is valid, combine

them. Section 15 tells us, Use a wage curve. We're meeting the requirements of

the law.

Now, the Respondent is accepting that the evidentiary basis for doing that is very

weak.

MR. RAVEN: I'm sorry to keep making this request, but I think because this is

the Respondent's methodology that we, as parties that have to reply, are entitled to

a simple "yes" or "no" to the question whether it is the position of the Respondent

that the sample supports a reliable basis for drawing regressions for each of these

groups. It's not enough just to say that it's weak. Either the position of the

Respondent is it's reliable or it's not. And if it's not, then we'll have things to say

in reply. And if it is, we'll have things to say in reply.

MR. FRIESEN: The Respondent's submission is made on the assumption that

the Tribunal will rule that the sample is not sufficiently reliable. [emphasis added]

147. We believe the Respondent is entitled to answer the complaints and to choose to make its

case by relying on the evidence led by the experts who testified before us. On the whole, we

believe this is not a situation where the Respondent is attempting to withhold factual evidence in

dispute. Therefore, the Tribunal finds no reason to draw an adverse inference in these

circumstances and we reject the Alliance's submission in this regard. However the Respondent's

decision not to lead evidence about its wage adjustment methodology leaves the Tribunal in a

quandary. The Tribunal is placed in an untenable position concerning the reliability of the

Respondent's methodology if we are to accept the Respondent's interpretation of s. 11 of the Act

and s. 14 of the Guidelines.

148. Our decision not to draw an adverse inference does not imply an acceptance by the

Tribunal of the Respondent's contention that the opinions of the pay equity and statistical experts

as to the issue before us in this phase of the hearing ("Phase II") should be disregarded or

minimized. The nature and scope of wage adjustment methodology in the context of pay equity,

particularly with large groups of employees, is complex. The experts provided the Tribunal with

analysis and opinions of wage adjustment methodologies for both "direct"and "indirect"

comparisons on an individual to individual basis and also in comparisons involving large groups

of employees. The Tribunal was introduced to the concept of regression analysis, a statistical

formulation, applied in doing indirect comparisons in large group complaints.

149. The Tribunal believes the evidence of the pay equity and statistical experts should be

respected and given careful consideration due to the highly complex nature of the subject

matter. The Tribunal is not prepared to circumscribe the expert testimony which was presented

in regard to wage adjustment methodology applicable in indirect comparisons for large groups of

employees. We believe the interpretation of the principles found in the legislation must be made

with due regard for the statistical, technical and scientific disciplines which the application of

these principles entail. The Tribunal is entitled to the benefit of that evidence when considering

the issues before us in Phase II.

150. The Tribunal is of the view that the selection of a wage adjustment methodology requires,

for the size and complexity of the complaints before us, the assistance of experts trained in pay

equity, compensation and statistics. At the same time we recognize the methodology chosen by

the Tribunal must meet the requirements of s.11 of the Act.

V. THE EMPLOYER'S CLASSIFICATION SYSTEM

151. The Respondent is advocating an interpretation of the Act and Guidelines that occupational

groups form the basis of comparison in group complaints. Occupational groups are groups of

employees within the Respondent's classification system. Mr. Sadler testified about the

development of the Respondent's classification system. He testified that in 1962 the Public

Service was known as the Civil Service which operated in a totally non-union environment. The

classification structure that existed at the time was very complex with three classes of public

servants. According to Mr. Sadler, there was the "continuing indeterminate" public servant who

fell under the Civil Service Regulations and the Civil Service Act; the "prevailing rate" employee

covered under separate legislation from the Civil Service Act; and the Ships Officers and Ships

Crews also covered by separate legislation.

152. Mr. Sadler testified there were approximately 700 classes of indeterminate public servants

and these classes were subdivided into grades with each class having its own pay scale. The

"prevailing rate" employees were mostly in traditionally blue-collar employment. They were not

salaried employees but wage rated, that is, paid an hourly rate rather than an annual or monthly

rate. Mr. Sadler testified the "prevailing rate" employees were wage rated in accordance to the

wages being paid for other individuals doing the same job in the same location so that, for

example, government carpenters were paid in accordance with local carpenter rates. (see

Section IX).

153. According to Mr. Durber prior to the introduction of the current classification system there

were over 2,000 different job classifications in the Public Service. He testified a simplification

of the classification structure took place between 1966 and 1971.

154. Mr. Durber testified that in July 1965 a report was submitted to Prime Minister Pearson

from the Chairperson of the Preparatory Committee on Collective Bargaining in the Public

Service. The Preparatory Committee was appointed in August 1963 to make preparations for the

introduction into the Public Service of an appropriate form of collective bargaining and

arbitration. The Committee's mandate also included the duty to examine the need for reforms in

the system of classification and pay structure applicable to all civil servants (Exhibit HR-

21). The Committee proposed a classification system comprised of occupational categories and

groups. The resulting system is essentially the system that exists today.

155. The Report proposed two levels of groupings for employees. The first level of grouping

would encompass six major occupational categories. The Committee conceived this

categorization to be a broad horizontal division of the Federal Public Service useful for planning

and development of personnel policy. It was to consist of occupational groups linked together in

a general way by educational requirements and a common approach to classification and pay

administration.

156. The second level of grouping would be the occupational group forming a subdivision

within an occupational category. It was intended that each occupational group be composed of

employees with similar skills, performing similar kinds of work and bearing a relationship to an

identifiable outside labour market wherever possible. Each occupational group would have its

own pay plan so that its rates of pay could be adjusted independently in response to changes

outside the Public Service.

157. The report of the Committee identified six occupational categories and 67 occupational

groups. The Committee concluded this system of classification and pay would be able to

respond with flexibility, without loss of integrity, to a number of pressures and

requirements. Two of the pressures identified were the outside labour market and considerations

of equity flowing from evaluations of the relative worth of jobs within an organization.

158. According to Mr. Sadler the rationale behind the proposed structure was to group into six

occupational categories areas of employment that were similar in a broad context. Mr. Sadler

testified the categories were fine-tuned and as a result approximately 66 smaller occupational

groups were formed. We note that occupational groupings from five of the six occupational

categories participated in the JUMI Study conducted by the JUMI Committee. The sixth

category, not included, was the Executive (EX) category.

159. It is noted the current classification structure has two further levels of grouping that were

not identified by the Report of the Preparatory Committee. Some of the occupational groups

were further designated into sub-groups and/or levels where applicable. The male-dominated

occupational groups tended to comprise a smaller number of employees than the female-

dominated occupational groups. The system included more sub groups in the male-dominated

occupational groups than in the female-dominated occupational groups.

160. Most of the occupational groupings have both sub-groups and levels, a level being a

smaller subdivision than a sub-group. However not all of the occupational groups have sub-

groups. The Tribunal was not provided with any rationale for the existence of sub-groups and

levels or how they were formulated. We note, for instance, the largest occupational group, the

CR occupational group, has no sub-groups but is subdivided into seven separate levels. The CR

occupational group is a female-dominated group composed of approximately 50,000 employees

and is one of the complainant groups before the Tribunal.

161. Following the passing of the Public Service Staff Relations Act, 1966-67, c.72, s.1 and the

implementation of collective bargaining, pay rates were established through collective

bargaining for the positions within an occupational group. Pay rates were negotiated between

the bargaining agents and the Treasury Board for each of the bargaining units established

pursuant to that Act. Pay rates are associated with the levels in an occupational group, if there is

one, or with the sub-group if there are no levels.

162. For the most part the bargaining units have continued to mirror the Respondent's

classification structure of occupational groups. Each of the occupational groups had its own

classification standard or job evaluation plan. There is no uniform job evaluation plan within the

Federal Public Service. That is part of the reason for the necessity of using the Willis Plan in the

JUMI Study. According to Mr. Durber job classification has been delegated to management

personnel within a particular government unit or department. Mr. Sadler testified most

departments of government have personnel branches staffed by a group of classification

specialists whose duty it is to evaluate jobs. A classification standard is used to determine which

group a particular job belongs.

163. Mr. Durber testified the classification standards are established on the authority of the

employer who may choose to consult with the bargaining agents. But the ultimate authority rests

with the employer who decides under which occupational group and level a job should fall.

164. There are two sub-groups of one occupational group which the Commission has requested

the Tribunal treat as distinct and separate occupations for purposes of wage adjustment. This

arises within the Data Analysis (DA) occupational group which has two sub-groups, the Data-

Conversion (DA-CON) sub-group and the Data-Production (DA-PRO) sub-group. Overall the

gender predominance of the DA group is female. Mr. Sunter's calculations using the level-to-

segment methodology produced a positive adjustment for the DA-CON sub-group and a negative

adjustment to the DA-PRO sub-group. For purposes of calculation Mr. Sunter split the DA

occupational group into two separate groups. His calculations demonstrated one sub-group was

clearly male-dominated and the other was clearly female-dominated.

165. Mr. Durber testified Mr. Sunter's calculations demonstrate the distinctiveness of the wage

structure between these two sub-groups which arises because of the distinct work of each sub-

group. According to Mr. Durber, the DA-CON sub-group is female-dominated and their work

involves keyboarding which is not dissimilar in nature to the work of a typist. In contrast the

DA-PRO sub-group is male-dominated and their work is quite distinct from

keyboarding. Because of the distinctiveness of the work the Commission advocates the DA-

CON sub-group and the DA-PRO sub-group be treated as separate occupations for purposes of

wage adjustment. Mr. Durber testified this treatment falls within the Commission's view of the

objective of s. 11 of the Act, that is to say, particular attention ought to be given to traditionally

female work.

166. Mr. Sunter's calculations also demonstrated a small "pocket of maleness"within level 2 of

the Court Reporter (COR) sub-group, a sub-group in the Secretarial, Stenographic & Typing

(ST) group. Mr. Sunter's calculations triggers a small adjustment for the first level of the COR

sub-group and a negative adjustment for the second level. The Commission is not proposing to

split the two levels into separate groups but requests that level 2 not receive any

adjustment. Applying Statistics Canada characteristics of occupational groups Mr. Durber's view

is that the ST-COR sub-group should be treated as a separate occupation for wage

adjustment. The ST-COR sub-group comprises employees hired as court reporters. Mr. Durber

sees this work as distinct from other secretarial work.

167. We note the Government of Canada is in the process of simplifying the job classification

system in the Federal Public Service through an initiative entitled PS2000. A task force has been

mandated to examine the design and administration of the existing classification system within

the context of values and objectives for a renewed public service. Some details of the initiative

are documented in the Executive Summary of a report entitled "Public Service 2000: Report of

the Task Force on Classification and Occupational Group Structures", (Exhibit PSAC-60). The

report is dated July 20, 1990 and refers to the conclusion of the Task Force made in an earlier

preliminary report dated January 31, 1990, of the need to reform the existing classification

system by a pragmatic and reasonable approach. There is, according to the Task Force, also a

need for significant reduction in the number of occupational groups and levels. Within the

context of that conclusion, the report reads at p. 1:

The new system must enhance career/work development and job enrichment.

It must be free of any systemic discrimination and sexual bias.

[emphasis added]

168. Further along, the Task Force Report reads as follows at p. 2:

The system must be consistent with the requirements of human rights legislation that introduced

the concept of "equal pay for work of equal value". This concept requires that predominantly

male groups and female groups working in the same establishment receive equal pay for work of

equal value although the work may be different in nature. This obviously places great emphasis

on internal relativity as the underlying principle of compensation. The Task Force feels that its

recommendation to combine groups and reduce levels will greatly help the pay equity

question. However, the existing classification system does not respond to pay equity

issues. Because of the number of classification standards and plans, cross-group comparisons

required to deal with pay equity matters cannot be made unless a common classification plan is

used.

The answer lies in a new rating plan which would support the concept of equal

pay for work of equal value and rely on internal relatively as the underlying

principle of compensation in the Public Service.

For each new occupational group four factors will be present in the rating plan

and which will be consistent with the requirements of the Canadian Human Rights

Act. Within a new occupational group, under each factor the rating scale and its

contents will be structured to measure the significant differences that are

applicable to the group. The rating scales and their description can therefore vary

from group to group. [emphasis added]

169. A report and a reference guide has been prepared for public sector employees by the

PS2000 Classification Simplification Task Force. These were tendered into evidence during the

cross-examination of Mr. Durber. They describe the concept of gender neutrality in the

preparation of work descriptions. Also tendered into evidence is a third document entitled

"Work Description Substantiating Data - A Writer's Check List", published October 5, 1992, to

be used by all employees under the new system. Mr. Durber agreed with counsel for the

Alliance that this later documentation confirms the PS2000 initiative and the proposed changes

to the classification plan are, in part, a recognition and admission that the existing classification

system fails to comply with the provisions of the Act. Respondent counsel sought clarification

of his response and in particular to the above noted passages from the Report. Mr. Durber

testified as follows in Volume 155 at p. 19324, line 1 to p. 19325, line 8:

THE WITNESS: Mr. Friesen, I would read all of this passage in the light of what is highlighted

and that is at the beginning where it says the system must be consistent and I would assume that

one has to read that it is not now consistent. I hope you will agree with me that it isn't consistent.

MR. FRIESEN:

Q. I am going to suggest to you that ---

A. Why else would they say it in this passage would be my question.

Q. They are addressing the question. But I am going to suggest to you -- and I

don't think we need to debate this to death.

A. Indeed not.

Q. -- but I am going to suggest to you that all that this passage means is that you

can't use the existing standards to determine whether it's consistent with the

Act. It doesn't say it's inconsistent or violates the Act.

A. And clearly the Commission isn't proposing the UJEP [Universal Job

Evaluation Plan] as a remedy at this point at all.

Q. That wasn't my question. My question was whether you are able to point to

anything in any of these passages that amount to an admission that the existing

classification standards violate section 11.

A. I would say the lead in to that paragraph is pretty close to doing that.

Q. We will leave that for argument, then.

The Tribunal received no further evidence concerning PS2000 or submissions

concerning the testimony of Mr. Durber.

170. The organization of work into groupings of traditionally male and female work and the

creation and implementation of different classification plans (job evaluation plans) for the

various occupational groups in the Federal Public Service resulted in concerns even before the

government's proactive initiative in 1985. Prior to the JUMI Study the Alliance took issue with

representatives of the Treasury Board concerning the lack of compliance of some of the

classification standards with the requirements of s. 11 of the Act. (see Section X, A).

VI. SECTION 11 OF THE CANADIAN HUMAN RIGHTS ACT

A. Systemic Discrimination

171. Since the Tribunal's Phase I decision it has had the benefit of the decision of the Federal

Court of Appeal in Public Service Alliance of Canada v. Staff of the Non-Public Funds,

Canadian Forces et al. (1996), 199 N.R. 81 (F.C.A.).

172. In an appeal from the decision of a Human Rights Tribunal on the question of retroactive

wage adjustment, the Federal Court of Appeal, in Non-Public Funds, supra, ventured to elaborate

on the nature of systemic discrimination under s. 11 of the Act. This is the first time since the

decision of the Supreme Court of Canada in Canadian National Railway, supra, that a Superior

Court expanded and elaborated on the comments of Chief Justice Dickson with respect to

systemic discrimination.

173. In Non-Public Funds, supra, a complaint was filed with the Commission on February 12,

1987 alleging the employer was not paying female employees wages equal to those paid to

certain male employees performing work of equal value contrary to ss. 7 and 11 of the Act. Prior

to the complaint being heard by the Tribunal the parties agreed to resolve the s. 11 complaint by

adjusting pay rates in the complainant group in accordance with a proposal made by Dr. Weiner,

the same expert who testified before this Tribunal. The issue before that Tribunal was whether

there should be a retroactive wage adjustment for a specified time period which was to

commence one year prior to the filing of the original complaint. The Tribunal had concluded

that no pay adjustment should be made for any part of the period in question and this decision

was upheld by the Federal Court Trial Division. That decision was then appealed to the Federal

Court of Appeal.

174. Hugessen J.A., writing for the majority, embarked on a discussion of the nature of systemic

discrimination in circumstances wherein the employer admitted having engaged in a

discriminatory practice that was systemic in nature. In view of those admissions Mr. Justice

Hugessen reasoned that an understanding of the phenomenon of systemic discrimination was

critical to an appreciation of the issue of retroactivity. He noted that the employer's job

classification system, which lay at the root of the pay equity problem, had existed since 1986 and

the discrimination resulted from a system which had undervalued women's work.

175. Hugessen J.A. reviewed the decision of Dickson C.J. in Canadian National Railway, supra.

That case dealt with the subject of systemic discrimination in the context of employment

equity. Referring to Dickson C.J.'s remarks concerning the Abella report, a study of systemic

discrimination in Canada on equality in employment, Hugessen J.A. quoted from the following

passage by the Chief Justice at p. 87:

Later in the same judgment, the Chief Justice returned to the subject and stressed

the historical, attitudinal and continuing nature of systemic discrimination:

I have already stressed that systemic discrimination is often unintentional. It

results from the application of established practices and policies that, in effect,

have a negative impact upon the hiring and advancement prospects of a particular

group. It is compounded by the attitudes of managers and co-workers who accept

stereotyped visions of the skills and "proper role" of the affected group, visions

which lead to the firmly held conviction that members of that group are incapable

of doing a particular job, even when that conclusion is objectively false. An

employment equity program, such as the one ordered by the Tribunal in the

present case, is designed to break a continuing cycle of systemic discrimination.

176. Hugessen J.A. then quoted at length from the 1991 Human Rights Tribunal decision, in

regard to the HS complaint filed by the Alliance in 1981. That Tribunal had dealt with the

subject matter of systemic discrimination. (see Public Service Alliance of Canada v. Treasury

Board (1991), T.D. 4/91 (C.H.R.T.)). That Tribunal's comments are found at p. 88 of Justice

Hugessen's decision:

The concept of systemic discrimination is perhaps as hard to define as such discrimination is to

identify. It is not identical in concept to indirect or adverse impact discrimination. Adverse

impact discrimination involves requirements which do not, on their face, discriminate on a

prohibited ground, but which affect a group identifiable on a prohibited ground in such a way as

to have a discriminatory effect on that group.

While adverse impact discrimination may be quite subtle in its operation, often the effect is

fairly obvious. Most people today, for example, recognize that minimum height and weight

requirements discriminate against women. Similarly, it takes only a fairly rudimentary

knowledge of religious diversity to realize that a hard hat requirement will adversely affect one

particular religious group.

The concept of systemic discrimination, on the other hand, emphasizes the most

subtle forms of discrimination, as indicated by the judgment of Dickson, C.J. in

CN v. Canada (Human Rights Commission), [1987] 1 S.C.R. 1114, at 1138-9. It

recognizes that long-standing social and cultural mores carry within them value

assumptions that contribute to discrimination in ways that are substantially or

entirely hidden and unconscious. Thus, the historical experience which has

tended to undervalue the work of women may be perpetuated through

assumptions that certain types of work historically performed by women are

inherently less valuable than certain types of work historically performed by men.

177. Following the reference to both Canadian National Railway, supra and the Tribunal's Phase

I decision, supra, Hugessen J.A. comments as follows at p. 88:

It is arguable, indeed, that the type of discrimination which pay equity is designed to counteract

is always systemic. Thus, Weiner and Gunderson say:

Regardless of what it is called, pay equity is designed to address a kind of

systemic discrimination. Systemic discrimination is found in employment

systems. It is the unintended byproduct of seemingly neutral policies and

practices. However, these policies and practices may well result in an adverse or

disparate impact on one group vis-à-vis another (e.g., on women versus

men). This differs from interpersonal discrimination where one individual

discriminates against another. Pay equity requires changes to pay systems to

ensure that women's jobs are not undervalued.

[emphasis added]

178. He concludes his coverage of this topic at p. 88 with the following comments:

Systemic discrimination is a continuing phenomenon which has its roots deep in history and in

societal attitudes. It cannot be isolated to a single action or statement. By its very nature, it

extends over time. [emphasis added]

(i). Concept of Causality

179. The Respondent refers to the Tribunal's Phase I decision, supra, to support its submissions

that the concept of causation is a requirement of, or is to be read into, the provisions of s. 11 of

the Act. Section 11 itself is not predicated upon nor does it invoke the idea of causation. We

now find it necessary to review briefly the earlier decision of this Tribunal on which the

Respondent relies.

180. The issues arising from the complaints before this Tribunal were separated into three

segments or phases. The Respondent's position in this phase, in respect to causation, has shifted

from its earlier position which it held in Phase I. In Phase I the Respondent submitted the job

evaluation results generated by the JUMI Study were unreliable for purposes of this

adjudication. The Respondent alleged the job evaluation results were biased in that the

individuals who performed the evaluations were treating the male-dominated employee

questionnaires and the female-dominated employee questionnaires differently from the Willis

consultants, who were involved in the study and who had evaluated a sampling of the same

questionnaires as the committees. (see Phase I decision, supra).

181. The issue before the Tribunal at that time was whether the job evaluation results obtained

from the questionnaires in the JUMI Study were reliable for purposes of the s. 11

complaints. The Commission had used these job evaluation results to conclude the wage gap had

not been closed as a result of the Respondent's unilateral adjustments of 1990.

182. The Commission and the Alliance requested the Tribunal accept the evaluation scores as

evidence of the "value" of work. They contended the evaluation results were sufficiently reliable

to establish the equality of work between male and female employees in the s. 11

complaints. Anecdotal and statistical evidence was led by the parties to explain these differences

between Willis consultants and the job evaluation committees who had evaluated the same

questionnaires. The position of the Commission and the Alliance was that the differences in

evaluation scores arose from factors other than gender.

183. The Respondent sought a broad interpretation of the phrase "sexual bias" referred to in s.

9(a) of the Guidelines which it alleged affected the treatment of questionnaires by the evaluation

committees. (see Section I, C, Paragraph 18 for s. 9(a)). The Respondent's position then was

that the proper interpretation of "sexual bias" did not flow from or include the concept of

causality.

184. Each of the parties composed separate questions for the Tribunal to address concerning the

meaning of "sexual bias". The Respondent posed its question as follows: Is there a pattern of

different treatment of male and female questionnaires? The Commission and the Alliance were

dissatisfied that a pattern of differential treatment was all that was necessary to show "sexual

bias" in the evaluation results and formulated their question with a causality provision, as

follows: "Is there a pattern, a systematic variance of different treatment of male and female

questionnaires (in the evaluation process) that was caused by or is attributed to gender-bias or

gender-related bias." [emphasis added]

185. At that time the Respondent sought a meaning of "sexual bias" which eliminated the

necessity for cause. The Respondent's position is summarized by the Tribunal in its Phase I

decision, supra, at p. 32, paragraph 111:

111. In formulating the question for the Tribunal to address, Respondent

Counsel argues that their formulation does not require a causative factor for the

different treatment of male and female questionnaires. The disagreement between

the parties lies not in assigning a broad meaning to the words "sexual bias" but

instead arises as to whether s. 11 requires the existence of a cause when different

treatment of male and female questionnaires is found or whether, on the other

hand, it is simply a matter of differential treatment of male and female jobs

without the necessity of assigning cause. In support of the Employer's

submission, they rely on a meaning of bias which in their view does not require a

causal link or relationship under s. 11 of the Act. [emphasis added]

186. In ascertaining whether a causal relationship existed within the meaning of "bias",

the Tribunal canvassed the meaning of a "wage gap" which s. 11 of the Act is intended to

correct. The Tribunal referred to the expert testimony of Dr. Armstrong, an expert in job

evaluation and pay equity. Dr. Armstrong had made significant comments on the overall wage

gap prevalent in rates of pay earned by females as compared to males. The Tribunal proceeded

to differentiate Dr. Armstrong's description of the overall wage gap to that arising under s. 11 of

the Act.

187. The Tribunal then stated at p. 23:

85. A wage gap is not something clearly delineated. The Tribunal recognizes

that salary differentials between male and female jobs can be a function of job

requirements making some jobs intrinsically more valuable to the employer than

other jobs. Such differentials are in contrast to differentials which are based

entirely on gender differences and it is the latter resulting wage gap which the

Tribunal believes s. 11 is intended to eliminate.

188. Now, in Phase II, the Respondent relies on paragraph 85 at p. 23 of the Tribunal's earlier

decision to support its submission that causation is a factor required under s. 11 of the Act. The

Respondent seeks further support from paragraph 99 of the Tribunal's decision at p. 28:

99. We must be assured the complaints seek to redress a wage gap based on wage differentials

that are gender based and not resulting from other factors. It seems apparent that the existence of

a wage gap per se is not proof of discrimination. To hold otherwise would negate the entire

evaluation process which has, as its purpose, the comparison of jobs according to a plan or

system for rating work according to the criteria prescribed in s. 11(1) of the Act. [emphasis

added]

100. We also find s. 11 is designed to eliminate economic inequality created by gender based

wage discrimination. The discrimination is unintentional as the decision of Dickson C.J. in the

CN case, supra, makes clear. It is nevertheless a subtle form of discrimination built into

employment practices as they have existed over the years since females have become

contributors to the work force. We recognize from the expert testimony of Weiner, Armstrong

and Willis that systemic discrimination operates in systems and becomes incorporated into the

wage setting practices of organization and that classification of jobs may be the by-product of

systemic discrimination. Since systemic discrimination is part of a system never designed to

discriminate, Weiner says that it cannot be corrected instantly nor can pay equity be achieved

quickly.

189. In addressing the interpretation of s. 11 of the Act, the Respondent refers to paragraph 131

of the decision found at p. 37 in support of the notion of causation under s. 11 of the Act:

131. The wage gap to be redressed by s. 11 must be caused by gender based

discrimination. Section 9(a) of the Guidelines is subordinate to the enabling legislation, the Act,

and is authorized by s. 27(2) of that Act. There is a presumption in favour of the validity of

regulations in the light of their enabling statute. In the Interpretation of Legislation in Canada,

2nd Edition, Pierre Andre-Côté at p. 310 the learned author comments as follows:

Finally it must be pointed out that the regulations are not only deemed to remain intra vires, but

also to be formally coherent with the enabling statute.

190. In the final analysis the Tribunal found, in its Phase I decision, supra, the question posed by

the Respondent was restrictive when taken in the context of the reliability of the JUMI Study

results. Because of the opinions expressed by the pay equity expert, Mr. Willis, and the gender-

based nature of the discrimination intended by s. 11 of the Act, the Tribunal held the difference

in treatment of the male and female questionnaires had to be gender-based. Therefore, the

Tribunal held the differential treatment of the questionnaires by the evaluation committees had to

be gender-related or gender-based within the intent of s. 11 of the Act to render the evaluations

unreliable. The Tribunal formulated the question to be addressed as follows: Is there a different

treatment of male and female questionnaires in the evaluation process that was caused by or

attributed to gender-bias or gender-related bias? [emphasis added]

191. The Tribunal acknowledges the notion of causation, which arose during the debate on the

issue of gender bias as it relates to the concept of systemic discrimination and to the problem

posed by the differential treatment by the committees and by the consultants of the job

questionnaires, has now provided the Respondent the opportunity of introducing causation as a

factor to be included in the interpretation of s. 11 of the Act. This argument has been advanced

despite the Respondent's earlier submission that the notion of causation was not a factor to be

considered in the treatment of male and female questionnaires. Also, the Respondent then

argued it was not necessary to establish a causal link or relationship when addressing the

meaning of "bias" within the context of the legislation.

192. The Respondent relies on the causation argument to support its preferred method of wage

adjustment methodology, that is, to use the lowest paid "male" occupational group as a

comparator. The Respondent claims this methodology is aimed at redressing only gender-based

discrimination. The Respondent cautions the Tribunal against including in the male comparator

group other higher paid "male" occupational groups performing work of equal value to the

"female" complainant occupational group. The Respondent contends the effect of including

other male-dominated occupational groups will widen the wage gap because the wage gap will

then be caused by factors other than gender-based discrimination, contrary to s. 11 of the Act.

193. The Tribunal's earlier decision rendered in Phase I must be read in the context of the issue

then before the Tribunal. There were differences in evaluation scores between the Willis

consultants and the job evaluation committees. The Tribunal was asked to decide if it was

necessary to account for these differences within the meaning of "bias" intended by s. 11 of the

Act and the Guidelines. This Tribunal needed to examine closely the intent of s. 11 of the Act

and what it aims to remedy.

194. The reference to the wage gap was used in a descriptive sense to distinguish the problems

s.11 was addressing from other issues raised by Dr. Armstrong and about how the educational,

scientific, economic and historical spectrum of the work place in which females continue to play

increasingly important roles has affected their pay.

195. The Tribunal is of the opinion the notion of causality may be appropriate in other

situations. It is not, in our opinion, appropriate when the discrimination complained of is

systemic in nature. The use of the phrases "caused by gender bias" and "caused by gender-

related bias" needs to be read in the context of the issues then before the Tribunal. Without

endangering the Tribunal's meaning and taken in context, it is apparent that s. 11 of the Act is

premised on gender difference.

196. This would avoid any misunderstanding arising from the Tribunal's use of those phrases in

its previous ruling. It would also put to rest the sophistry of the Respondent's submissions on

this issue.

197. It is noteworthy that the Tribunal, in its Phase I decision, supra, did not elaborate on the

evidentiary basis necessary to establish a wage gap under s. 11 of the Act. Accordingly the

Tribunal will now examine the problem from the perspective of what constitutes the essential

elements which must be proven to support a prima facie case of discrimination under s. 11 of the

Act.

(ii). History of Section 11 of the Canadian Human Rights Act

198. Mr. Durber, an acknowledged expert in pay equity and job evaluation generally, provided

the Tribunal with information relating to the historical background, motivation and eventual

implementation of the Act. More specifically he testified to the meaning and effect given to s.

11 of the Act in the Commission's approach to the application of that section. According to Mr.

Durber the historical context begins with the International Labour Organization (ILO), an organ

of the United Nations. The ILO passed a Convention dated June 29, 1951 entitled "Equal

Remuneration Convention" (Volume 145, p. 17933). Article II of the Convention, at p. 104,

provides for international recognition of the principle of equal remuneration for male and female

workers for work of equal value. The Convention encouraged "measures" be taken in order to

achieve that principle in practice.

199. The next significant event towards the implementation of s. 11 of the Act, according to Mr.

Durber, concerned the report of the Royal Commission on the Status of Women in Canada

published on September 28, 1970. That report identified occupational segregation as one of the

reasons for women's lower earnings. Furthermore it found that predominantly female

professions tended to be paid less than those which were predominantly male.

200. The introduction of a human rights bill was referred to in the Speech from the Throne on

October 12, 1976 (Exhibit PIPSC-82). The bill was undertaken on behalf of the Government of

Canada to prohibit discrimination on specified grounds. In particular a portion of the bill was

intended to establish the principle of equal compensation for work of equal value. An extract

from the Speech reads as follows:

...In a similar effort to remove obstacles to information and to equal opportunity,

the government will introduce a Human Rights Bill. The major effect of the bill

will be to prohibit discrimination on the grounds of race, colour, national or ethnic

origin, religion, age, sex, marital status, or physical handicap. In particular, the

Bill will establish the principle of equal compensation for work of equal value

performed of persons of either sex... [emphasis added]

201. The Tribunal was provided with a copy of the minutes of proceedings and evidence of the

Standing Committee on Justice and Legal Affairs convened in 1977 while Parliament was

considering Bill C-25, the Canadian Human Rights Act. (Exhibit HR-236). The then Minister of

Justice and Attorney General of Canada, the Honourable S.R. Basford, appeared before the

Committee and spoke as follows:

...[W]e should legislate the principle and through the Commission and through its efforts at

setting out guidelines, solve those problems, presumably of definition and application. To this

end, the Commission convened a task force in 1977 known as "equal wages task force" to study

and report on how value ought to be defined. The task force reported in 1978. It concluded, that

on a broad basis, job evaluation practices were sufficiently widespread to afford some guidance

and definition for the Commission, and that it ought to be possible to measure, albeit

subjectively, the value of work as required under s. 11 of the Canadian Human Rights Act.

[emphasis added]

202. According to Mr. Durber's testimony it was the ILO Convention and the Royal

Commission on the Status of Women's Report that lead directly to the enactment of the Act.

203. The Act was assented to on July 14, 1977. However, s. 11 was only proclaimed in force

on March 1, 1978. In December 1979 the General Assembly of the United Nations adopted the

earlier UN Convention on the elimination of all forms of discrimination against women, which

included wage inequality. (Exhibit HR-237).

204. Mr. Durber further testified the terms of s. 11 found in the Act are consistent with the terms

of article 11 of the ILO Convention, (Exhibit HR-237). Article 11 of the convention reads:

Article 11

1. States Parties shall take all appropriate measures to eliminate discrimination

against women in the field of employment in order to ensure, on a basis of

equality of men and women, the same rights, in particular:

(d) The right to equal remuneration, including benefits, and to equal treatment

in respect of work of equal value, as well as equality of treatment in the

evaluation of the quality of work;

[emphasis added]

B. Prima Facie Case of Discrimination

205. In Phase I the Tribunal identified four elements required to satisfy the legal burden on the

Commission and the Alliance in order to prove a prima facie case of discrimination under s. 11

of the Act. Based on the provisions of s. 11 of the Act, the companion Guidelines and the state

of the pleadings, the four elements were listed by us in Phase I decision, supra, at p. 46,

paragraph 165 as follows:

(i) The complainant groups are female-dominated within the meaning of the

Equal Wages Guidelines;

(ii) The comparator groups are male-dominated within the meaning of the Equal

Wages Guidelines;

(iii) The value of work assessed is reliable; and

(iv) A comparison of the wages paid for work of equal value produces a wage

gap.

206. There is no outstanding issue with respect to elements (i) and (ii). The parties agreed

before this Tribunal the groups that were included in the JUMI Study are female-dominated

occupational groups and male-dominated occupational groups within the requirement of s. 13 of

the Guidelines. That section prescribes the criteria which defines sex predominance (by

gender). The Tribunal addressed the third element in Phase I and found the job evaluations were

reliable for purposes of wage gap calculations.

207. Therefore wage differences, which arise from a comparison of the wages paid for work of

equal value which produces a wage gap listed as item (iv) above, are the focus of this decision

because it is the only remaining issue required to establish a prima facie case of discrimination.

208. It is common ground in complaints under the Act that the Complainant bears the initial

onus of establishing a prima facie case of discrimination following which, if proven, the burden

lies with the Respondent to establish justification for the discriminatory treatment. (see Ontario

Human Rights Commission et al. v. Borough of Etobicoke, [1982] 1 S.C.R. 202 and Simpsons-

Sears, supra).

209. In Simpsons-Sears, supra, a prima facie case was defined by McIntyre J. at p. 558 as "one

which covers the allegations made and which, if they are believed, is complete and sufficient to

justify a verdict in the complainant's favour in the absence of an answer from the respondent-

employer". In general, according to Phipson on Evidence, 14th ed. (London: Sweet & Maxwell,

1990), paragraph 4-10(b) et seq. the rule which applies is "he who invokes the aid of the law

should be the first to prove his case." This rule is founded on considerations of good sense and

as well, the general observation that, in the nature of things, a negative is more difficult to

establish than an affirmative. (see Robins v. National Trust Company [1927] A.C. 515

(PC)). Referring again to Simpsons-Sears, supra, it is helpful to refer to the following passage as

per McIntyre J. at p. 558:

To begin with, experience has shown that in the resolution of disputes by the

employment of the judicial process, the assignment of a burden of proof to one

party or the other is an essential element. The burden need not in all cases be

heavy -- it will vary with particular cases -- and it may not apply to one party on

all issues in the case; it may shift from one to the other. But as a practical

expedient it has been found necessary, in order to insure a clear result in any

judicial proceeding, to have available as a 'tie-breaker' the concept of the onus of

proof...Where adverse effect discrimination on the basis of creed is shown and

the offending rule is rationally connected to the performance of the job, as in the

case at bar, the employer is not required to justify it but rather to show that he has

taken such reasonable steps toward accommodation of the employee's position as

are open to him without undue hardship. It seems evident to me that in this kind

of case the onus should again rest on the employer, for it is the employer who will

be in possession of the necessary information to show undue hardship, and the

employee will rarely, if ever, to show its absence.

210. The burden of proof in respect to element (iv), that is a comparison of wages paid for work

of equal value which produces a wage gap, rests on the Alliance and the Commission. To satisfy

that burden the Alliance and the Commission are required to prove, on a balance of probabilities,

that the complainant has been discriminated against pursuant to the provisions of the Act and in

particular by its treatment of the female employees who are employed in the same establishment,

contrary to the provisions of s. 11 of the said Act.

211. L'Heureux-Dubé J., in her dissenting opinion, in the Supreme Court of Canada decision,

Syndicat des employés de production du Québec et de l'Acadie v. Canada (Canadian Human

Rights Commission), [1989] 2 S.C.R. 879, canvassed dual concepts of "value" and "equality"

under s. 11 of the Act. Madam Justice L'Heureux-Dubé remarked that the application of the

principle of "equal pay for work of equal value" in s. 11 of the Act gives rise to considerable

difficulty and that difficulty lies in the dual concept of "equality" and "value". She stressed the

concept of "equality" ought not receive a technical or restrictive interpretation. Her Ladyship,

however, did not elaborate on the meaning of equality in terms of quantitative comparisons

forming a basis for wage adjustment methodology. Neither were quantitative comparisons

elaborated on by Sopinka J. who delivered the majority opinion.

212. In S.E.P.Q.A., supra, the Commission had received a s. 11 complaint alleging the

employees in the Section fabrication et Manipulation des décor of the CBC, who were

predominantly male, were paid more for work of equal value than employees in the Section

fabrication et Manipulation des costumes, who were predominantly female. After the

Commission conducted its investigation it dismissed the complaint under s. 36(3) of the

Act. The Supreme Court was asked to determine whether the decision of the Commission was

one that was required to be made on a judicial or quasi-judicial basis within the meaning of s. 28

of the Federal Court Act and, if so, whether the Commission committed a reviewable error.

213. The majority of the Court held the Commission's decision was not one that was required to

be made on a judicial or quasi-judicial basis. In dismissing the appeal, Sopinka J. also held that

the Commission correctly applied s. 11 of the Act to the facts of the case.

214. Within the context of the Commission's investigation Mr. Justice Sopinka described the

application of generally accepted job evaluation techniques to assess the relative value of the

jobs in issue. He outlined a three-step process used by the Commission. Without further

elaboration he listed as one of the steps, a quantitative comparison of the relative value of

jobs. He writes at p. 887:

The investigation of an equal pay complaint requires the application of generally

accepted job evaluation techniques to measure the relative value of the jobs in

issue. This process is a three-step procedure:

1. The investigator must gain a thorough understanding of the job content of

each job. This information is obtained from up-to-date job descriptions or

position specifications obtained from the employer, and when there is doubt, from

the incumbents of the job who are asked to complete the questionnaires. These

are known as job fact sheets.

2. The jobs are then evaluated using a job evaluation plan. The plan will contain

techniques used to measure job content according to factors and criteria specified

in the plan. This permits a quantitative comparison of the relative value of the

jobs. [no details provided]

3. The quantitative measures of job value are then co-related to appropriate

levels of compensation.

[emphasis added]

215. We note Madam Justice L'Heureux-Dubé devoted a section in her dissent to "prima facie

discrimination" under s. 11 of the Act. The argument raised by the Appellant before her

Ladyship was that the job segregation at the CBC established of itself a prima facie case of

discrimination. The bargaining unit was composed of a minority of female employees who

occupied certain "female jobs" and claimed to be paid less than employees occupying other

"male jobs" in the unit despite similar working conditions and objectives. The Appellant argued

proof of job segregation established prima facie that the wage disparity was discrimination based

on gender.

216. During the course of her analysis on the legal definition of wage discrimination under s. 11

of the Act, L'Heureux-Dubé J. recognized the direction of the Supreme Court of Canada in

establishing that intent is not a prerequisite element in finding adverse discrimination. (see

Robichaud v. Canada (Treasury Board), [1987] 2 S.C.R. 84 and Canadian National Railway,

supra). In noting at p. 924 that statistical evidence of professional segregation "is the most

precious tool in uncovering adverse discrimination" under ss. 7 and 10 of the Act, her Ladyship

stated that the scope of protection provided under s. 11, differs from ss. 7 and 10 by the concept

of "equal value". At p. 925, her Ladyship writes:

That provision does not prevent the employer from remunerating differently jobs

which are not "equal" in value. Wage discrimination, in the context of that

specific provision, is premised on the equal worth of the work performed by men

and women in the establishment. Accordingly, to be successful, a claim brought

under s. 11 must establish the equality of the work for which the discriminatory

wage differential is alleged. [emphasis added]

217. L'Heureux-Dubé J. ultimately found the evidence of professional segregation does not, in

itself, constitute a prima facie case under s. 11, unless such evidence "independently establishes

the equal worth of the work under consideration" which was not the case in the evidence before

her Ladyship.

218. Within the context of L'Heureux-Dubé J.'s discussion of prima facie discrimination under

s. 11 she comments at p. 926 that whether the work performed in the Section that was

predominantly female and the work performed in the Section that was predominantly male was

of equal value within the meaning of s. 11, "is intimately linked to the procedure followed by

the Commission in this case." We note some details of the Commission's procedure for

conducting job evaluations were provided. However, there was no explanation of the method of

job comparisons.

(i). Concept of "Equal Value"

219. L'Heureux-Dubé J. elaborated more fully on the concept of "equal value" in S.E.P.Q.A.,

supra, pointing to s. 11(2) of the Act which defines in general terms the manner in which the

value of work is to be assessed. That section and s. 3 of the Guidelines identify four criteria to

be applied in assessing the value of work, namely, the composite of skill, effort and

responsibility required in the performance of the work and the working conditions under which

the work is to be performed. L'Heureux-Dubé J. noted these four criteria were also the same

criteria recognized in the Equal Pay Act of 1963 in the United States. Her Ladyship concluded s.

11(2) of the Act encompassed the application of job evaluation plans to determine whether jobs

are of equal value. She refers to the application by the Commission of the Aiken Plan, a job

evaluation plan, to assess the value of the jobs.

220. We conclude from L'Heureux Dubé's dissenting judgment that there must be sufficient

evidence of "job value" to meet the requirements of s. 11 of the Act. We believe that aspect has

been established through the system of job evaluation conducted in the JUMI Study.

221. The Willis Plan was used in the JUMI Study to evaluate the sample of questionnaires

completed by the Respondent's employees. Points were assigned by evaluation committees to

each of the four factors, skill, knowledge, responsibility and working conditions identified under

s. 11(2) of the Act, for each employee questionnaire. The points assigned for each factor in each

questionnaire were then totalled to give an overall point score or rating for each employee

questionnaire. These scores are also referred to as the job evaluation results.

222. This Tribunal found as a fact in its earlier decision rendered on February 15, 1996, supra, at

paragraph 204, that the Willis Plan was an appropriate tool within the requirements of the Act

and Guidelines for the job evaluations. This Tribunal further found at paragraph 205 at p. 55:

The Willis Plan provides a tool to be used in assessing the relative value of work. But in and of

itself it does not give a methodology to determine what is the wage gap between female positions

and male positions. The determination of any wage gap is a function of comparing evaluations

between male and female jobs. The system itself does not do that without a further step.

223. We have heard from Mr. Durber that the Commission does not view "equal value" as a

purely technical issue but rather an issue of relative worth. Mr. Durber testified the "equal

value" concept has to be assessed in the context of the work being performed. According to Mr.

Durber, a job evaluation plan measures characteristics of male and female work and it should be

able to measure equivalencies between male and female work. These equivalencies are an

integral part of the evaluation process.

224. Mr. Durber further testified the Commission views pay equity as an evolving issue. As part

of the explanation for the Commission's approach to a broad interpretation of s. 11 of the Act,

Mr. Durber testified, in Volume 145 at p. 17942, lines 1-14 as follows:

Essentially, I think I have to say that our view is that pay equity is an evolving

issue. The fact that we read it broadly and liberally and because it's part of, as

you have so rightly said, the Human Rights Act and fundamental issues, means

that we try not to take a technical interpretation of that Act.

I think these conventions, as they continue to be promulgated, also show that this

business of fundamental rights is an evolving one and not, in other words,

something simply in a historical context. I think we have to see these conventions

and the Act itself as something where we learn and we interpret broader issues as

we go.

And further on he testified in Volume 145 at p. 17943, lines 11 - 22:

A. We certainly don't look on equal value as a purely technical issue. It's a matter of finding

equivalencies in value between the work of men and women and by "equivalency" we also don't

mean equal pay for equal work. We don't mean exact equivalencies.

I am sure we have heard that there was preceding legislation before the Human

Rights Act which provided for equal pay for equal work; that is, substantially the

same work. So, we believe that we have to go beyond that exactness to a broader

definition of "equality". [emphasis added]

225. Dr. Weiner outlined two basic principles behind the concept of equal pay for work of equal

value. The first principle is to evaluate all the jobs to be compared using the same set of

criteria. By this process one identifies equal pay for work of equal value. She described these

principles in Volume 16, p. 2105, line 9 to p. 2106, line 23:

THE CHAIRPERSON: What I would like to do is step back, to take a step to before you make

your decision rules and look at the concepts that we are dealing with here. We are dealing with

the concept of equal pay for work of equal value. Is there any way that you could outline what

would be the basic principles behind that concept?

THE WITNESS: The basic principles are that: One, you evaluate all the jobs to

be compared on the same set of criteria.

THE CHAIRPERSON: I want to write these down so I can understand what you

are saying. You evaluate...?

THE WITNESS: All the jobs -- all the female jobs and all the male jobs -- on the

same set of criteria. When I use the word "criteria" here, I would be referring to

the job evaluation system with its sub-factors, in the weighting of those sub-

factors.

I guess a second basic principle is that you have to pick the phrase "equal pay for

work of equal value". We have now talked about how we identify equal value

and we then have to make sure we define "equal pay" in the same way. By that I

mean that if you are going to use the maximum salary for the female jobs, you

compare that to the maximum salary for the male jobs. If you cost benefit, you

have to cost them the same way. So, you make sure that you are picking up a

salary point that is the same for both groups.

Those would be the two basic principles, I think, that come out of the phrase

"equal pay for work of equal value".

THE CHAIRPERSON: Are there any other ones?

THE WITNESS: I think anything else starts to get into decision rules where

there could be advantages and disadvantages to different methodologies.

[emphasis added]

226. The process of job evaluation used in the JUMI Study measured the relative value of skill,

effort, responsibility and working conditions of the sampled positions. The job evaluation results

of the JUMI Study have been tendered before us as reliable evidence of job value of the different

positions evaluated by the Job Evaluation Committees. A wage adjustment methodology allows

for the kinds of quantitative comparisons referred to by both Sopinka J. And L'Heureux-Dubé in

S.E.P.Q.A., supra. Through the application of the methodology relative values will be compared

to determine equivalencies. That, we believe, is how the concept of "equal value" found in s. 11

of the Act is administered in a practical sense.

(ii). Principle of "Equality"

227. All parties have advocated that s. 11 of the Act expresses a principle which must be given a

broad interpretation. There exists between the Respondent, on the one hand, and the

Commission and Alliance on the other, differing views about the technical and statistical basis

upon which the concept of "equality" is required to operate in large group complaints under s. 11

of the Act and under s. 14 of the Guidelines. The Respondent has raised the argument that the

"unit of analysis" for making comparisons in s. 11 complaints for large groups is the

"occupational group." The Respondent has led no evidence as to its understanding of the

meaning of "occupational group" in its wage adjustment methodology, relying only on parts of

Mr. Durber's evidence.

228. The Commission's approach, supported by the Alliance, is to disregard the framework of

occupational group for making comparisons once gender predominance is determined. Both Mr.

Sadler and Mr. Durber testified the Commission's task under s. 11 is to compare "work" not

"groups".

229. Mr. Durber testified as to the Commission's views of the Act. He described the Act in

general terms as fundamental legislation, quasi-constitutional in nature, which is designed to

eliminate discrimination. In particular, s. 11 is intended to eliminate discrimination in the

workplace as between males and females (based on sex). He testified the Commission believes it

has a duty to interpret the Act in a broad and liberal fashion. He indicated the Commission

frequently has technical issues coming before it and where choices must be made it tries to go

behind the technical aspects in order to understand the broad purpose of the legislation, which is,

to eliminate discrimination by creating conditions of equality.

230. The critical issue raised in this hearing requires a consideration of statutory construction

and the approach adopted by the Supreme Court of Canada in interpreting human right

legislation.

231. The focus of the Respondent is on the plain meaning of the language of s. 11 of the Act. It

argues the approach should be one of ordinary grammatical construction. The Commission and

the Alliance seek a purposive approach to an interpretation of s. 11 of the Act.

232. It is patently apparent the legislation has enshrined the principle of equal pay for work of

equal value in s.11 of the Act without however articulating a scheme on how this principle is to

be implemented. The Commission was given the responsibility for implementing an appropriate

mechanism by which to achieve that goal by the power conferred on it pursuant to s. 27(2) of the

Act. That provision reads:

(2) The Commission may, on application or on its own initiative, by order, issue

a guideline setting out the extent to which and the manner in which, in the opinion

of the Commission, any provision of this Act applies in a particular case or in a

class of cases described in the guideline.

233. This Tribunal is guided by the decisions of the Supreme Court of Canada on the proper

interpretive attitude towards the Act. The following passage from the decision of Dickson C.J.,

in Canadian National Railway, supra, is instructive of the Court's approach to interpretation of

the Act's provisions so as to achieve a fair, large and liberal interpretation. Dickson C.J. says at

p. 1134:

Human rights legislation is intended to give rise, amongst other things, to individual rights of

vital importance, rights capable of enforcement, in the final analysis, in a court of law. I

recognize that in the construction of such legislation the words of the Act must be given their

plain meaning, but it is equally important that the rights enunciated be given their full

recognition and effect. We should not search for ways and means to minimize those rights and

to enfeeble their proper impact. Although it may seem commonplace, it may be wise to remind

ourselves of the statutory guidance given by the federal Interpretation Act which asserts that

statutes are deemed to be remedial and are thus to be given such fair, large and liberal

interpretation as will best ensure that their objects are attained. See s. 11 of the Interpretation

Act, R.S.C. 1970, c. I-23, as amended. As Elmer A. Driedger, Construction of Statutes (2nd ed.

1983), at p. 87 has written:

Today there is only one principle or approach, namely, the words of an Act are

to be read in their entire context and in their grammatical and ordinary sense

harmoniously with the scheme of the Act, the object of the Act and the intention

of Parliament.

[emphasis added]

234. Dickson C.J., stressed the remedial nature of the Act and the importance of fully

recognizing the rights conferred by it and giving to them the effect intended. It is in essence the

impact of the discriminatory act upon the person affected which is decisive in considering a

complaint. The Chief Justice affirmed the Court's rejection of the necessity to prove intent in

discrimination cases. Intent is not a factor or an element of systemic discrimination whether in

the context of employment equity or pay equity.

235. Having regard to the objectives and goals of the Act we accept the purposive approach and

believe a wage adjustment methodology must be consistent with the purpose of s. 11 of the Act

which is to redress systemic discrimination for work performed by employees. The right to

equal pay for work of equal value recognized in s. 11 of the Act must be given the broad, liberal

interpretation recognized by Dickson C.J. in Canadian National Railway, supra.

236. In its oral and written submissions, the Respondent contends the phrase which is contained

in s. 11 and which reads as follows:

...differences in wages between male and female employees who are performing

work of equal value...

is severable into two parts, namely:

(i) ...differences in wages between male and female employees; and

(ii) who are performing work of equal value.

The Respondent thus argues the unit of analysis is expressed in terms of

employees in occupational groups and not in terms of work.

237. In our opinion the phrases referred to above are not severable. If one were to insert the

conjunctive "and" between "employees" and "who are performing work of equal value", the

phrase would read thus:

...differences in wages between male and female employees and who are performing work of

equal value...

Even with the insertion of the conjunctive " and" the meaning, it seems to us, remains focussed

on work, although the phrase is noteworthy because of its awkwardness. The Respondent has

given no authority for severing the phrases quoted above to impute a different meaning to them

than is intended by the legislation itself. Therefore the Tribunal rejects this interpretation of s.

11.

238. The concept of equality "...is an elusive concept...and lacks precision of definition..." to

paraphrase McIntyre J. in Andrews v. Law of Society of British Columbia, [1989] 1 S.C.R. 143

at p. 164. His Lordship in that case goes on to state as follows:

It is a comparative concept, the condition of which may only be attained or

discerned by comparison with the condition of others in the social and political

setting in which the question arises.

In Andrews, supra, the Court was addressing the application of s. 15(1) of the Charter of Rights

and Freedoms in an action by a permanent resident of Canada, who was not a citizen and was

barred from admission to the Bar of British Columbia pursuant to s. 42 of the Barristers and

Solicitors Act of that Province. In his minority decision, McIntyre J. adopted the criticism of the

"similarity situated test" enunciated by Kerans J.A. in Mahe v. The Queen in Right of Alberta

(1987), 42 D.L.R. (4th) 514 (Alta. C.A.) at p. 546 in which he states as follows:

...the test accepts an idea of equality which is almost mechanical, with no scope for considering

the reason for the distinction. In consequence, subtleties are found to justify a finding of

dissimilarity which reduces the test to a categorization game. Moreover the test is not

helpful. After all, most laws are enacted for the specific purpose of offering a benefit or

imposing a burden on some person and not on others. The test catches every conceivable

difference legal treatment.

McIntyre J. then comments as follows at p. 168:

For the reasons outlined above, the test cannot be accepted as a fixed rule or formula for the

resolution of the equality questions arising under the Charter. Consideration must be given to the

content of the law, to its purpose, and to its impact upon those to whom it applies, and also upon

those whom it excludes from its application. The issues which will arise from case to case are

such that it would be wrong to attempt to confine these considerations within a fixed and limited

formula. [emphasis added]

239. The historical, political and social basis for the introduction of s. 11 of the Act was to

counter systemic discrimination which had deprived female workers of wages of which they are

equally deserving compared to male workers. In our view, s. 11 of the Act focuses on and is

directed towards the notion of equality as between male work compared to female work as it has

been traditionally understood in a historical and societal perspective. Section 11 of the Act is an

attempt to remedy what had become an issue of social justice. The fundamental purpose and

goal of the Act is to give effect to the principle of equality which, in our opinion, favours equal

treatment for female work relative to male work. Therefore, we believe that s. 11 is aimed at

redressing a wage gap between male and female employees who perform "work" of equal

value. Where the work is deemed to be of equal value it should be paid the same.

240. The principle of equality that arises in s. 11 is a comparative concept. In our view it is

difficult to achieve exactitudes with a job evaluation procedure which is, by its nature,

subjective. Job evaluation, which identifies job value and the mechanism of wage adjustment

methodologies for testing and measuring the relative value between job factors, which in

themselves are quite distinct, are not simple quantitative processes.

241. Section 11 of the Act has expanded the concept of discrimination into the emerging field of

pay equity. The federal government approached the problem of systemic discrimination in 1978

by using the complaint-based process provided for in the Act as opposed to the proactive

approach undertaken by some of the provincial legislatures. As we have seen, five of the

provinces Manitoba, Ontario, New Brunswick, Nova Scotia, and Prince Edward Island have

passed pay equity legislation requiring public sector employers to implement pay equity plans

within their establishments. Québec deals with pay equity in its Human Rights Charter on a case

by case basis. British Columbia and Newfoundland have agreements to introduce pay equity and

Alberta and Saskatchewan have no legislative program at the present time.

242. According to Dr. Weiner, the phrase "pay equity" is often referred to in federal writings, as

"equal pay". Both of these phrases are shorthand for "equal pay for work of equal value." The

principle of "equal pay for work of equal value" has been treated as synonymous with the more

familiar phrase "pay equity". Recently Hugessen J.A. in Non-Public Funds, supra, wrote at p.

81: "This case concerns pay equity."

243. The Federal Government instructed the Commission to devise a process for determining

how the principle of equality that arises in s. 11 of the Act might be implemented in a practical

sense. The Government empowered the Commission through s. 27(2) of the Act to establish

guidelines that would implement the principle of equal pay for work of equal value.

244. Over a period of time guidelines were enacted and amended as complaints were filed under

the Act. The current Guidelines were proclaimed in November 1986 following active input from

the various stakeholders, interest groups, federal employers and federal unions. It is to be noted

that no precise wage adjustment methodology for group complaints was spelled out in the

Guidelines. We believe it reasonable to infer from this the Commission favours a flexible

approach as opposed to a regulated one. Flexibility allows the Commission to adopt a

methodology best suited to the circumstances of the case.

245. Absent a "fixed rule or formula for the resolution of equality questions..." per McIntyre J.

in Andrews, supra, the Commission is free to call upon and rely on the expertise of persons such

as Dr. Weiner, Mr. Sunter, Dr. Shillington and Mr. Durber when applying compensatory

techniques commonly used in pay equity exercises in order to determine if a wage gap exists.

246. We are of the opinion, therefore, if a difference in wages is established through a

methodology that adheres to the principles of "equal value" and "equality" it will be sufficient

proof of a prima facie discrimination based on gender under s. 11 without recourse being had to

the notion of causation. We have already referred to the concept of "causality" in this

discussion. It is, in our opinion, incompatible within the prevailing historical and social context

which must be recognized when addressing the injustices generated under systemic

discrimination based on gender.

247. The separate methodologies of comparison favoured by each party will be addressed with

these elements in mind. Before doing so, we believe it is essential for determining the existence

of a wage gap that the relevant Guidelines be understood.

VII. THE EQUAL WAGES GUIDELINES

A. History of the Equal Wages Guidelines

248. Parliament empowered the Commission pursuant to s. 27(2) of the Act to issue guidelines

"setting out the extent to which and the manner in which, in the opinion of the Commission, any

provision of the Act applies in a particular case..." The current Guidelines dated November 18,

1986 and gazetted in December 1986 incorporate provisions relating to group complaints found

under the heading "Complaints by Groups". They read:

Complaints by Groups

12 Where a complaint alleging different wages is filed by or on behalf of an

identifiable occupational group, the group must be predominantly of one sex and

the group to which the comparison is made must be predominantly of the other

sex.

13 For the purpose of section 12, an occupational group is composed

predominantly of one sex where the number of members of that sex constituted,

for the year immediately preceding the day on which the complaint is filed, at

least

(a) 70 per cent of the occupational group, if the group has less than 100

members;

(b) 60 per cent of the occupational group, if the group has from 100 to 500

members; and

(c) 55 per cent of the occupational group, if the group has more than 500

members.

14 Where a comparison is made between the occupational group that filed a

complaint alleging a difference in wages and other occupational groups, those

other groups are deemed to be one group.

15 (1) Where a complaint alleging a difference in wages between an

occupational group and any other occupational group is filed and a direct

comparison of the value of the work performed and the wages received by

employees of the occupational groups cannot be made, for the purposes of section

11 of the Act, the work performed and the wages received by the employees of

each occupational group may be compared indirectly.

(2) For the purposes of comparing wages received by employees of the

occupational groups referred to in subsection (1), the wage curve of the other

occupational group referred to in that subsection shall be used to establish the

difference in wages, if any, between the employees of the occupational group on

behalf of which the complaint is made and the other occupational group.

249. The first version of the Guidelines, published in 1978, did not address the application of s.

11 to group complaints. Sections 13 through 15 were first introduced into the revised Guidelines

issued and gazetted in November of 1986. Prior to the last revision the Commission had

identified employer and employee associations seeking to comply with s. 11 of the Act. Mr.

Durber testified in order to assist the Commission in investigating complaints under s. 11 there

was a need to know whether groups of workers who had filed complaints were predominantly

male or female. The Commission had the task of drafting a guideline for that purpose.

250. During the process of revising the Guidelines the Commission published "Background

Notes of Proposed Guidelines - Equal Pay for Work of Equal Value" in March 1985, (Exhibit

HR-18, Tab 5). This document was sent to 70 employers, unions and interest groups with a

request for comments. Appended to the Background Notes were draft guidelines on sex

predominance, group complaints, the same establishment and value.

251. The Tribunal did not receive the full text of the Commission's Background Notes but they

did include a reference to group complaints in the document tendered into evidence. It is noted

the Commission, in its proposed draft guidelines, introduced the concept of averaging the male

group wage comparisons. This concept is prevalent in each of the methodologies advanced by

the parties. The Background Notes provide an illustration of a method proposed for adjusting

wages using a regression line at p. 7:

II - GROUP COMPLAINTS

A number of problems are encountered in dealing with groups of employees and

with individuals as members of groups.

Subsection 1 of the proposed guideline states the requirement for sex

predominance and emphasizes that the sexual composition of the group to which

an individual belongs must be considered in determining whether sexual

discrimination exists.

Subsections 2 and 3 set out the concept of indirect comparison of employees

who are members of groups.

Indirect comparison is already Commission practice, and it represents a move in

the direction of comparable worth/pay equity as the terms are understood in the

United States.

Central to this guideline is the concept of the adjustment of female* group wages

to an average level of male group wages. This method ensures that women as

individuals will earn at least the average wage paid to men doing work of equal

value in the same establishment.

APPENDIX B: DRAFT PROPOSED GUIDELINES

GROUP COMPLAINTS

1) Where a complaint involves an occupational group or groups the complaining

group must be predominantly of one sex; and the group(s) to which a comparison

is to be made predominantly of the other sex. Where a complaint involves an

individual or a group of individuals who are members of a larger occupational

group, the sex characteristics of the larger group will be considered in deciding

whether the situation complained about is discriminatory on the basis of sex.

2) Where a complaint involves more than one individual or group of employees,

the individuals or groups to whom a comparison is being made will be considered

to be one group. For purposes of wage determination, the weighted average wage

paid to each cluster of employees in the comparison group who are performing

work of equal value will be employed.

3) Where the direct comparison of employees in one group with individuals or

employees of another group is not possible, the Commission will consider

proposals for the adjustment of salaries that are based on indirect comparisons,

including standard statistical methods such as regression analysis to establish a

salary trend line. In these cases, where an adjustment is indicated, the wages of

employees in the complaint group shall be adjusted to the level predicted by the

salary trend line of the comparison group for the value of the work performed by

those employees.

[emphasis added]

252. Ms. Millar testified on behalf of the Alliance that the 1986 proposed guidelines on group

complaints and the accompanying background notes (Exhibit HR-18, Tab 5), particularly p. 7

of the background notes which illustrates wage lines representing average male salaries, describe

essentially what was agreed between the Alliance and the Treasury Board to settle the GS

complaint. (see Section I, B, Paragraph 8).

253. According to Mr. Durber the Commission received submissions from 39 organizations,

including the Respondent. The Commission met with many of the 39 organizations. Mr. Durber

explained there were two areas of common concern from those organizations. The first area was

a requirement for sex predominance, now covered in s. 13 of the Guidelines. The second area of

concern was regional rates (see Section IX). Other concerns raised were the issue of ratcheting

(see Section VIII, B), how to assess value, indirect comparisons (see Section VII, E(i)), and

utilization of regression analysis (see Section VII, E(i)). Mr. Durber testified that after myriad

correspondence and a number of meetings the Commission made "fairly noteworthy changes" to

its proposal. (Volume 186, p. 18101).

254. In a memorandum dated February 4, 1985 from Hanne Jensen, who is identified as the

Director, Complaints and Compliance Branch of the Commission and addressed to Members of

the Commission, (Exhibit PIPSC-3), both the Commission's preferred approach to adjusting

wages in group complaints and the Respondent's approach was described. The relevant portions

of the memorandum read as follows at p. 3:

The approach detailed in the proposed guideline was used in calculating the

retroactive portion of the General Services groups complaint. The method of

adjustment ensures that women as individuals will earn the average wage paid to

men. There will still be some women earning more and there will be some men

earning less than women. This is unavoidable, in any approach using an average

wage as a basis for adjustment. It should be noted that this approach may be

criticized by women's groups as a "dilution" of S. 11.

The proposed approach will also raise objections from Treasury Board, who are

still proposing a form of averaging that the Commission rejected in 1981 as a

method of settling the GS complaint. Briefly, it proposed averaging male wages,

averaging female wages , calculating the percentage difference and increasing

female wages by that percentage. (method III) This approach would have

ensured that women as a group received the average male wage. One of the

problems with this approach is that some women will continue to receive less than

the average male wage, and some more. Indeed, in the GS complaint the wages

for a number of women's jobs would have exceeded the highest paid male

jobs. The Treasury Board staff argument in favour of this approach is that it

preserves the relativity existing among female groups. Where one female group

is higher paid than another, it is argued that this difference is non-discriminatory

and cannot be addressed by Section 11. The counter argument is that these

differences are evidence that some female groups have been discriminated against

more than others. The Treasury Board staff approach is sound only if their line of

reasoning can be accepted. The draft guideline is based on the premise that

differences in wages among women are due to varying degrees of

discrimination. If that view is accepted, the method described [p]roposed

guideline best addresses that situation.

255. There is some evidence before the Tribunal in the testimony of Ms. Lise Ouimet about the

Respondent's reaction to the draft Guidelines circulated by the Commission in 1986. During the

Voir Dire, supra, concerning the admissibility of the data generated during the JUMI Study, Ms.

Ouimet testified about the Respondent's position on the use of the information gathered in the

JUMI Study. Ms. Ouimet was the Respondent's co-chair of the JUMI Committee. As part of her

testimony, Ms. Ouimet admitted she had presented a paper at a seminar at York University,

Toronto, in March 1987. The university was hosting a conference on Equal Pay for Work of

Equal Value. Her paper (Exhibit PSAC-20) reveals, to some extent, the Respondent's positive

reaction to the revised Guidelines and reads in part:

The Canadian Human Rights Commission has recently approved Guidelines which address a

number of the issues related to the implementation of equal pay for work of equal value. Two of

the interpretations we e[s]pecially welcome are related to regional rates as a reasonable factor for

wage differences, and the acceptance of using the weighted average of wages paid comparison

groups in calculating equal pay adjustments for a complainant group as opposed to settling to the

highest comparison wage. This will eliminate the risk of creating grounds for reverse

discrimination. [emphasis added]

256. The final version of the Guidelines pertaining to group complaints, issued and gazetted in

November 1986, included the new section "Complaints by Groups" sections 12 through 15. We

note the four sections under the heading "Complaints by Groups" in the final version of the

Guidelines does not specify a methodology for wage adjustment in determining how the work of

male and female employees can be compared so as to be consistent with s. 11 of the Act.

257. We further note this final version introduced the term "occupational group" into the section

without defining its meaning. No explanation was provided concerning the introduction of this

phrase.

258. According to Mr. Sadler a feature of the 1986 Guidelines and specifically ss. 11 through to

15, notwithstanding the term "occupational group", enables the Commission to apply the

Guidelines in both a structured group situation and in an unstructured group situation. Mr.

Sadler testified the formal grouping of jobs into occupational groups in the late 1960s within the

Public Service was established for the purposes of collective bargaining. That is part of the

reason why, in a s. 11 complaint, the Commission does not consider itself bound, under the Act,

to the Respondent's classification system.

B. Sections 12 and 13 of the Equal Wages Guidelines

(i). Occupational Groups

259. Mr. Durber testified that although s. 11 of the Act does not refer to groups, it implies

groups when it refers to male and female "employees". He said groups are referenced in the

Guidelines for s. 11 complaints, by the use of the language found in the heading "Complaints by

Groups", introducing ss. 12 through to 15.

260. Mr. Durber testified the Commission needs to understand what is male and female work at

the onset of a group complaint. For this reason the Commission "pays attention" to the issue of

percentage of males and females in occupations as required by s. 13 of the Guidelines. As

explained by Mr. Durber the Commission does not automatically accept the occupational group

designation of the employer's organization when determining sex predominance. He testified the

Commission applies a broad interpretation to the term "occupational group". He stated the

Commission may, for example, consider other official usage of the terminology "occupational

group" such as that used by the National Occupational Classification publication. According to

Mr. Durber the National Occupational Classification is a publication which embodies and refines

concepts which Statistics Canada and Employment & Immigration have used for some years. Its

purpose is to determine boundaries between occupational groups. One reason the Commission

does not automatically use the occupational groups designations of an organization is explained

by Mr. Durber in Volume 145 at p. 17977, line 20 to p. 17978, line 12 as follows:

Q. When you say that, does that mean that you don't automatically take the

occupational group designation of the organization which is named in the

complaint?

A. No, we don't. There are a number of reasons for that. One is that the

workforce is already highly occupationally segregated and that segregation may

give rise to difficulties for pay equity.

What we are really trying to do is enable ourselves to look at what is men's work

and what is women's work and if the structure the employer uses doesn't help us

to get at men and women's work, then I think we have to look at it very closely to

determine whether it's helpful in a complaints investigation or impedes. We have

criteria that we have to bring to bear in testing the occupational grouping.

261. The Commission uses other criteria to designate a group including an examination of the

skill level of particular jobs. This means assessing the amount and type of education, training,

entry-level experience, complexity and responsibilities of the work. The Commission may look

for occupational specialization or occupational mobility which examines whether jobs are inter-

related so as to form "job families". Still other criteria considered by the Commission are salary

structure, similar work and linkages in salary structures and career structures.

262. The Commission's approach to defining a "group" was further clarified by Mr. Durber

under cross-examination by the Respondent in Volume 162 at p. 20207, line 2 to p. 20209, line

23 as follows:

Q. Now, would it be the position of the Commission that an equal pay complaint

can be made on behalf of any group of individuals but would constitute an

occupational group the way you've just defined it?

A. Well, I think we ought to be somewhat clear here. The Commission isn't in

the business of inviting complaints. We try to play a neutral role. So when you

say a group can make complaints, I want to be quite clear that the Commission

isn't in the business of permitting complaints. What constitutes a reasonable

group for a complaint I think would first of all be up to the individuals

concerned. They might seek advice from the Commission.

We had an inquiry recently, for example, from nurses who were within a broader

bargaining unit in a Crown corporation, and they asked us whether they could --

whether they could lodge a complaint. Well, clearly, everyone has the right to

lodge a complaint.

Now, whether they were a group or not would remain to be seen probably during

investigation. But -- by the way, we've yet to see a complaint. Now, on the face

of it, of course, it's quite clear that nursing is a profession, it's accepted as a

profession, so that some of these answers as to whether a group is a group are

somewhat self-evident. But people, for example, in a specific job such as, let us

say, pay clerks, might consider themselves to be a group, even though they're part

of a broader, let us say, clerical group as in the public service. And then one

would have to look at the nature of issues they were bringing forward in order to

understand how discrimination, if at all, worked in respect of the people in that

specific job.

Q. That's a good example, Mr. Durber. And if you were satisfied after

investigation that the pay clerks meet the definition of occupational group or meet

your -- come within the meaning of that expression, I guess is a better way of

saying it.

A. Yes.

Q. I guess that's what I'm trying to ascertain, is whether the Commission would

then be prepared to treat them as an entity and deal with that complaint as a group

complaint in looking for comparison with other groups?

A. Well, I think the Commission has an obligation to investigate, in any event,

under the Statute, unless under section 41 there are some impediments. But, yes,

we would examine whether the characteristics of the work made it sensible to

treat all of these individuals together in an occupational sense. I think we would

have to recognize that those pay clerks did not have their own salary structure, if

we use the public service as an example, so we would then have to examine the

nature of the difficulties, the discrimination that was alleged.

It might, for example, relate to how their work was valued. And indeed, we've

seen instances of that. We had a complaint of registered nursing assistants, for

example, compared with orderlies. Now, each of those groupings, if you like, was

a job, and because it was a job with linkages in terms of the work, you could say it

was occupational. So that complaint proceeded, and indeed, it reached a

satisfactory conclusion of a settlement.

Q. And that's although the employer had established that as part of a larger

group?

A. Both of those jobs were part of a female predominant group, hospital

services, but the Commission was satisfied that the registered nursing assistants

were predominantly female and the orderlies were predominantly male. And the

issue then was not the discrimination in the broader salary structure, which of

course covered both, but in the value of the work, that is, should RNAs be the

same as or greater than, whatever, in value than the orderlies.

So again, that depended on the nature of the discrimination alleged, what one

focused on. But clearly, that was an occupational group in the broad sense of the

word. Both of them, I should say, were occupational.

263. Mr. Durber testified from the Commission's perspective the concept of "occupational

group" is only necessary during the Commission's investigation of a complaint. It is applied by

the Commission when determining the sex predominance of the complainant and comparator

groups identified in the complaint. Once this is accomplished the notion of occupational group is

set aside by the Commission investigator.

264. Mr. Durber testified the "occupational group" determination made for purposes of sex

predominance in the investigation of the complaint no longer applies in formulating a wage

adjustment methodology because the focus for comparison is then on the "work" rather than on

comparing groups. He testified to that effect in Volume 162 at p. 20210, line 10 to p. 20211, line

13:

Q. Thank you. Now, I would like to consider the term, the expression, rather,

occupational group, being applied to the comparator. So we've discussed it in the

context of the complainants, and now I'd like to consider it as we look for

comparators for the complainants.

As I understand it, the Commission is proposing in this case that the comparator

be all of the male questionnaires that fall within the range of the minimum and

maximum scores for a level of a female occupational group?

A. I believe that that's what the methodology for wage comparison comes down

to, yes. That is, we're looking at work of equal value, which is a somewhat

different question, I think, from the constitution of occupational groups.

Q. Well, all right. Now, are you saying that the comparator does not need to be

an occupational group?

A. No, no. What I am saying is that one needs to understand what is male and

female work at the outset of a complaint when one looks at groups. That is, one

must pay attention to the issue of percentage of males and females in

occupations. Once that is done, one then goes on to compare work as between the

complainant occupational group and males doing work of equal value to that

female occupational group.

Q. All right. And in deciding which males you will use as comparators you do

not apply as a criteria that they must have a similarity of work?

A. No, I believe that the issue then, occupational grouping, has been set aside,

because what we're then doing is we're then comparing work of equal value, not

groups of equal value.

265. The difficulty of attributing a precise meaning to the term "occupational group" was

acknowledged by Dr. Weiner in her evidence. She considered the definition found in a Canadian

personnel textbook fitting, it described an "occupational group" as "a grouping of jobs with

broadly similar content". (Volume 8, p. 1115). That definition accords with Dr. Weiner's

understanding of the term. She testified that within the occupation jobs have a commonality of

function such as that which exists with nurses and with accountants. Dr. Weiner's approach is to

distinguish a "position" from a "job". She described a "position" as an element of work, i.e., a

combination of tasks and duties. She defined a "job" as a grouping of positions and an

"occupation" as a grouping of jobs.

266. As we noted earlier (see Section I, C, Paragraph 27) because of the JUMI Study it was not

necessary for the Commission to determine the sex predominance of either the complainant or

the comparator group that was used in the JUMI Study. The JUMI Committee designated which

occupational groups were female-dominated and which groups were male-dominated using the

criteria found in s. 13 of the Guidelines. All parties, the Commission, the Alliance and the

Respondent have informed this Tribunal the sex predominance of the Complainant groups and

the comparator groups is not in issue. It seems the concept of "occupational group" as defined

by the Employer's classification system was applied by the JUMI Committee in its designation of

female predominant and male predominant groups.

267. We find in a s. 11 complaint the discrimination claimed is made on the basis of

gender. Therefore to be a valid complaint the complainant group and the comparator group must

meet the qualifications set down in s. 13 of the Guidelines. Section 13 of the Guidelines lays out

the requirement that the complainant group must be predominantly of one sex and the group that

is identified in the complaint as the comparator group must be predominantly of the opposite

sex. This is the essential element which must be achieved in order for the complaint to be

viewed as valid and worthy of investigation by the Commission.

268. Since s. 11 of the Act is aimed at comparing the value of work, it does not make sense to

restrict comparison in the Guidelines by the occupational classifications of the employer which

have, in part, contributed to the problem. The employer's classification system was developed

primarily for collective bargaining purposes which formalized into categories the occupational

segregation of work. The largest occupational group in the Federal Public Service is the CR

occupational group, comprising 50,000 employees, a female-dominated group. Compared to the

small number of female-dominated occupational groups, there are many more male-dominated

occupational groups in the Federal Public Service. In the JUMI Study there were nine female-

dominated occupational groups that participated compared to 53 male-dominated occupational

groups. We believe a whole occupational group approach only perpetuates the problem of

undervaluation of female work. Dr. Weiner pointed out in Volume 6 that occupational

segregation has given rise to the need for pay equity. She says at p. 895, lines 1-5:

Occupational segregation is one of the conditions that lead to the need for pay

equity. The second condition requiring pay equity is undervaluation and

underpayment of the work done by women.

269. In rejecting the whole occupational group approach as inappropriate in large group

complaints, Dr. Weiner described it as "cherry picks". We refer to her explanation found in

Volume 16 at p. 2208, line 8 to p. 2209, line 1:

A. The disadvantage would be that if someone -- as a colleague of mine likes to say -- "cherry

picks", they could say, look there is, let's say, four or five occupational groups that could be the

one we could make a direct comparison to. One side might have that it's to their advantage to

pick the male occupational group that would get them the highest adjustment; another side might

say, well let's pick the occupational group that would get them the lowest adjustment.

So if you go to pick one when you could use four or five then you are into this --

in addition to pay equity we either want to get the most money or we want to pay

out the least. If you say let's combine the information from those four or five,

then you get something closer to the middle and you get rid of this extreme or

having another agenda item.

270. We must now examine the meaning of the term "occupational group" found in ss. 14 and

15 of the Guidelines.

C. Section 14 of the Equal Wages Guidelines

271. Section 14 of the Guidelines reads as follows:

14 Where a comparison is made between the occupational group that filed a

complaint alleging a difference in wages and other occupational groups, those

other groups are deemed to be one group.

272. The Respondent argues that ss. 12 to 15 of the Guidelines mandates that occupational

groups must form the basis of comparison in group complaints. The Respondent contends s. 14

of the Guidelines is invalid because it provides for the combining of male comparator

occupational groups into a deemed group contrary to the concept of "causation" and "equal

value" required by s. 11 of the Act.

273. The Respondent challenges the validity of s. 14 of the Guidelines alleging that the

combined effect of that section taken with s. 11 of the Act renders s. 14 of the Guidelines

invalid. The Respondent claims s. 14 is inconsistent with the concept of "equality" contained in

s. 11 of the Act. According to the Respondent a combination of male comparator occupational

groups, which on an individual basis are not equal to the female complainant occupational

groups, are blended by virtue of s. 14 of the Guidelines to form a deemed group. The result, the

Respondent contends, offends the requirements of equality under s. 11 of the Act.

274. Critical to the Respondent's challenge to the validity of s. 14 of the Guidelines is the

Respondent's interpretation of "employees" as contained in s. 11 of the Act. It contends that the

word "employees" as it appears in s. 11 means "occupational groups" when dealing with group

complaints and that this meaning is defined by the employer's classification system.

275. Dr. Weiner's understanding of s. 14 of the Guidelines and the meaning of an occupational

group was provided in her examination-in-chief in Volume 7 at p. 1048, line 13 to p. 1049, line

19 as follows:

Q. Could I ask you to read Section 14 for yourself, please?

A. It seems there is a single occupational group, presumably female, which has

made a complaint. It should be compared against all male jobs of a similar value

range. I would say that is going to the segmented line and looking at male jobs

falling within the same value range as this female occupation.

Q. The section refers to groups. Is it your understanding that the guidelines do

not define what a group is?

A. That is my understanding.

Q. In compensation, what is a group?

A. The word "group" doesn't have a lot of meaning. If you are talking about an

occupational group, that would be jobs similar in function but different in

level. Within the probation officer group, there would be entry level probation

officers, senior probation officers and probation officers who supervise jobs. That

could be considered an occupation.

Q. In your opinion as a practitioner, is the use of composite line or segmented

line permitted under Section 14?

A. I would see a segmented line as consistent with Section 14.

276. Later, Dr. Weiner indicated s. 14 of the Guidelines also contemplated the composite line

approach. She stated in Volume 9 at p. 1223, line 22 to p. 1224, line 14 as follows:

Now, you referred to section 14 of the Guidelines and indicated that, in your view, section 14 of

the Guidelines supported a segmented line approach.

A. You could use a segmented line approach with section 14, that is right.

Q. Would you --

A. But, you do not have to.

Q. You do not have to. And that was my next question.

Would you agree with me that section 14 also supports a composite line

approach?

A. Yes, it is one of those decisions that the parties could decide.

Q. And section 15, equally, supports a composite line approach?

A. Right.

277. Dr. Weiner provided a further insight into her understanding of s. 14 of the Guidelines

when questioned by Respondent counsel. She believed the effect of s. 14 was to allow for group

boundaries to change and testified in Volume 10 at p. 1441, line 7 to p. 1443, line 4 as follows:

So, what I would like you to do now is remember the content of 14 and 15 and read them -- or

read 12 and 13.

A. Okay.

Q. Now, that you have read these provisions let me ask, do you agree, as a pay

equity expert, that even the word "group" because it is not defined, and in your

own statement -- and I quote: "Doesn't have a lot of meaning." I refer you to

transcript, Volume VII, Page 1049.

"Q. In compensation, what is a group."

And you said:

"The word "group" doesn't have a lot of meaning."

Although, after that, to be fair, you have qualified it and we will get into that

later.

But for the time being you will agree that even if it doesn't have a lot of

meaning, once we have chosen what it is, in terms of units, group there, it is

necessary and very important to stick with that unit throughout the process of pay

equity analysis and not change the rules of the game in mid-stream, to use your

own word?

A. The word "group" here seems to refer to an occupational group.

Q. Yes.

A. All right, and then in 14, it seems to say that the complaining occupational

group must -- that you can never change that. But if there is other occupational

groups, so more than one male occupational group against whom the complaint is

made, I read Section 14 to say that these occupational groups are deemed to be

one group. So, that the boundaries of that group change.

Q. We will get into that, but my question is, in your view, given we don't have a

specific notion of the term "group" here, because it is not defined, in your view, is

it important to stick with it once we have chosen that notion and not change the

definition of "group" as we go along?

A. Consistency always sounds like a good idea in general. Sometimes when you

get to specific pieces you find out why it's called the "hobgoblin of small

[l]ines." [emphasis added]

278. And further on in the same volume, she elaborated on s. 14 enabling comparisons between

ranges of values. She testified as follows at p. 1448, line 21 to p. 1449, line 10:

Q. My question is, isn't it the fact that out of these comparators to be used under Section 14, to

be amalgamated to form a new deemed to be comparator?

A. Right.

Q. Is it your understanding that each one of those units must be an equal unit to

the complaining group?

A. Okay, there are certainly no words to that effect in 14, but I would read --

Q. I am asking you if it is your understanding?

A. It is my understanding that you would look to jobs in a similar range of value,

but not equal.

279. Mr. Sunter provided testimony about the Commission's understanding of s. 14 of the

Guidelines in Volume 107, p. 12856, lines 8-21 as follows:

Q. Could we turn to tab 2, which is the equal wages guidelines. You have indicated that you

have seen these before.

Sections 14 and 15, to narrow this down -- what information in those sections

was relevant for your analysis?

A. Section 14 of the guidelines clearly implies, I think, that I need not -- and I

am coming back to my own analysis here -- that I need not differentiate between

male occupational groups, when I was trying to find a comparator for a particular

female group, that I would consider all groups on the male side as one group for

the purposes of this comparison.

280. Mr. Durber gave the Commission's approach to s. 14 of the Guidelines as providing a point

of reference to making comparisons in Volume 146 at p. 18081, line 10 to p. 18084, line 1 as

follows:

Q. Could you move to section 14 of the Equal Wages Guidelines.

A. Yes. This is a provision which is, I think, sometimes taken rather too

literally. It reads:

"Where a comparison is made between the occupational group that filed a

complaint alleging a difference in wages and other occupational groups, those

other groups are deemed to be one group."

I think the operative word there is "deemed". When I say literally, I really don't

quite mean literally because if people read the word "deemed", they would know

that we weren't in reality making those other groups one. But sometimes people

think that despite the word "deemed", that we are in fact creating one

occupational group. We are simply treating the same for purposes of creating a

point of reference.

This is rather analogous to subsection 11(2) where you will recall we create a

weighted average for purposes of having a reference point. So when we have

several groups, two or more, we put them together for purposes of treating all of

the observations together.

Q. This is the other groups, meaning?

A. The comparator groups. I would say that is the purpose of 14. It is in the

same spirit as 11(2), in my view.

Q. Again, you talked previously that the Commission interprets provisions of the

Act and the Guidelines liberally. You have just given one statement with regards

to the phrase "deemed".

Is there anything else in this section of the Guidelines which you could assist us

on with regards to the treatment of section 14 by the Commission?

A. I think that we will no doubt find at issue here the question of whole groups

and whether we must consider groups as a whole. I think we will find in a

number of instances throughout these Guidelines that we are talking about

comparisons of work. I am sure we will come back to that. So that what I do

here is give the term "group" a fairly broad definition; that is, I don't take it

simply to mean an occupationally segregated group established by an employer.

Q. How does that relate to section 11(1) of the statute which talks about work of

equal value, individuals or employees performing work of equal value?

A. I think that our duty is to examine the value of work, not to conserve existing

occupational structures as we are making comparisons.

In other words, I would say that if we find ourselves restricted by an

occupational structure of an employer restricted from carrying out the broad

comparisons of work required under section 11(1), then my view of it is that we

choose to look at work rather than groups, if we find ourselves restricted.

281. Similar to the approach of the Commission, Dr. Weiner focussed on the comparison of

work rather than the comparison of groups. For this reason she was not concerned if the male

comparator had high values and low values ("heads" and "tails"), from different male-dominated

occupational groups. (Exhibit R-18). Neither did it concern her that the resulting male

comparator using the female range of value was not an occupational group. She testified in

Volume 10 at p. 1459, line 2 to p. 1460, line 8:

Q. They [heads and tails] come from different occupational groups.

How can we then call them a group?

A. What is similar about them, what is relevant to the pay equity exercise, is that

they are work done by men and that they fall within a value range of work done

by women, and therefore they provide a way for us to develop a standard to do

equal pay -- to assess if equal pay for work of equal value is operating.

Q. Now, my question is: Are we changing the rules midstream when we define

the complainant occupational group in one way, but then construct a comparator

group from bits and pieces, heads and tails, of several occupational groups?

A. I would say no.

Q. We're not.

A. We are, in fact, taking bits and pieces versus a whole --

Q. Yes.

A. -- so in that sense -- but the rules are constructed for a purpose and the

purpose is to enable us to assess equal pay for work of equal value. And

unfortunately organizations haven't designed themselves in a way always to make

that a simple exercise again.

So we have to look at those principles and then look at what we need to do and

make sure what we are doing is consistent with those principles, even if

sometimes we seem to be inconsistent. [emphasis added]

282. Dr. Weiner and Mr. Sunter viewed s. 14 as effectively making possible comparisons

between female work and male work without the constraint of a group structure. Dr. Weiner's

response to Respondent counsel of the effect of s. 14 of the Guidelines is found in Volume 10 at

p. 1449, lines 8 - 10:

It is my understanding that you would look to jobs in a similar range of value, but not equal.

D. Validity of Section 14 of the Equal Wages Guidelines

283. Respondent counsel claims that s. 14 of the Guidelines is invalid on the grounds that it is

inconsistent with s. 11 of the Act because it allows for the selection of a male comparator beyond

the equality concept intended by s. 11 of the Act. Respondent counsel argues that s. 11 of the

Act directs the selection of a male comparator consisting of the lowest paid male-dominated

occupational group in order to eliminate discrimination based on gender. According to

Respondent counsel, the expression "equal value" found in s. 11 of the Act limits a comparison

to the lowest paid male-dominated occupational group.

284. According to the Respondent, s. 14 of the Guidelines imposes a requirement to combine

male-dominated occupational groups performing work of equal value. Such a requirement, it is

argued, purports to extend the purpose of the Act beyond that which is intended, namely, the

elimination of discrimination based on gender. The Respondent alleges the Commission has

exceeded its power under s. 27 of the Act to prescribe such a requirement which, it argues, goes

beyond the purpose and terms of the legislation and is inconsistent with the Act.

285. Respondent counsel contends that s. 14 of the Guidelines cannot operate without s. 15 of

the Guidelines because s. 14 only solves the problem of what to do when there is more than one

male comparator group. If we were to find s. 14 is valid and that the male comparator will be the

deemed group, Respondent counsel contends s. 15 when read with s. 14, provides the rules to

measure the wage gap. However, in the Respondent's view, if we find s. 14 invalid and select

the lowest paid male-dominated occupational group as the comparator s. 15 can operate

independently of s. 14. (Volume 243, p. 32463). Respondent counsel also contends by applying

s. 15 of the Guidelines the wage curve of the lowest paid male-dominated occupational group

can be calculated to provide the average wage and value of work of the group.

286. The Respondent argues s. 14 of the Guidelines extends the meaning of equality by allowing

for comparability on the basis of the Commission's argument of "on average fairness" (see

Section III, B) thus extending the principle of equality beyond gender-based discrimination in

wages. Evoking the principle of "on average fairness", according to Respondent counsel, is

inconsistent with the principle of "equal value" because it calls for something additional

involving policy considerations other than discrimination. Respondent counsel makes this point

in Volume 243 at p. 32398, line 18 to p. 32399, line 14:

THE CHAIRPERSON: How does it go beyond the purpose of the Act?

MR. FRIESEN: That the purpose of the Act is to eliminate discriminatory

practices and that, the Respondent says, is achieved by adjusting to the lowest

paid male comparator. But section 14 of the Guidelines calls for us to adjust not

only to the lowest paid male comparator, but up to the average of all comparators

performing work of equal value.

So it's a further adjustment, and that involves policy considerations other than

discrimination. It strives for on average fairness, in effect.

And in the submission of the Respondent there is a difference between on-

average fairness, which may be a very valid policy determination. Parliament

may at some point determine women should have wages that achieve on-average

fairness. But this pay equity legislation that is before the Tribunal is in Human

Rights legislation, which is directed expressly at discrimination and only at

discrimination.

So if the purpose of the Act is to eliminate discriminatory practices and the

Guideline calls for something additional involving other kinds of policy

considerations, then it goes beyond the purpose of the Act.

287. The Commission's argument concerning "on-average fairness" has been described in

Section III-B of this decision. In brief the Commission advocates an interpretation of s. 11 of

the Act that would provide fairness of opportunity in employment and fairness of result in wages

to employees in the female-dominated complainant groups. Essentially the Commission

contends employees in these complaints who perform work of equal value to individuals who are

performing male-dominated work should have an equal opportunity to receive, on average, the

same wages paid to employees in male-dominated groups. The Commission contends support

for that interpretation arises from s. 2 of the Act setting out the purpose of the

legislation. Section 2 of the Act reads in part:

2. The purpose of this Act is to extend the laws in Canada to give effect, within the purview of

matters coming within the legislative authority of Parliament, to the principle that every

individual should have an equal opportunity with other individuals to make for himself or herself

the life that he or she is able and wishes to have, consistent with his or her duties and obligations

as a member of society...

288. According to Respondent counsel achieving "on average fairness" is also incompatible with

the concept of causation because it not only eliminates a discriminatory practice based on gender

but goes further and calls for adjusting wages based on reasons other than gender.

289. Section 27(2) of the Act empowers the Commission to issue guidelines setting out the

extent to which and the manner in which, "in the opinion of the Commission", any provision of

the Act applies in a particular case. Section 27(2) reads:

(2) The Commission may, on application or on its own initiative, by order, issue

a guideline setting out the extent to which and the manner in which, in the opinion

of the Commission, any provision of this Act applies in a particular case or in a

class of cases described in the guideline.

290. We have reviewed the history of the Guidelines relating to group complaints pursuant to ss.

12 through 15. Section 14 of the Guidelines is subordinate legislation which is authorized by s.

27(a) of the Act. The question of interpretation of subordinate legislation arose in the Tribunal's

Phase I decision, supra, in reference to s. 9(a) of the Guidelines and to the term "sexual

bias". The Tribunal noted the presumption in favour of the validity of regulations authorized by

their enabling statute at the time. We referred to Pierre André-Côté in Interpretation of

Legislation in Canada, 2nd edition, and his remarks at p. 310 as follows:

Finally it must be pointed out that the regulations are not only deemed to remain intra vires, but

also to be formally coherent with the enabling statute.

291. The Alliance submits the Guidelines stand as a proper exercise of the Commission's

discretion under s. 27(2) of the Act and are consistent with the Act as a whole, including s. 11 in

particular, and thus constitute an appropriate basis for the achievement of pay equity. The

Alliance argues the central issue concerning s. 14 of the Guidelines requires an examination of

the enabling Act to determine if the Guidelines are valid.

292. The Commission argues for the well established principle of the presumption of coherence

governing legislative interpretations. The provisions of a statute are presumed to be coherent

and intended to be applied together as part of a functioning whole. The presumption of

coherence is applied to both statutes and regulations. It is presumed that regulatory provisions

are meant to work together, not only with their own enabling legislation, but also with other Acts

and other regulations as well. The governing principle is set out in Driedger on the Construction

of Statutes by Ruth Sullivan, 3rd ed., (Toronto: Butterworths, 1994) at p. 176:

Governing principle. It is presumed that the provisions of legislation are meant

to work together, both logically and teleologically, as parts of a functioning

whole. The parts are presumed to fit together logically to form a rational,

internally consistent framework; and because the framework has a purpose the

parts are also presumed to work together dynamically, each contributing

something toward accomplishing the intended goal.

The presumption of coherence is also expressed as a presumption against

internal conflict. It is presumed that the body of legislation enacted by a

legislature does not contain contradictions or inconsistencies, that each provision

is capable of operating without coming into conflict with any other. As La Forest

J. wrote in Friends of Oldman River Society v. Canada (Minister of Transport):

There is a presumption that the legislature did not intend to make or empower

the making of contradictory enactments.

In J.A. MacKeigan v. Royal Comm. (Marshall Inquiry) McLachlin J. wrote:

I start from the fundamental principle of construction that provisions of a statute

dealing with the same subject should be read together, where possible, so as to

avoid conflict...In this way, the true intention of the Legislature is more likely to

be ascertained.

293. Cases submitted by the Respondent raise the question as to whether a particular body

exceeded powers granted to it by its enabling legislation in making the regulation in

question. These cases illustrate an exercise of excessive powers by regulating authorities relating

to the specific subject matter specified in their respective parent statutes.

294. The first of these cases is Utah Construction & Engineering Pty. Ltd. v. Pataky, [1966]

A.C. 629 (Privy Council). This case involves an appeal from the Supreme Court of New South

Wales challenging a regulation made pursuant to the Scaffolding and Lifts Act of New South

Wales. The respondent had sustained injuries while employed by the appellants in tunnelling

operations and had claimed damages on the ground it had failed to comply with the requirements

of the regulation. The Supreme Court of New South Wales reversed an earlier decision which

had found against the respondent on the grounds the regulation was ultra vires the legislation.

295. The Privy Council allowed the appeal by restoring the initial judgment and finding that the

regulation in question had extended the scope or general operation of the enactment and could

not be read in conjunction with the empowering provision of the parent legislation. The Privy

Council held the ancillary regulation, which imposed an absolute duty of protecting the drive and

tunnel in which the employees worked, extended the scope of the subject matter in the parent

enactment relating to the manner of carrying out excavation work. The Privy Council found the

regulation was not a valid exercise by the government of New South Wales of powers granted by

the Act and ultra vires the powers conferred by the general enactment.

296. Another case concerned a decision of the Supreme Court of Canada in Metropolitan

Toronto v. Village of Forest Hill, [1957] S.C.R. 569. The issue in that case was whether the

Municipality of Metropolitan Toronto was empowered under s. 41 of its charter to pass a bylaw

to bring about fluoridation of its water supply. By s. 41 of the Municipality of Metropolitan

Toronto Act, the Council was empowered to pass bylaws, inter alia, "to secure to the inhabits of

the Metropolitan Area a continued and abundant supply of pure and wholesome water." The

Municipality had taken the step of fluoridation for the purpose of promoting the health of the

teeth and the elimination of tooth decay amongst the inhabitants of the metropolitan area. The

majority decision by Rand J. held the addition of fluoride to the water served a distinct and

different purpose than promoting the ordinary use of water as a physical requisite for the

body. The Court found the special health purpose of fluoridation was beyond the power of

Municipal Council which was to provide a supply of pure and wholesome water. Respondent

counsel referred to comments of Cartwright J. who agreed with the majority at p. 580:

The question is as to the power of the council to enact the impugned by-law,

and the answer depends upon the nature of the subject-matter to which it

relates...Its purpose and effect are to cause the inhabitants of the metropolitan

area, whether or not they wish to do so, to ingest daily small quantities of fluoride,

in the expectation which appears to be supported by the evidence that this will

render great numbers of them less susceptible to tooth decay. The water supply is

made use of as a convenient means of effecting this purpose. In pith and

substance the by-law relates not to the provision of a water supply but to the

compulsory preventive medication of the inhabitants of the area. In my opinion

the words of the statutory provisions on which the appellant relies do not confer

upon the council the power to make by-laws in relation to matters of this sort.

297. The Respondent also relied on Ainsley Financial Corp. et al. v. Ontario Securities

Commission et al. (1993), 106 D.L.R. (4th) 507 (Ont. Gen. Div.). The proceedings before the

court concerned the validity of a policy statement issued by the Ontario Securities Commission

(O.S.C.) and its jurisdiction to promulgate policy statements. The O.S.C. had issued a policy

statement stipulating that it expected security dealers to comply with a policy which contained

detailed and restrictive measures regarding the trading of speculative penny stocks. The Court

held that the O.S.C. lacked the statutory mandate providing it with jurisdiction to issue such a

policy. In his decision, Blair J. referred to Pezim v. British Columbia (Superintendent of

Brokers) (1992), 96 D.L.R. (4th) 137 (B.C.C.A.) which involved a somewhat analogous situation

to Ainsley, supra. Blair J. quoted a passage from Pezim, supra, at p. 527:

Without reaching any decision about whether there is any power in the Commission to inquire

into and impose penalties for conduct falling short of what the Commission judges to be a proper

standard of conduct for those engaged in the securities business, it is my opinion that where the

particular type of conduct that is being considered is conduct that is so closely governed by

legislative provisions as is the conduct relating to disclosure of material changes or material

facts, the Commission does not have the power to impose different and more exacting standards

than those specifically adopted and imposed by the legislature and then to make penal orders for

a breach of those standards which is not a breach of the legislative standards...

That is not to say that higher standards are not desirable. That is a question of

careful policy judgment. But they should not be regarded as mandatory where the

legislature, in balancing the policy considerations, has specifically chosen not to

make them mandatory.

298. Having referred to Ainsley, supra and Pezim, supra, the Respondent concludes as follows

in Volume 244 at p. 32625, lines 9-16:

Now, that's what we're saying here, is that some people may consider very desirable to go to on-

average fairness. But if the legislature hasn't provided for on-average fairness it's not in the

power of the Canadian Human Rights Commission to say, Well, we think pay equity requires on-

average fairness when the legislature has really provided for eliminating discrimination and that

there is a difference between those. On that premise, we say that the regulation is invalid.

299. Section 27(2) of the Act, which gives the Commission its power to issue guidelines is not

restrictive in the sense that it does not specify or limit the scope of the subject matter of the

regulations as is found in the cases referred to by the Respondent. The key phrase in s. 27(2) of

the Act reads "...the extent to which and the manner in which, in the opinion of the Commission,

any provision of this Act applies..." The cases relied on by Respondent counsel do not confer on

the subordinate regulatory agency's broad discretionary power found in s. 27(2) of the Act and

are distinguishable from the power conferred on the Commission by that section. The issue is

not whether the Commission has exceeded its authority under s. 27(2) of the Act. The crucial

issue is whether s. 14 of the Guidelines provides a means for implementing the principle of

equality and "equal value" which s. 11 of the Act requires.

300. Section 14 of the Guidelines provides little or no guidance to the parties or to the Tribunal

in their search for a suitable mechanism for implementing these principles. One is left in a state

of uncertainty due to a lack of direction in the wording of s. 14, when addressing the technical

difficulties and complexities which must be sorted out in its application. The problem with the

phrase "occupational group", for example, is illustrated by the testimony of Dr. Weiner, which

we have already referred to, when she was being cross-examined about s. 14. We again refer to

Volume 10 at p. 1442, line 7 to p. 1443, line 4:

A. The word "group" here seems to refer to an occupational group.

Q. Yes.

A. All right, and then in 14, it seems to say that the complaining occupational

group must -- that you can never change that. But if there is other occupational

groups, so more than one male occupational group against whom the complaint is

made, I read Section 14 to say that these occupational groups are deemed to be

one group. So, that the boundaries of that group change.

Q. We will get into that, but my question is, in your view, given we don't have a

specific notion of the term "group" here, because it is not defined, in your view, is

it important to stick with it once we have chosen that notion and not change the

definition of "group" as we go along?

A. Consistency always sounds like a good idea in general. Sometimes when you

get specific pieces you find out why it's called the "hobgoblin of small

lines." [emphasis added]

301. According to Respondent counsel, s. 14 of the Guidelines infringes the so-called "zone of

non-discrimination" thereby raising policy considerations other than discrimination. (Volume

243, p. 32479). Simply put if four male-dominated occupational groups receive different wages

but perform work of equal value to each other and equal with the female-dominated occupational

group are then blended into a single comparator group, we are no longer correcting for

discrimination based on gender. Since the differences in wages among the male groups cannot

be attributed to gender-discrimination, combining the four male-dominated occupational groups

into a "deemed" group would effect differences in wages caused by factors other than

discrimination.

302. The goal of s. 11 of the Act is to provide for an effective remedy for wage

discrimination. The historical context leading up to the inclusion of s. 11 of the Act in 1979 was

a recognition by such agencies as the ILO, the United Nations, the Royal Commission on the

Status of Women as well as the general public that women's work was being undervalued and

that remedial legislation was necessary.

303. Both Dr. Weiner and Dr. Armstrong testified about pay practices failing to value female

work which differs historically from male work, which differentiation continues to this

day. Gradually over time, with an increasing number of women in the work force and the

changing nature of the tasks to be performed, a more objective and analytical approach to the

value of work done by women has occurred. This led to the recognition of "systemic

discrimination" by legislators, by the courts and by society in general.

304. In our view, s. 14 of the Guidelines is intended to facilitate how comparisons between male

work and female work can be accomplished. It allows for the blending together of the male

work. We are also of the opinion, s. 14 of the Guidelines is compatible with s. 11 of the Act. It

simply provides a means by which comparisons can be made between male work and female

work. It identifies the male values from male-dominated groups which are available for

implementing a wage adjustment methodology.

305. We find the reference to "occupational group" refers to the groups which have been

designated by the application of s. 13 of the Guidelines as either female-dominated or male-

dominated. The term "deemed to be one group" describes how the male comparators will be

treated for purposes of wage adjustment methodology. We further find the reference to

"occupational group" does not mean that comparison between male work and female work has to

be by whole occupational groups designated by the Respondent's classification system.

306. As we have heard from the pay equity expert, Dr. Weiner, there is a variety of different

wage adjustment methodologies which can be used to achieve pay equity, none of which is

referred to in s. 14 of the Guidelines. We have determined if the difference in wages is

established through a methodology that adheres to the principles of "equal value" and "equality"

pursuant to s. 11 of the Act, that is sufficient proof of a prima facie case of discrimination based

on gender. We will examine the different methodologies proposed by the parties within that

framework.

307. Based on our findings we reject the arguments and submissions of the Respondent

concerning the invalidity of s. 14 of the Guidelines.

E. Section 15 of the Equal Wages Guidelines

308. Some guidance is provided in s. 15 of the Guidelines in determining how comparisons are

to be made. It refers to direct and indirect comparisons. Section 15 of the Guidelines reads:

15(1) Where a complaint alleging a difference in wages between an occupational

group and any other occupational group is filed and a direct comparison of the

value of work performed and the wages received by employees of the

occupational groups cannot be made, for the purposes of section 11 of the Act, the

work performed and the wages received by the employees of each occupational

group may be compared indirectly.

(2) For the purposes of comparing wages received by employees of the

occupational groups referred to in subsection (1), the wage curve of the other

occupational group referred to in that subsection shall be used to establish the

difference in wages, if any, between the employees of the occupational group on

behalf of which the complaint is made and the other occupational group.

309. This case deals with group complaints and each party had acknowledged that direct

comparisons between complainants and comparators are not possible in these group

complaints. The difficulty with direct comparisons is aptly explained in paragraph 139 at the

Commission's written submissions, as follows:

(139) Due to the improbability of every male observation [job evaluation] from an occupational

group in a sample or population from the employer's workforce having exactly the same value

for their work as that identified for the female work in the employer's establishment, direct

comparisons are impossible in most group complaints.

310. An admission by Respondent counsel on that point is found in Volume 239 at

p. 31742, line 23 to p. 31743, line 6. After referring to s. 15 of the Guidelines, Respondent

counsel submits as follows:

I think that it's not controversial before this Tribunal that we cannot compare the work and

wages of occupational groups directly in this case. The work cannot be -- the work and wages,

rather, cannot be compared directly. We have that from Mr. Sadler and Mr. Durber and, in my

submission, there is no controversy over that. So, this provision [s. 15 of the Guidelines] says

then it may be compared indirectly. [emphasis added]

And further on he states in Volume 239 at p. 31745, lines 7-10 as follows:

We know that we cannot make direct comparisons in the case before the Tribunal, we must

make indirect comparisons. That's the only option open to us.

311. Counsel for the Alliance informed the Tribunal earlier on during the hearing about the

necessity to do an indirect comparison. In Volume 249 at p. 33337, line 9 to p. 33338, line 2:

Whenever you have got something that is more complicated than that, or to put, I suppose, in

the positive, in a situation where you have got varying pay plans, varying declassification

standards, and you have got jobs that don't line up like that, then you are into indirect

comparisons and you are into regression lines.

That is why when we get to the provincial legislation virtually everybody --

everybody goes to a regression system of one sort or another because it is just not

possible to make that nice, clean job-to-job comparison that is one method that

Dr. Weiner spoke about.

So in my submission, what we are dealing with here clearly is an indirect

comparison within the meaning of section 15 and so the wage curve that is

contemplated in that section is applicable to this case, and I don't understand

anybody to be saying otherwise.

312. The Respondent's interpretation of the meaning of "direct comparison" and "indirect

comparison" in subsection 15(1) of the Guidelines is summarized in its "Notes For Oral

Submissions of the Respondent". Paragraphs 20 and 21 on p. 6 and paragraphs 22 and 23 on p.

7, which are reproduced here, are illustrative of the circuitous arguments of the Respondent in

this respect:

20. If the expression "direct comparison" means comparison with a group performing work of

equal value, then the expression "indirect comparison" in subsection 15(1) of the Guidelines

would mean comparison with a group performing work of unequal value, and that provision

would be invalid as being inconsistent with section 11 of the Act. Dr. Weiner may have

described this kind of indirect comparison, which is allowed in Ontario, but no party has

contended for that interpretation of subsection 15(1) of the Guidelines.

21. In any event, it is not necessary for the Tribunal to make a determination on

this point because subsection 15(1) provides for indirect comparisons only when

direct comparisons "cannot be made". In order to hold that "direct

comparison...cannot be made", the Tribunal would have to find that there is no

male occupational group performing work of equal value. In this case,

comparison can be made with a group performing work of equal value for every

complainant group.

22. If the expression "direct comparison" in subsection 15(1) means comparison

without use of a wage curve, then the expression "may be compared indirectly"

means comparison using a wage curve to represent the wages and value of work

of the occupational group which is the comparator. In that event, section 15 is

consistent with section 11 of the Act, and has operative effect according to its

terms. It is respectfully submitted that this is the interpretation that ought to be

adopted by the Tribunal.

23. Thus, a correct interpretation of section 15 of the Guidelines and section 11

of the Act leads to the conclusion that, as a matter of law in this case, the wage

gap is to be established by comparison with the wage curve of an occupational

group performing work of equal value. [emphasis added]

313. Dr. Weiner explained indirect comparison in reference to s. 15 of the Guidelines in Volume

7 at p. 1049. In her opinion, an indirect comparison is one that involves the use of a wage curve

(wage line) to determine equal value.

314. Mr. Durber from the Commission was asked to describe the Commission's interpretation of

s. 15 of the Guidelines. He addressed s. 15(1) in Volume 146 at p. 18084, lines 2-25:

Q. Let's move to section 15(1) of the Equal Wages Guidelines.

This is the provision that says where direct comparisons cannot be made.

A. Yes. I think that it is an interesting progression of thought here; that is, from

direct comparisons to indirect comparisons.

It is analogous to section 11(1) and (2) under individual complaints where I

suggested that under 11(1) we had two individuals, we made direct

comparisons. Under 11(2) we could not make a direct comparison but made a

comparison based on a weighted average, an indirect comparison.

This is in the group context. It is difficult, actually, to conceive of a direct

comparison between groups, but I can give you an example. What this section

does is very similar to what 11(2) does. It doesn't use the word weighted average,

of course, but it does say you can make indirect comparisons; that is, what you

can do is you can create that one point of reference for the females and one point

of reference for the males that you have to compare.

And further on he says:

I would say that without some measure of indirectness, the business of pay equity, except where

we have two individuals involved, would be impossible. Without indirect comparisons it might

be very difficult for anyone to implement section 11(1) of the Act. [emphasis added]

315. Mr. Durber testified regarding the Commission's views on s. 15(2) in the same volume at p.

18090, line 25 to p. 18092, line 12 as follows:

Q. Can you continue on with the Commission's view of section 15(2) of the Equal Wages

Guidelines?

A. Yes. This subsection of the Guidelines makes reference to a concept which

has come before the Panel before; that is, a wage curve or a wage line.

The wage curve is to be used for the comparator. I would read this subsection

broadly again. I mentioned my interpretation under section 14 of the term

"group". I would view the group as that group of comparators of equal value. So

whatever observations one has, whatever that group is, whether it be more than

the six draftspersons who I called a group, I wouldn't have to draw a wage line.

The wage line for those 12 draftspersons would be a single point, so I wouldn't

have to draw a wage line. But if I had technicians and other jobs thrown in, I

might well wish to reduce all of them to a series of points. And when we draw a

line between those series of points, of course, what we have is a curve. So a wage

curve is simply a moving series of points across a range of value.

The only reason for that, again, is to permit comparison. I view this section as

enabling comparison through the use of wage curves, whatever they look like.

I don't think there is anything in this subsection that tells us what shape a wage

curve is to have, how many of them we are to have. I view this as an injunction

that we are to use all of the information we have and create a moving reference

point which we are calling a wage curve that is reasonable, that summarizes the

information on comparators.

316. Mr. Sunter provided the following understanding of s. 15 of the Guidelines in Volume 107

at p. 12856, line 22 to p. 12857, line 13 as follows:

Section 15, if I may go on here -- and it seemed to be fairly clear from section 11 that the intent

of section 11 was to compare work of equal value and that that could not be interpreted to mean

that, for every particular female job, you have to find a particular male job of the same value. It

seemed to say, and indeed section 15 seems to make clear, that indirect comparisons are

acceptable, which I take to mean through the kinds of technique that I have been illustrating for

regression analysis, for example, which is an indirect way of making comparisons.

It was my belief that everything I did, and it is still my belief that all the analyses

I did were completely in accordance with section 11 of the Act and sections 14

and 15 of the guidelines.

(i). Direct and Indirect Comparisons

317. According to Dr. Weiner comparisons between male and female jobs may involve either

direct or indirect comparisons.

318. Dr. Weiner described direct comparison in the following manner. Points are allocated to

jobs evaluated by a job evaluation plan specifically designed for pay equity job evaluation. The

points assigned are based on certain criteria similar to those outlined in s. 11 of the Act, i.e., skill,

knowledge, responsibility and working conditions. In a direct comparison a male job having the

same number of points as a female job is selected for comparison. Dr. Weiner testified male and

female jobs assigned the same number of points are considered to be of equal value. The wages

of the two jobs are then compared to determine whether they are receiving the same

compensation. If a differential exists the female wage is adjusted to the male wage. Dr. Weiner

described this type of comparison as a "job-to-job" approach.

319. Dr. Weiner testified in indirect comparisons pay equity is achieved for female jobs based

on their value relative to male jobs regardless of whether there is an actual male job of a

particular point value comparable to a female job. Comparisons are accomplished through the

use of a "wage line". According to her wage lines are used in conventional compensation

systems. Dr. Weiner explained that in indirect comparisons the wage line is treated as the male

wage line and female jobs at a particular point value are then compared to the male line.

320. Dr. Weiner testified in pay equity wage adjustment methodology a wage line is drawn to

represent a "pattern" of male jobs. In that way the pattern of male jobs is captured by the wage

line.

321. A "wage line" is also called a "pay line", a "pay term line" or a "policy line". According to

Dr. Weiner the line is called a "pay line" because it provides information about salary and value

and the relationship between them. Generally as job values rise salaries also rise and the line

tends to slope upward. For this reason it is also called a "trend line" to express an upward

sloping trend. Dr. Weiner said the wage line approach has been used in some of our Provincial

jurisdictions including Newfoundland, New Brunswick, Prince Edward Island and Manitoba.

322. Dr. Weiner explained that in statistical terms the line is called the "regression line" or "line

of least squares". Through a statistical technique called regression analysis the line is placed

such that the total squared distance between all the points in the line is minimized. Regression

analysis produces a mathematical equation which expresses wages as a function of value. The

regression line can be illustrated by way of a graph with intercepting horizontal and vertical

lines. The horizontal line (the x axis) represents job value and the vertical line (the y axis)

represents wages. Male jobs are plotted on the graph using job value and wages. According to

Dr. Weiner a wage line is drawn to capture the employer's pay policy for the male jobs. In the

statistical process a computer is used to summarize the information and draws a line using the

formula that best fits the particular set of male job data plotted on the graph.

323. The pay equity and statistical experts testified that regression analysis can produce different

kinds of wage lines for the male data. The line can be segmented, composite, straight, quadratic

(curved), with bends or "dog-legs", stepped or any combination of these various options. There

are sophisticated statistical techniques to measure the quality and reliability of the line.

324. Dr. Weiner testified a wage line is an averaging process because it recognizes jobs at lower

and higher values and therefore, by definition, some jobs will be below the regression line and

some jobs above the regression line.

325. She testified before the advent of pay equity wage lines were used by organizations to

identify their policy as between value and wages. In the following excerpt from her testimony

found in Volume 7 at p. 1020, line 19 to p. 1021, line 16, she describes the unique aspect of

using wage lines in the context of pay equity:

A. These charts all use the very same placement of the male jobs and the female

jobs. Again we've used "M" to denote male jobs and "F" to denote female jobs.

With this methodology you use the male jobs to draw the male wage line. This

aspect of using wage lines is unique to pay equity, otherwise people tended to

draw wage lines looking at all the jobs. But the methodology adapts itself

beautifully into being able to identify if there are any different patterns of

relationship between value and salary for female versus male jobs.

THE CHAIRPERSON: Excuse me, what did you say was unique?

MS. WEINER: It's this idea of drawing wage lines with the just the male jobs

and doing something like a wage line for female jobs. It's taking that existing

methodology and saying, well, if we look at the wage line of a male job and a

female job we will have a sense of having any inequities. So it's an existing

methodology applied to the problem.

326. The rationale for adjusting wages using a male line provides a methodology for comparison

in situations where there is no male job of equal value to the female job, (i.e. direct

comparisons). According to Dr. Weiner, the line tells what the male jobs at the same value of

the female jobs should be paid to be consistent with other male jobs. She testified in Volume

106 at p. 893, lines 1-11 as follows:

Indirect comparison is a means of assessing any pay equity adjustment due to female jobs,

based on their value relative to the compensation for male jobs based on their value, regardless

of whether there is an actual male job of a particular value or not. What we will see when we use

wage lines is that you can identify how much a particular male job at 100 points would or should

be paid, regardless of whether there is in fact a male job of 100 points that exists in this

organization.

Then in Volume 7 at p. 1023, lines 13-25, she stated:

So when you draw the one line, the male wage line, any female job is brought up to it. One of

the things I kind of referred to yesterday, we can now see here is a female job that has no male

job of equal value, but it can easily go to the wage line. There is a stretch of wage line covering

that value, even though there's no male job, so we do the same thing with this female job running

to the wage line as we do with all the others. The fact that there is no particular male job there is

irrelevant. We know what the male job should be paid, given that value, to be consistent with

other male jobs.

327. Dr. Weiner testified to eliminate systemic discrimination in an employer's pay system,

different wage line approaches are used to achieve pay equity. She discussed both the

Commission's level-to-segment approach and the Alliance's level-to-composite approach as

common techniques for achieving pay equity.

328. Dr. Weiner described for the Tribunal the difficulty of finding appropriate male

comparators of the same value as the female jobs when dealing with large group

complaints. The problems are amplified because comparisons are based on subjective

judgements and one has to grapple with the question of whether jobs at 100 points differ

significantly from jobs at 110 points. There may be a variety of wages paid for male work in a

range of values which may not have points that are of equal point value to the female work in

issue. According to Dr. Weiner the methodology of drawing regression lines has been adapted to

identify whether or not there is a different pattern of relationships between value and salary as

between female jobs compared to male jobs. (Volume 7, p. 1021).

329. Mr. Sunter testified he was required to do indirect comparisons because the data from the

JUMI Study did not have male points for each of the female points which would enable him to

make a point-to-point comparison. By the application of regression analysis Mr. Sunter was able

to draw separate segmented lines for all the male scores that fell within the range of values for

each female level. Thus he had the ability to do a comparison between the female and male

work and to identify the existence of a wage gap. Mr. Sunter explained this approach in Volume

108 at p. 13012, line 8 to p. 13014, line 22 as follows:

THE WITNESS: Let me step back a little. What I would like to do, if I had the

data and if the data were available, would be to have a set of male wages for each

of these female points and corresponding to each female point I would have a set

of male wages. Let's say, for this point here I would have a set of male

observations at the same value, and then I would take the average of their wages.

That is not possible because we don't have such a set of points. But, if that were

the case, then I would have a very simple estimate of the wage gap. All I would

do would be to take for each point the distance from the female wage to the

average of the corresponding male wages and then I would average those

distances. That would be a simple estimate within the descriptive class of

applications, and there would be no argument about it. No model would be

involved --

THE CHAIRPERSON: You wouldn't need a regression line.

THE WITNESS: You wouldn't need a regression line.

THE CHAIRPERSON: You would just be going point to point.