Human Rights: Statistical Evidence

The Ontario tribunal in a 1984 case noted that the use of such evidence may be an effective means of proving the case by such circumstantial evidence: ¹;

Often discrimination is not overt. Rarely does an employer expressly state that it refused to hire a qualified applicant because she was a woman. Acts of discrimination and intent to discriminate are often proved by circumstantial evidence (Re: Windsor Board of Education and Federation of Women Teachers’ Associations of Ontario (1982), 1982 CanLII 5080 (ON LA), 3 L.A.C. (3d) 426, at 430). “Statistical evidence is an important tool for placing seemingly inoffensive employment practices in their proper perspective” (Senter v. General Motors Corp., 532 F. 2d 511 (1976)) …

The tribunal is this instance considered the submission made by the employer that the use of such evidence was similar to similar fact evidence in that it may be used to support a propensity rather than direct evidence. This submission was rejected in a 1982 labour arbitration case. ² The tribunal found that statistical evidence may be used to demonstrate patterns of behaviour and represent a form of circumstantial evidence:

Statistics show patterns of conduct rather than. specific occurrences. Statistics represent a form of circumstantial evidence from which inferences of discriminatory conduct may be drawn (Davis v. Califano, 613 F.2d 957 (1979) at 962). It is within the rubric of “circumstantial evidence” that statistical evidence in human rights cases should be considered. Like all circumstantial evidence, statistics are to be considered along with all surrounding facts and circumstances (International Brotherhood of Teamsters v. U.S., 97 S.Ct. 1843 (1977), at 1857).

Statistical evidence is not the same as similar fact evidence. This evidence may establish a pattern of conduct which will then be considered as circumstantial evidence from which it may be possible to make inferences. It must be considered in the context of all surrounding facts.

The tribunal in Blake continued to observe that such evidence may be used to show racial or sexual distinctions in decisions to hire or promote; or the number of women hired for a particular position compared to the number of such qualified women in the market available for such a position; or reveal that the employer has made subjective decisions in a discriminatory manner; or that tests imposed by the company has a discriminatory impact; or that the employer’s rebuttal evidence intended to show a non-discriminatory reason not to hire is a “mere cover-up for a discriminatory reason”. ³

This being said, such evidence, as noted in Blake, should be best used with oral testimony setting out the specific details of adverse treatment. Such evidence is more significant in the context of a highly trained applicant where the number of persons hired is too small for effective statistical analysis.

In the same decision, it was noted that the use of such evidence is not confined to proving the case. It can also be used to defend the claim. The employer may use statistical evidence to show that it did not discriminate, or that an apparent discriminatory standard is a BFOR, or that the requirements of the position and relevant tests are job related, or to show that the discrimination is justified.

The Blake decision was decided in an era in which it was necessary to prove the intent to discriminate. Even in this context, it was noted that statistical evidence alone may be sufficient to establish a prima facie case. It was then noted that statistical evidence should show “gross disparities” in the treatment of the affected group and such distinctions are “unlikely to be the result of random selection”.

Given a prima facie case, the employer may yet assert that the statistical evidence must be connected to the specific event on which the complaint is based. There may be established, for example, evidence to show a propensity to treat a minority group unfairly. The employer may, however, assert and prove that on this specific occasion, it did not do so.

This in fact was the result in Blake. The employer succeeded in demonstrating that the applicant was interviewed fairly and that the employer made a honest decision that she lacked the qualifications for the position.

Similarly, the employer may assert that the distinction is a BFOR. It may even introduce its own statistics to show this.

Further, the defence may challenge the statistics offered by the complainant by critiquing the qualifications of the statisticians or the methodology employed.

The use of such evidence was considered in a 1996 case ⁴ of the Canadian Human Rights Tribunal. The complaint alleged that the employee had been treated in an abusive manner by his immediate superior. Both sides introduced evidence of expert witnesses intended to influence the outcome based on statistical evidence.

The tribunal determined that the issue in this instance was not relevant in a case of direct harassment. At best, it may provide a context. Even had this evidence demonstrated systemic discrimination towards visible minorities, the tribunal concluded, it would not be proof of harassment or that the employer had encouraged this behaviour. The expert report of the complainant was considered irrelevant in this instance.

In 1998, the Federal Court, upon a review of the tribunal decision ⁵ which disallowed statistical evidence, reversed this decision on this basis and directed a new hearing. ⁶ The applicant had attempted to introduce this evidence to show that visible minorities were not represented in proportion to their availability at certain levels of employment hierarchy. The intent was not to prove systemic discrimination but rather to support an individual complaint. ⁷ The tribunal had ruled that this evidence was not admissible as the applicant was not allowed to introduce evidence to show a pattern of systemic discrimination as it was prejudicial to the employer and it would have allowed a claim to be brought which was not in the complaint.

The Federal court cited the authority of Blake to allow such evidence to show evidence of a systemic problem as circumstantial evidence to infer discriminatory conduct in this instance. The complaint in this case was based on discriminatory practice on the basis of race, colour, and/or national or ethnic origin. It was not one of directly abusive behaviour as in Air Canada.

The matter was returned to the tribunal which bifurcated the hearing. In the liability hearing, statistical evidence was allowed and liability was found. The Tribunal concluded that Dr. Chopra was denied the opportunity to serve in the acting position as a result of discrimination on the basis of his national or ethnic origin.  ⁸

The evidence was admitted before the second tribunal hearing on liability. It was, however, found to be flawed in its creation. Even it were to be accepted, the tribunal found that there was no meaningful difference between the expected results, presuming no adverse treatment, and the actual results. The predicted number of visible minorities expected from the report was three to four while the real number was two. Liability was nonetheless found.

This issue was again considered by the Canadian Human Rights Tribunal in a 2001 decision. ⁹ The evidence was admitted but it was found to be of limited probative value. The persons who were reviewed were too small a number to be of any meaningful significance. This was so, in particular, with respect to the five persons in the applicant’s relevant course. The determination of the statistical methods employed were flawed.

When arguing for the use of such evidence, small data groups should not be used as the results are statistically insignificant. When using such a small group, the results should be subjected to statistical testing to determine if any differences are meaningful. ¹⁰

This remains the law today. As noted by Heureux-Dubé, J. ¹¹

As intent is not a prerequisite element of adverse discrimination, a complainant may build his or her case under ss. 7 and 10 by presenting evidence of the type adduced by the complainant in the present case.  Statistical evidence of professional segregation is a most precious tool in uncovering adverse discrimination.

The Respondent countered that the size of the number of individuals being studied, particularly regarding the total of five persons in Mr. Morris’s QL7 course, is just too small a group from which to draw any conclusions, and any differences that show up between age groups may simply be a result of random fluctuations. ⁽¹⁹⁾ By contrast, in cases such as Blake v. Mimico Correctional Institute ⁽²⁰⁾ and Singh, ⁽²¹⁾ where employment statistics were taken into consideration in assessing the circumstantial evidence of discrimination, the data related to hundreds of individuals. When dealing with small data groups, on the other hand, statisticians regularly subject their findings to statistical testing in order to determine if any differences in the results are statistically significant. ⁽²²⁾ No such testing was conducted in the present case.

👤 About the Author •
📚 Human Rights Index •
🎥 YouTube •
⬆️ Back to Top

1. Blake v Ontario
2. Re: Windsor Board of Education and Federation of Women Teachers’ Association of Ontario (1982)
3. Blake
4. Danhjal v Air Canada
5. Chopra v. Canada (Department of National Health and Welfare), [1996] C.H.R.D. No. 3 (QL)
6. “Chopra” #1 Canada (Human Rights Commission) v. Canada ; upheld in brief reasons by the Federal Court of Appeal
7. S. 10 Canadian Human Rights Act deals with a discriminatory policy or practice
8. August 2001 liability hearing, no review was taken of this decision “Chopra #5”; the remedy decision followed in 2004; this decision was reviewed on the application of Chopra to Federal Court which was dismissed in 2006; that decision was appealed further by Chopra to the Court of Appeal which dismissed the appeal in 2007
9. Morris v Canada
10.   Morris v Canada
11. dissenting, but not on this point SCC 1989 Syndicat des Employés de Production d Québec et et l’Acadie v. Canada (Canadian Human Rights Commission)