Employment Discrimination and Statistics: Census Data and its Use: Part 1
Besides the 4/5’s or 80% rule to assess discrimination in an organization’s selection, promotion, or termination practices, one may also use census data for determining whether selection practices are discriminatory. EEOC has developed databases in conjunction with the U.S. Census Bureau from the 2000 census data to aid in such analyses.
In the recent proposed definition of Internet applicants, the guidelines appear to give much emphasis to census data for determining disparate impact:
Q: Are all the search criteria that employers use subject to
disparate impact analysis?
A: Yes. All search criteria used are subject to disparate impact analysis. Disparate impact analysis can be based on Census or workforce data. If a disparate impact is shown, the employer must demonstrate that its criteria are job-related and consistent with business necessity for the job in question.
What are some examples of cases where census data have been used?
Plaintiffs have used census data in a number of cases to support their cases involving disparate impact. In Butler v. Home Depot (1997), for example, the plaintiffs’ expert argued that although 15.20% of people hired as merchandising/sales associates were women, 48-49% of the available and qualified sales associates in the labor market were women. Similarly, although 6.40% of assistant manager trainees hired were women, 36-39% of the available and qualified assistant manager trainees in the labor market were women.
In another case, which involved about 454 restaurants, plaintiffs in Rhodes v. Cracker Barrel (2002) attempted to show race discrimination through a number of statistics. One set of statistics involved census data, where the number of African-American entry level restaurant workers were compared to their representation in the county in which they were hired. For the relevant years, the plaintiffs’ experts found that for the dishwasher positions, African-Americans were hired in greater proportion than their representation in the relevant labor market. However, for server positions, African-Americans were hired in much lower numbers than their representation in the relevant labor market.
In Part 2 on Census Data, I will discuss some criticisms of Census data. Stay tuned!
If you enjoyed this post, please consider leaving a comment or subscribing to the feed or by email.
Notice how the examples are “macro’ comparisons.
Unlike the Griggs v. Duke-type focus on a specific job qualifications (see my last post, with the accompanying “rant”), in these cases employers stand accused of discrimination based on the most diffuse statistical “disparities” between workforce representation and general population data, which may have innumerable lawful explanations.
As we shall see, much more refined analysis is available and needed before one may conclude anything unlawful has occurred.