Wednesday, July 15, 2009

Some Thoughts About Ricci [Ricci v. DeStefano, 557 US 557 (2009)] and Career Self-Selection

A common statistical problem gets little mention in race and job testing issues. The test takers are not a random sample of the underlying group. The two test taking groups likely have different abilities and characteristics. There is heterogeneity and self-selection.

When one tests similar representative population samples of blacks and whites, one can legitimately expect equal passing rates across sub-groups, i.e., close to the same black and white passing rate for college graduates, high school only graduates, and by family income levels, etc. A failure of a minority group to pass would then be suspicious. However, the minority and white New Haven firefighter test takers self-selected and chose to take the exam. They are not representative samples of their whole New Haven group population.

There are no reasons to think that the self-selection within the black population and the white population were similar. Within each ethnic, minority and racial group there are different job aspirations and job rankings. It is highly likely that the black test takers have different characteristics than the area's general black population and than the white firefighter test takers. For example, in NYC at times, there were many more Irish firefighters and police officers than other ethnic groups. It was self-selection and not bias test results.

To blame test failure as discriminatory, one would first have to show that the characteristics of all the black test takers were similar to all the white test takers. We do not know if the two groups of test taker characteristics were close to equal. If they were not, then we need to determine if the self-selection within the group was due to an institutional bias in the system that prevented minorities from taking the test, such as using ways to make whites more aware of the exam than blacks were or the existence of other institutional barriers to inhibit minority test taking.

Most likely, in these days of civil service job postings, the internet, concerns about racial bias, etc., self-selection within the minority and racial groups led to the disparate results. The blacks with the ability and characteristics to pass the test and be good firefighters chose to try to find other types of jobs and careers and did not apply to be firefighters. Those that chose to try to become firefighters may not have had the skills and characteristics to pass the test and be good firefighters. The 100 percent black failure rate is more a statement about the self-selection process to want to be firefighters in that black community than it is a statement about the discriminatory effects of the test.

One has to remember that individuals do not want to do all the possible jobs and careers. A self-selection process goes on all the time. Different subgroups, such as minorities, education levels, family income levels, etc. will choose different career paths. The characteristics of any minority subgroup trying to enter a particular career or job function may be different and the subgroup's performance in that career path may diverge over time due to the dissimilar characteristics of the different subgroups in that job function or career. Due to job and career self-selection, talking about racial and minority bias in a job or career is different than talking about racial bias affecting an entire minority population. It is legitimate to fret about the high school and college graduation rates of blacks versus whites, because it is about the entire population of the group. However, the freedom to choose to try to develop one's own career path and not discrimination may have led to the different passing rates between blacks and whites on the New Haven firefighter's examination.

1 comment :

  1. Thankz for such a great and informative post, i am sure it will be of great help to the readers.

    ReplyDelete