September 20, 2010
Still Thinking About Gender Gap IssuesAccording to data from the federal Bureau of Labor Statistics, 2.6% of female workers in the finance industry left the industry in the past 10 years, while male workers increased by 9.6%. In the same period the number of women in the work force increased by 4.1% Young women, in particular, have become rare in banks, brokerage firms, and insurance. There is no clear explanation for this trend. WSJ, Ranks of Women on Wall Street Thin
TrackBack URL for this entry:
Listed below are links to weblogs that reference Still Thinking About Gender Gap Issues:
In analyzing gender bias issues, one of the major assumptions made is that the characteristics of applicants and hires are identical between the two gender groups. I do not mean that either group of applicants or hires are more or less capable of doing an excellent job. I mean the two subgroups' preference rankings for the non-performance related part of the job are different. There is no reason to assume that sociological or demographic subgroups have the same preference rankings for the non-monetary part of the job.
A self-selection process filters applicants that can result in different characteristics and preferences among applicants and hires in different demographic/sociological subgroups.
For example (hypothetically), a significantly larger percentage of men than women may prefer (or dislike less) a significant part of wages as bonus payment or dislike less a highly variable total annual pay; similarly for business travel, working late, and many other (including many unidentified) intangible job qualities.
If one subgroup has a lower preference for some non-monetary, non-essential to job function job trait that is a major part of an industry's job but that some other job in some other industry may not possess, then employees will leave that do not prefer or actually dislike that job characteristic. Similarly, applicants that do not prefer that job characteristic will self-select and not apply for those jobs.
For example, many fewer women than men take jobs where there is a substantial risk of on-the-job injury. Likewise, there is anecdotal (and some statistical) evidence that women will accept a lower salary for increased benefits more often than men will. Men will accept more pay with fewer benefits. Women will accept lower pay with more benefits.
In any measure of two subgroups job performance ability, the two groups will appear equal. However, qualities needed to perform a job do not measure employee job satisfaction, potential self-retention or application rates.
There are also problems when looking for hiring bias in applicant subgroups as opposed to total subgroup populations because different subgroups rank their occupation preferences differently.
As a hypothetical, suppose there are two sociological/demographic groups that rank occupational categories differently. Suppose group A ranks Doctor over Lawyer and suppose Group B ranks Lawyer over Doctor as potential occupations.
The best of group A applies to medical schools and the next tier of its applicants applies to law school. Group B does the reverse with its best applicants applying to law school and their next best to medical school.
Medical schools will either accept students of equal quality with more of group A than Group B or accept equal numbers and then Group A's measurable law school performance qualities will be higher than group B's. The reverse situation will be true for Law schools. The different occupation preference rankings will cause either different acceptance numbers or different measurable qualities, and there is the potential for a claim of bias where none exists at the applicant acceptance level.
The above is the major shortcoming in civil service tests bias lawsuits and decisions. For example, the entire population of two subgroups may have an equal number (percentage) of qualified individuals for the civil service job. However, do to self-selection based on subgroup occupational preferences, there is no reason to assume that the members of each subgroup that decide to take a civil service test are either representative of their entire subgroup population or equal to the other subgroup's test takers. This is why so many municipal civil service tests, such as firefighter, seem to show bias. In some white subgroups, fire and police officer are high occupational preferences, higher than doctors or lawyers. In some minority groups, lawyers, doctors, teachers are a higher occupational preference than fire or police. When the tests are scored, the minority subgroup will appear to underperform the white subgroup. The underperformance is not due to any inherent bias in the test. It is because the two subgroups have different job preference ranking and the subgroups taking the test self-select and do not reflect their entire group's populations and are not equal to each other in ability.
Posted by: Milton Recht | Sep 20, 2010 11:14:44 PM