- November 29, 2018
[Author’s Note: A widespread modern canard is the notion that women continue to face major discrimination in pay and employment throughout American society, including in academia. Irrelevant or misinterpreted statistical data are typically put forward in support of such claims. A recent example (1) was notable because it was not only put forward by a reputable statistician but also published in a prestigious statistical journal: Significance, the official magazine of the Royal Statistical Society and the American Statistical Association. The commentary below is a critique of that article, submitted to the same journal on April 27, 2016. The response was quick and positive. On April 29, Editor Brian Tarran wrote, “Many thanks for your detailed response to Amanda Golbeck’s article. We don’t have space for this in our upcoming June issue, however, I would like to include it in our August [issue], with a reply from Amanda to the points made.” But over following months publication kept being postponed, and Golbeck came onto the editorial board. Finally, after ten months, Editor Tarran wrote to say that Significance could not publish it because at more than 1500 words it far exceeded “the ideal length (of between 240 and 350 words)” for such a piece, a limitation nowhere else mentioned for the journal. So here we are. Download as a PDF.]
Hurlbert is Professor of Biology Emeritus at San Diego State University.
Amanda Golbeck's essay, "Are women underrepresented in the American statistics profession" (1) will ring a bell with -- and often raise the hackles of -- many academics whose feet have been held to the fire by administrators bearing down on admissions committees and search committees, demanding "more diversity, whatever it takes." The essay misleads in three ways: by starting with a false premise, by its implicit equating of "gender imbalance" and "underrepresentation" with unjust discrimination, and by its use of a biased measure of true availability of women in the relevant labor pools.
Golbeck's key premise is that, "Following standards established in the early 1970s for affirmative action planning towards federal compliance -- standards that continue to be widely used today in diversity planning in the USA -- we would want women and men to be represented among our [statistics] graduates and faculty in approximately equal numbers, as this would mirror the gender distribution in the general population."
Who she intends to be included by her editorial "we" is unclear. But the original idea of affirmative action was to increase equality of opportunity, not to force equality of result, not to increase skin-color or gender "diversity", or to promise a "critical mass" (e.g. in a student body) of any sub-group that might demand it. If her editorial "we" is meant to imply that the government, most academics or society at large approve taking measures to make the sex ratio (or, for that matter, the racial, religious or political composition) of members of any profession or occupation mirror that in the "general population," she is simply wrong. They do not so approve. To even begin moving in that direction would require major violation of key principles in the U.S. Constitution and law, including equality of opportunity, freedom from unjust discrimination, and treatment of individuals on their merit.
The data on the statistics profession seem not a special case in any way. Golbeck summarizes new data on the representation of women in doctorate-granting statistics (or biostatistics) departments in 2014: women earned 46% of the PhDs in statistics that year, constituted 35% of untenured, tenure-eligible professors in statistics departments, and only 20% of tenured full professors in such departments. Golbeck acknowledges that there are many other job opportunities for statisticians in addition to those in doctorate-granting institutions, and presumably knows what psychologists and sociologists have been telling us for some time, that men and women in the same field, in the absence of any discrimination, often markedly tend to choose different types of jobs or career tracks for a host of reasons.
Yet Golbeck describes the trend she documents in draconian terms such as "disturbing," "especially unbalanced," "a giant hole in the academic career pipeline," and, more cryptically, mentions the need for hiring faculties to "exercise better critical thinking." Some administrators may take this as a call to make sure persons capable of a very particular type of "critical thinking" are made chairs of faculty search committees. We are far from women statisticians making up the 50% desired by Golbeck for statistics faculties, and mere turnover of faculties is unlikely to eventually get us there absent de facto quota systems.
If Golbeck's criteria for using loaded terms such as "gender imbalance" are not appropriate, what criteria might be? What kinds of data might actually constitute at least prima facie evidence of some sort of discrimination going on?
The best easily available data on the relative availability of the two sexes in the job market for any field is the sex ratios of the applicant pools elicited by position advertisements, ideally averaged over many searches. My opinion on these matters is based on just such an analysis I did several years ago for my own Department of Biology at San Diego State University, when we were being attacked by administrators for being insufficiently "diverse."
My analysis examined data on applicants in 30 faculty searches by our department during the period 1989 to 2002, as well as the departmental record of position offers and hirings during that period (2). In a nutshell, the findings were these. Women comprised, on average, 22% of applicants despite comprising about 41% of recent (1983-2001) biology PhDs awarded to U.S. citizens. Women nevertheless received 42% of our first offers but also rejected offers, first or otherwise, at a higher rate (44%) than did men (28%). Women made up 18-22% of our department's faculty during the period 2000-2002.
A few years prior to my 2003 report on the matter, there was a purposeful twisting of statistical data by our Director of Diversity and Equity to create the impression that search committees in Biology, as well as in most other departments at SDSU, had failed and were continuing to fail to hire women and minorities in proportion to their availability in the relevant workforce. The director objected to me carrying out an independent analysis. When she saw my report her response was that "the raw data on Ph.D. production [is] a more accurate indicator of the sex ratio of the 'available' applicant pools than is the record over many searches of the sex ratios in actual applicant pools." In essence, she felt it reasonable to criticize search committees and administrators for not considering or hiring persons who had not submitted applications!
But her behavior had been exposed, she further embarrassed herself by canceling a search in another department because she didn't like the (fairly diverse) composition of an applicant pool, said silly things to a reporter from our student newspaper, and then took early retirement a couple of years later.
Nevertheless, across academia in the U.S., there remain thousands of diversity czars and czarinas and their acolytes carrying on with this same particular sort of intimidation.
And certainly, it often works to generate preferential treatment of women applicants. Perhaps the most egregious and well-documented case has been provided by the University of California. In the late 1990s, a few women faculty members at UC Davis began vociferously complaining that women professors were getting paid less than males with the same rank and qualifications and were also being discriminated against in faculty searches. California State Senator Jackie Speier than asked the Bureau of State Audits to analyze five years’ worth of faculty hiring data for the entire University of California system looking for discriminatory practices.
And what were the core findings in the Bureau's 111-page report issued in May 2001 (at a presumed cost of many hundreds of thousands, if not millions, of dollars for the whole investigation) (3)? First, it concluded there was no evidence of gender discrimination in salaries. Second, it concluded, "that a significant disparity appears to exist between the proportion of female professors hired [29%] and the proportion of female doctorate recipients nationwide [33%]."
Whether such a difference merits the label "significant" is a subjective matter, but the difference is certainly smaller than that often found between actual applicant pools and recent Ph.D. recipients. The report's own Table 5 shows that for 287 different faculty searches over the nine UC campuses, applicant pools averaged only 20% female while recent Ph.D. recipients averaged 31%. That could be prima facie evidence that in the UC system, as in the SDSU Department of Biology, women applicants had actually been getting preferential consideration.
These facts notwithstanding, the report goes on at great length to claim serious discrimination against women and to recommend a plethora of steps to keep the malevolent, misogynistic forces in the UC system punished, well-monitored, very busy, and under control. Not surprisingly, the UC administrators of the day buckled under and made millions of dollars and dozens of positions available for correcting the putative "gender disparities" in the UC faculties.
The most accurate measure of the availability of women in the relevant labor pool for any academic discipline will be their representation in the actual applicant pools, averaged over many searches by a given department. Women not applying for a position are not available for it, obviously. Yet the Bureau's report argues, without evidence, that discrepancy between the representation of women in applicant pools vs recent PhDs must be due to "deficiencies" in advertising and outreach processes. As such, the deficiencies must be searched out and corrected, and proper punishment meted out to the responsible males.
There is a veritable cornucopia of additional reports and articles on these matters for both the UC system and other universities in the US. Most reflect very poorly on academia's capacity for objective, non-ideological analysis. Shouldn't the American Statistical Association be taking the lead in turning this ocean liner around?
Give Amanda Golbeck the captain's hat and put her on the bridge with a new mandate from the ASA. She's good at multi-tasking and did great things for the statistics programs at San Diego State University during her time here (1983-1996).
(1) Golbeck, A. 2016. Are women underrepresented in the American statistics profession? Significance 13(2): 8-9. https://rss.onlinelibrary.wiley.com/doi/pdf/10.1111/j.1740-9713.2016.00892.x
(2) Hurlbert, S.H. 2003. Race, sex, and faculty searches, Department of Biology, SDSU, 1988-2002, with commentary on policies and actions of the SDSU administration. San Diego State University, San Diego CA. 16 pp. (Now available as: NAS Article, National Association of Scholars, New York NY, September 11, 2017: https://www.nas.org/articles/race_sex_and_faculty_searches_department_of_biology_sdsu_1988_2002 )
(3) California State Auditor. 2001. University of California: Some academic departments need to take additional steps to resolve gender disparities among professors. California Bureau of State Audits, Sacramento CA. 111 pp. https://www.bsa.ca.gov/reports/summary/2000-131