Tuesday, May 24, 2016
The Pipeline Problem (V): Dear Patriarchy, Pull Up A Chair
Pull up a chair.
I just finished reading Why Women Earn Less: Just Two Factors Explain Post-PhD Pay Gap by Helen Shen in Nature. It's based on a recent study by Catherine Buffington et al (2016) that examines data obtained from the UMETRICS project at the University of Michigan. Both Buffington et al (2016), Nature, and a similar commentary in Science do a serious disservice to the human race and academia by publishing and commenting on the study in its present form.
Firstly, it has been known for many years that the pipeline problem for women in STEM, of which the pay gap is a component, is the result of a gradual accumulation of negative effects (Valian, 1998). Studies such as Buffington et al (2016) that examine too short a time-frame are likely to seriously distort the influences of factors that contribute. Although both Mason and Weinberg mention problems with time frame durations, even five to ten years is not long enough. She Figures 2015 (European Commission, 2015; Figure 6.1), as one example, shows that predominant effects of gender occur over a more prolonged time window.
Secondly, let's examine data selection and statistics. Buffington et al (2016) lump together variables that have intersectional effects such as race and age, and omit categories with potential for insight such as single mothers or fathers with childcare responsibilities. There are other variables not scored in the original data sets such as sexual orientation and couples that chose not to declare their relationship status.
The authors also appear to disregard graduate students that for reasons of insufficient support, harassment, locational restrictions, or otherwise not fitting the traditional academic mold, dropped out during their graduate years in spite of substantial investment in training. In addition, the study neglects undergraduate students that failed to be admitted to graduate programs due to financial and location-specific stressors, or merit-based, ivory tower biases (Goldrick-Rab & Kendall, 2016). These longitudinal biases reflect societal stressors associated with race, gender, class, disability, age and veteran status that result in substantial downward spirals over time. Thus, even though Buffington et al (2016) include some key variables in their study, their inclusion in this short time frame without regard for interactions is nothing short of misrepresentation.
Another problem of data selection and representation concerns the numbers for gender. Many studies, including the European Commission Report, have indicated that the numbers of PhDs granted to women and men are nearly equal in recent years. The fact that there are more than double the number of men in Buffington et al (2016) calls the UMETRICS and Census Bureau's Protected Identification Key (PIK) data into serious question. Numbers for other variables such as race are not reported — the percentages mentioned in the Methods show Black and Hispanic populations are negligiblely accounted for. Because least squares regression aims to minimize the overall variation, the numbers representing variables such as gender and race will affect the results.
Third, I would like to point to an effect that goes unnoticed by Buffington et al (2016), perhaps unsurprisingly given the normalization of sexist bias in societies across the world; namely, women consistently shoulder the burden and blame for motherhood, or lack thereof. Quoting Buffington et al, "Nineteen percent of females and 24 percent of males had children at the time of the 2010 census." A five percent difference in parental status is, in the words of Buffington regarding other observations, a robust effect. More importantly, the observation suggests more men 'get pregnant' than women. Considering the effect of motherhood on women's careers, the finding could further be taken to imply that men are conditioned by capitalist patriarchal expectations to neglect their wives and children. It's beyond time capitalist patriarchy acknowledge and accept responsibility for its share of care.
Finally, Buffington et al (2016) claim that the large majority of the pay gap in their study is accounted for by choice of STEM field. However, they neglect to mention other studies (see Miller, 2016; Slaughter, 2015; and Short Takes on Slaughter in Signs, 2016 ) showing the devaluation of fields as more women are accepted and vice versa.
The titles and recognition resulting from the Nature and Science commentaries also will undoubtly contribute to the longevity of the interpretation and citational rankings. It's likely to cause a serious waste of research dollars, result in unnecessary delays, and negatively affect the lives of dedicated women researchers in STEM all over the world. What a shame.
...A more formal rebuttal will be submitted after I'm finished with my move.
Posted by 真行 at 15:09