Counterintuitively, the representation of women and minorities and bias were uncorrelated

From What Happens Before? A Field Experiment Exploring How Pay and Representation Differentially Shape Bias on the Pathway into Organizations by Katherine L. Milkman, Modupe Akinola, and Dolly Chugh. Abstract.

Little is known about how bias against women and minorities varies within and between organizations or how it manifests before individuals formally apply to organizations. We address this knowledge gap through an audit study in academia of over 6,500 professors at top U.S. universities drawn from 89 disciplines and 259 institutions. We hypothesized that discrimination would appear at the informal “pathway” preceding entry to academia and would vary by discipline and university as a function of faculty representation and pay. In our experiment, professors were contacted by fictional prospective students seeking to discuss research opportunities prior to applying to a doctoral program. Names of students were randomly assigned to signal gender and race (Caucasian, Black, Hispanic, Indian, Chinese), but messages were otherwise identical. We found that faculty ignored requests from women and minorities at a higher rate than requests from White males, particularly in higher-paying disciplines and private institutions. Counterintuitively, the representation of women and minorities and bias were uncorrelated, suggesting that greater representation cannot be assumed to reduce bias. This research highlights the importance of studying what happens before formal entry points into organizations and reveals that discrimination is not evenly distributed within and between organizations.

The study seems in most ways to be much more robust than many I see. My primary concern is that the study does not include raw numbers. They indicate that 67% of professors responded in some fashion but other than that, nothing. All the other measures are relative or percentages which can sometimes mask issues. Other than that, this seems reasonably solid.

This is the second study in a couple of weeks that I have seen indicating that the bias is institutional rather than individual, in other words, that African-American professors discriminate to the same degree as White professors against minority applicants ("Counterintuitively, the representation of women and minorities and bias were uncorrelated, suggesting that greater representation cannot be assumed to reduce bias.")

I have long assumed that there was a lot of discrimination going on against all parties, and not necessarily from a group perspective (race, gender, class, religion, orientation, etc.) but from an operational/institutional perspective. Doesn't make it less bad, but sometimes it makes it more understandable and therefore more amenable to targeted solutions. I have always suspected that the victim categories (race, gender, etc.) are something of a red herring obscuring a much more complex reality.

What are the factors that might be causing professors to be less responsive to minorities and women (independent of their own race and gender)? Being an economist, I of course instantly fall back on trying to understand incentives. What are the risks and rewards to which professors are responding?

And why are there differences based on gender and race? Chinese Females appear to be the most discriminated against followed by Indian Males, then Chinese Males, then Indian Females, then African-American Females.

Why are the STEM fields most receptive to unsolicited contacts from prospective PhD students as opposed to the nominally more people oriented fields of Business, Education, Human Services and Health Sciences?

We already know some of the issue related to gender. Women depart from full time employment at higher rates and for longer durations than do men. If you are a professor, how might that influence your inclination to respond to a female expressing interest in the field? To what degree are response rates shaped by professor's anticipated network ROI calculations? In other words, given limited time and competing demands, there is a limit to the amount of time that can be spent on career coaching and mentoring. There is a real return on creating dense homophilic affiliation networks in most fields, presumably including academia. The implication though is that you probably want to invest your limited time in responding to, coaching and mentoring candidates who are most likely to 1) be around, 2) be continually productive, and 3) have some probability of elite accomplishment. Does that informal heuristic calculation lead professors to consciously or unconsciously favor male candidates? I don't know but it is a logical inference.

But what about minorities? And in particular, why such a strong negative showing for what are otherwise characterized as "model minorities" (i.e. Asian and Indian)? I can only speculate. I am guessing that there might possibly be three things going on.

Most insidious would be if there is indeed, as is widely speculated, an informal cap on Asian/Indian applicants to the most prestigious programs. If that really exists, then it would serve as a disincentive to spend time on such interested candidates. I kind of doubt that this is the case 1) because I hope it is not true and 2) even if it is, my impression is that all the speculation is centered on undergraduates, not graduate and PhD programs.

Beyond that obvious one, the only strong explanatory candidate I have is the possibility that such professors may see a large volume of international Asian and Indian candidates who present complications towards their acceptance in a program. If you are interested in maximizing the ROI on your affiliative homophilic network, you are likely to discount time invested in candidates where there might be language, culture, financing, foreign credential validity issues and green card issues. This hypothesis could have been tested if there had been in a sampling of the letters an explicit disclaimer of citizenship/residency issues. According to this hypothesis, those already demonstrating residency/citizenship should have had higher response rates. I am guessing residency status concern might be a significant factor in the differential response rates.

Why else might minority professors along with their majority colleagues respond differentially to minority candidates? Is there a markedly different attrition rate between minorities and majority? That would certainly be a reason. Are there issues in the grant application process that makes it harder to fund work by minority (non-citizen) candidates? Is there something about the unsolicited contact process that holds some bias? I am running on fumes here. I am certain someone in academia might highlight additional operational reasons that might be relevant.

All in all a very interesting study that formalizes, structures and documents that there is a real pattern of disparate response. But whether the disparate response is warranted for non-obvious operational issues remains unclear. What does seem clear is that disparate impact is based more on operational issues than on majority bias. If all professorial respondents (minority and majority) are equally disparate in their response, then that suggests there is either shared discriminatory biases, which seems unlikely, or that they are responding to some set of shared experiences/knowledge that we are not taking into account.

One other factor suggests to me that this is more operational than discriminatory bias. It is well documented that universities are sharply skewed in political orientation compared to the general populace. Very roughly the general population runs self-identified conservative 40%, moderate 40% and liberal 20% whereas most humanities fields run self-identified conservative 10%, moderate 10% and liberal 80%. From an ideological perspective the stereotype is of liberals being the most unaccepting of discrimination. Now there is no reason that prevents them from holding the view that discrimination is completely wrong and yet being severely discriminatory. Everyone is self-deceptive of the gap between belief and practice. However, these gaps are pretty large.

AN interesting study that provides a platform for much deeper investigation. My suspicion is that the root causes are going to be found to be operational rather than personal bias, but that remains to be determined.