Over at ScienceDaily there is a report on a new paper on affirmative action and academia, Understanding the Impact of Affirmative Action Bans in Different Graduate Fields of Study. The paper is gated, but the regression model used really doesn't seem to do much more than confirm intuition. The descriptive details are more interesting and straightforward. Here's the major point: states which banned affirmative action in higher education seem to see a proportionate drop off in "minority" enrollment in many graduate disciplines. I put minority in quotes because if you read through the paper there is the consistent semantic confusion which elides important dynamics at play. The author admits that Asians are not included in the analysis, because they are a varied group. More precisely: "I do not include Asian American/Pacific Islanders students in my definition of 'underrepresented' students of color because the category is too broadly defined to allow me to capture the educational disparities that exist within the various subgroups included in the category." This seems a dodge. The reality is that "Asians" are not an underrepresented minority, period. Rather, they are an overrepresented minority. If you want to make science reflect America, you better start reducing the number of Asian Americans who are taking the slots of underrepresented minorities! (international students are excluded from this analysis) An interesting finding is that the largest effects are in engineering, then the natural sciences, social sciences, and finally humanities. There is no impact on graduate education in business. In other words, banning affirmative action has no impact on the proportion of underrepresented minorities in business graduate school. Why the pattern? The author is straightforward on this point:
Given that students of color are generally underrepresented at higher score percentiles on standardized tests and generally overrepresented at lower percentiles
(see e.g., Bowen & Bok, 1998; Diaz, 1990), it is not surprising that the impact of affirmative action bans is greater in science-related fields like engineering (26%), natural sciences (17%), and social sciences (15.2%), compared to the humanities (11.8%) or education (where the impact is not statistically significant). This is because the overall mean score for standardized tests like the GRE, particularly in the quantitative portion of the test, is generally higher in these fields than in the humanities and education fields. Moreover, as I noted in the hypothesis section, the general characteristics of graduate students in less technically focused fields of education (and similarly humanities) suggest that past work experience may be an important factor that plays into students’ decisions to pursue study in these fields or in faculty’s admissions decisions. These other factors may thus balance the negative impact of affirmative action bans in these fields.
Notice the term "students of color." In this context it's obvious that she's talking about underrepresented minorities, but people might be confused and think engineering is "lily white." It isn't. Additionally, there is the standard convention whereby the Latino/Hispanic ethnicity is contrasted with the white race, when about half of American Latinos identify as white, and a substantial minority have far less "color" than the Asian students excluded from this analysis (to the point of being omitted from data tables!). In any case the overarching variable is the quantitative GRE. What would happen if affirmative action was enacted, and the quantitative GRE standard was relaxed for underrepresented minorities? Consider the Marcus Cole effect:
Let me illustrate my point. I am willing to bet that I am the only member of this list who feels compelled to put his standardized test scores and National Merit award on his CV. Why do I do this? For those of you who do not know me personally, it is not a matter of braggadocio. Every September I have to deal with nearly 60 prima donna first year law students whose first and only (initial) reaction to my skin color is that they have been cheated out of a "real" Contracts professor, and are stuck with an "Affirmative Action" instructor. Many of them come around when, as some "gunners" often do, they look up my CV and find that I have outscored virtually every single one of them on the test around which they have centered their lives, the LSAT. Others usually come around by mid semester when they have had an opportunity to compare my teaching to that of their other instructors. If numbers (standardized test scores and teaching evaluations) could obscure my skin color, my life would be heavenly.
Academic science is a test case in vulgar Social Darwinism. Only a minority of graduate students end up with the academic job that most initially desire. There's a lot of luck and social intelligence which goes into this, but smarts matter. In many of the more mathematical sciences there are often introductory core sequences, and one reason to set the bar for the quantitative GRE high is that graduate schools are not excited at the prospect of students flunking out due to low aptitude or preparation. Do you want your civil engineer to have received a passing grade for effort? I doubt it. Laced throughout the paper is the assumption that America needs a diverse scientific work force which includes blacks and Latinos to compete with the rest of the world. I don't think that that's true. Are the Chinese diversifying their scientific establishment in such a manner? But even granting a value to diversity, I think one can make the argument that that value varies by discipline. In an area like literature cultural priors matter. Discussion and exchange of perspectives are the bread and butter of many humanities programs. In an area like mechanical engineering cultural priors don't matter. Most people don't care about the ethnicity or sex of the engineers who help maintain and produce the superstructure of modern civilization. They want their engineers to be competent. In fact, if your engineer has a "different perspective" because of their cultural background they're likely to be a crank. Academic treatments like this presume that differences in ethnic proportions across disciplines can be changed by fiat. It isn't so easy. I have reported before that cultural preferences and expectations can influence the disciplinary path one takes. When you add the extra layer of lesser academic aptitude or preparation the goal of proportionality seems farcical to me. Some of the points made the paper are well taken. Ethnicity and cultural background do matter in the social ecology of science. I have noticed a trend of labs tending to be somewhat skewed in ethnicity, likely because of the nature of social networks in employment opportunities. But simply increasing the number of first year engineering graduate students won't change that a bit, because they'll be filtered out by the time they're up for a tenure track position.