Editor's Note: Portions of this submission have been previously published on Professor Tim Groseclose's blog, specifically his analyses of tables in the Mare report.
By Tim Groseclose
In an Oct. 30, 2012 Daily Bruin column, a group of 57 professors criticized a Daily Bruin news article and column, which documented evidence that UCLA is using race in admissions, a violation of Proposition 209.
The 57 faculty also criticized a report by law professor Richard Sander, who described statistical analyses showing that UCLA is using race in admissions.
The 57 professors cite a report by UCLA sociologist Robert Mare. They write that “(Mare’s) report found no signs of race-based reader bias in the awarding of applicant holistic scores.”
The professors either did not read the Mare report carefully, or they are intentionally trying to misrepresent its findings.
Mare analyzed two major parts of the admissions process: the scores that each applicant receives from two initial reviewers in the first round of the process, and the scores that some applicants receive in “second chance” (Mare’s term) reviews by senior admissions staff. The latter reviews include “Final Review,” “Supplemental Review,” and “School Review.”
Sander and Mare found little, if any, evidence of racial bias in the initial reviews. However, both researchers found evidence of bias in the second-chance reviews. Sander was not given data about particular aspects of the second-chance reviews; he could therefore only conclude that the total effect of all aspects of the second-chance reviews contained racial bias. Meanwhile, Mare analyzed each aspect of the second-chance reviews separately.
With the latter analysis at least three of his statistical estimates imply racial bias.
The first of these estimates is the .391 number in column F of his Table 10. Because it is positive and statistically significant, this means the following: Suppose you take a black and a white student who are identical on every other variable in Professor Mare’s data set. That is, they have identical grades and SAT scores, have parents with identical incomes and educational backgrounds, etc. They’re also identical on the “Limits to Achievement” variable that Professor Mare created.
To construct this, Mare recorded such things as whether the applicants’ life experience includes homelessness, whether their life experience includes incarceration, whether their life experience includes being a victim of discrimination, and so on.
The .391 number means that the black student has a significantly higher probability of being selected for “Supplemental Review.” Remember the two students are identical on everything but race. Thus, it indicates a violation of Prop. 209.
The second of Mare’s estimates that implies racial bias is the -.706 number in column G of his Table 10. It indicates the following: suppose you take two students who have been selected for supplemental review.
Suppose one is black and one is white, but otherwise they are identical on all the variables that Professor Mare included in his analysis. The fact that the number is negative (and highly significant statistically) means that the black student is more likely to receive a lower holistic score than the white student. Lower scores are better, which means that the black student is more likely to be admitted. Once again, that’s a violation of Prop. 209.
A third estimate by Mare that implies a racial bias is the -.865 number in column D of his Table 10. This number indicates that black students receive significant racial preferences in the “Final Review” stage of the admissions process. (The latter occurs when the scores of two initial readers differ by more than 1.0. When this happens, a senior staff member conducts a third holistic review of the applicant. The applicant’s final holistic score is determined by that senior staff member.)
Mare’s Table 10 contains eight columns. Five do not show any statistically significant evidence that UCLA is giving racial preferences toward African Americans; however, three do. It is thus false to conclude that Mare found “no signs of race-based reader bias.”
Further, when he calculates the net effect of the entire admissions process (that is, all eight aspects, represented by the eight columns of Table 10), Mare finds that the net effect is substantial. Specifically, on page 74, he writes: “Absent the adjusted disparities estimated in this analysis 121 fewer Black applicants would have been admitted, which amounts to approximately 33 percent of the actual number admitted.”
The 57 professors also claim the following about Mare’s report: “An extensive, independent analysis of UCLA’s holistic review process concluded that it works as intended by our faculty.”
Here, however, is what Mare actually wrote: “The holistic ranking process for freshman admissions at UCLA appears to work much as intended.” Note that the 57 professors omitted the word “much.”
Again, they either did not read the report carefully, or they are intentionally trying to misrepresent its findings.
Groseclose is a professor of political science.