EXTRA!! Michigan Sky Did Not Fall!

Yesterday, in the post immediately below, I took the editors of the Michigan Daily to task for their rejection of the principle of colorblind equality, for their implication that minorities are afraid to compete on a level playing field, and for encouraging the UM administration to violate the law.

That editorial was presumably a response to recently released information revealing that the sky in Michigan did not fall as a result of the passage of the Michigan Civil Rights Initiative.

The preliminary admission number for the freshman class represents a decrease of approximately 9 percent from the same time last year and reflects University efforts to manage the size of the 2008 entering class….

At U-M, the number of applications from underrepresented minorities fell 2.0 percent to 2,771 from last year. Of those who applied, 1,310 were admitted, an admission rate 2.3 percent lower than last year. Among underrepresented minorities, the percent of those admitted who have paid the enrollment deposit remains relatively constant…. A year ago, underrepresented minorities accounted for 10.85 percent of the projected freshmen class. This year the number is relatively unchanged at 10.47 percent.

Discussing this same information, Peter Schmidt reports on the Chronicle of Higher Education News Blog that these preliminary figures reveal that the University of Michigan

has a avoided a substantial drop in black, Hispanic, and Native American enrollment despite being barred from considering race and ethnicity in admissions.

Come on, Peter! Since Michigan’s data also reveal that the projected size of next year’s entire “freshman class represents a decrease of approximately 9 percent from the same time last year,” to write that the 2% decline in minority enrollments avoided “a substantial drop,” while true, would seem to be a considerable understatement. That phrasing emphasizes the fact of the decline rather than its small size.

Finally (this must be my give Peter Schmidt a hard time day), in a similar vein I have a problem with the way Schmidt reported, without comment, the acceptance rate figures provided by Michigan. He writes:

The university’s figures show that its acceptance rate for black, Hispanic, and Native American students dropped only slightly, by 2.3 percent, over last year. The university admitted 47.3 percent of applicants from those three minority groups and 42 percent of all applicants.

The group of “all applicants,” of course, includes “black, Hispanic, and Native American students.” The relevant comparison is thus not between “underrepresented minorities” and all applicants; it is between “underrepresented minorities” and whites, Asians, and other minorities who are not “underrepresented.”

I’m sure the reason Schmidt did not report the more relevant comparison is that Michigan did not release the acceptance rates of white, Asian, and other applicants, but if that is the case its failure to do so deserves to be mentioned, and questioned. Is Michigan afraid of what a relevant comparison of racial and ethnic rates would reveal?

ADDENDUM

A numerate reader emails that the UM press release contains enough information to tease out the non-minority acceptance rate. All it takes is the application of a little lower mathematics, i.e., arithmetic:

The original release states the number of total applicants and the number of minority applicants. Subtract: 29794-2771=27023. We also know total accepted and total minorities accepted. Subtract: 12533-1310=11223. Divide: 11223/27023=41.5%.

So, the acceptance rate for preferred minorities was 47.3% and for whites, Asians, and other non-preferred groups it was 41.5%. Perhaps there is an explanation for this difference in acceptance rates that does not involve awarding under the table racial preferences. Perhaps UM’s minority recruiting was so successful it attracted a group of minority applicants whose credentials were stronger than the other, non-minority applicants. If so, Michigan officials should be proud and quick to explain.

My numerate correspondent, however, doubts that the explanation is innocent. Even equal rates of acceptance, he writes, would probably be indicative of racial preferences at work.

What they do not give in the press release is average test scores and grades for different groups. Just knowing the SAT, one can infer that unless the group of minority students applying is at the higher end of the group’s IQ, relative to the placement of the non-favored, who must represent a much wider and/or much lower performing sample of their ‘group.’

Based on this analysis, I’m betting that the average Verbal and Math SAT scores among the minorities is 100 pts (1 standard deviation) lower on each exam relative to White applicants, with a similar ratio on the ACT. (more accurate for those identifying as “Black.” “Hispanic” etc is more like 75 pts down relative to White). These numbers are based on data from the college board.

Lets assume that the white students accepted into M go Blue are at least one standard deviation above the white mean. That is about 630 on both V and M. That means that only ~16% of the Black applicants are likely to reach that level on the SAT (Again, starting from College Board numbers). The situation gets worse quickly as the actual cutoff moves up.

I wrote back to the Numerate One:

You probably recall my posts on the Center for Equal Opportunity’s most recent studies of the numbers at Michigan, using 2005 data (before, obviously, MCRI passed and was implemented. In case not:

http://www.discriminations.us/2006/10/university_of_michigan_scoffla.html

In the most recent year (2005), the median black admittee’s SAT score was 1160, versus 1260 for Hispanics, 1350 for whites, and 1400 for Asians. High school GPAs were 3.4 for the median black, 3.6 for Hispanics, 3.8 for Asians, and 3.9 for whites.

Even if all MCRI did was reduce the SAT gap by 100 points, at least that was something.

Say What? (6)

  1. Chetly Zarko June 19, 2008 at 12:46 am | | Reply

    But is that statistically significant, John? Sure, its a difference, but as we all argue on the other side when talking about disparate impact, small differences may or may not be statistically significant, so we must give U-M some benefit of the doubt here.

    I suspect it is statistically significant, but not so statistically significant that U-M couldn’t defend itself in court. But there have to be better ways to slice the data.

  2. John Rosenberg June 19, 2008 at 6:51 am | | Reply

    But is that statistically significant…?

    Maybe. Maybe not. Easy to check, though, or it would be if UM would release SAT/ACT scores and high school grades by race and ethnicity, instead of cherry-picking what race/ethnic data they release.

  3. John Rosenberg June 19, 2008 at 6:52 am | | Reply

    But is that statistically significant…?

    Maybe. Maybe not. Easy to check, though, or it would be if UM would release SAT/ACT scores and high school grades by race and ethnicity, instead of cherry-picking what race/ethnic data it releases.

  4. David June 19, 2008 at 10:47 am | | Reply

    As Carl Cohen reported in his 1996(!) Commentary article, a more insightful set of data is the GPA/SAT/acceptance rate grid, which shows how acceptances vary by race for the SAME qualifications. It’s the best way to compare the level of racial favoritism, but it apparently requires FOIA requests to get the data. As Ward Connerly has asserted, the non-academic qualities supposedly prized by admissions offices that are used to justify racial preference are not likely to be concentrated in one race over another. The grid shows that these non-academic qualities are indeed just a fig leaf for racial preference.

  5. John Rosenberg June 19, 2008 at 11:07 am | | Reply

    I believe UM was forced to abandon use of that grid as a result of Jennifer Gratz’s successful lawsuit.

  6. willowglen June 20, 2008 at 12:17 pm | | Reply

    Of course the sky did not fall. This is so because if I were a betting man I would absolutely bet on the fact that UM is still in the business of extending admission based on racial preference – and it likely sure as heck is a lot more than 100 SAT points. I have got to hand to the apparatchiks at Michigan – the right pigment count matters a heck of a lot to them – and they have accomplished a lot with the twin towers of geographic based analysis (one might pause to think that ongoing segregation is in their interest, hardly a position they would otherwise openly endorse) and holistic – or whole person, admissions. And the problem is – UM can make it fairly expensive to litigate – because it takes more to prove that they aer using zipcodes as a proxy for race, and finding positive “holistic” factors at rate far more frequent with minority students. How clever of them?

Say What?