« Satisfied and Unsatisfied Clients | Main | "Extreme" and "Non-Extreme" Awards: More on Punies and Comps »

19 September 2006


Steve Sailer

To evaluate changes in test passage rates when tests get harder or easier over time, as in this case wher they got harder, you should use the bell curve method, as explained by La Griffe du Lion (and recently endorsed by Charles Murray in the WSJ):



OK, ABA schools then... fun with google:

Whittier had a 45% bar passage rate in 1998, although they are doing better now, Pepperdine had 65% that year, Stanford averages 90%

Whittier is half minority students, Stanford, one third. Minority enrollment is down across the board this year.

Rick Sander

In response to Corey, let me reiterate that the California bar data is for ABA schools only. I'm not including unaccredited schools in the analysis. And I can vouch from personal knowledge that the UCLA patterns closely track the general California-ABA patterns.


"These are seat-of-the-pants estimates, based on limited available data."

And yet, you are willing to make them and feel that you are shielded from culpability for their likely impact because they involve real data?

California bar passage is notoriously variable by school attended, which could be anything from Stanford to unaccredited inner-city night schools. You are using this data as if minority candidates are evenly distributed across that range of schools. That assumption flies in the face of logic and demonstrable fact.

If minority candidates are more prevalent in the schools with low bar passage rates, then minority candidates are going to have lower bar passage rates.

You MUST find a way to limit your data set to similarily situated white and black candidates or else stop drawing inferences that will lead policy-makers to further cluster black students into underperforming schools.

Gary Rosin

Kelin & Bolus have updated their December 2004 report on the July 2004 Texas Bar exam (link at the bottom of the comment). The new report (June 2006, link at the bottom of the comment) looks at the eventual BPRs of the cohort of first-takers in July 2004. They reported first-time and eventual passage rates by ethnic subgroups as:

Group, First, Eventual
White, 84%, 94%
Asian, 75%, 92%
Hispanic, 68%, 89%
Black, 51%, 77%

As found in the LSAC Bar Passage Study, Black Bar candidates had a lower "persistance" rate: 92%, as against 97% overall. A persistance rate is the percentage of failing candidates who retake the Bar after failing it. The 2006 Report looks at only persistance after failing the Bar on the first attempt.

Unlike the 2004 report, the 2006 report does not discuss average LSAT, UGPA, or adjusted LGPA for the subgroups, or otherwise delve any deeper into the data.

In the 2004 report, changes in the average LSAT of the ethnic subgroups accounted for 84.6% of the variance in their average total scores (p=0.027). A one-point increase in a group's average LSAT increased (i) its expected total score by almost 5 points (p=0.027), and (ii) its expected BPR (odds of a candidate passing) by just over 20% (residual Chi-Sq = 3.488 [p=0.322])(here, p>0.05 is good).

Note that these statistics are based on the results for all takers (first-takers and repeaters). Also, they are not in the 2004 Report; I calculated them using data drawn from it.

2004 report: http://www.ble.state.tx.us/one/analysis_0704tbe.htm
2006 update: http://www.ble.state.tx.us/announcements/klein%20report%200606.doc

Gary Rosin
South Texas College of Law

The comments to this entry are closed.


June 2024

Sun Mon Tue Wed Thu Fri Sat
2 3 4 5 6 7 8
9 10 11 12 13 14 15
16 17 18 19 20 21 22
23 24 25 26 27 28 29

Site Meter

Creative Commons License

  • Creative Commons License
Blog powered by Typepad