Are Black/White Disparities in Graduation and the Bar Getting Better, or Worse? - Sander 2

Previously: Introduction - Sander 1

Before delving into the debate on the mismatch effect, it is worth considering a prior question – have the terrible racial disparities documented by the Bar Passage Study improved since the mid-1990s?

The dismal statistics documented by the BPS and explored in Systemic Analysis are fairly familiar now: only 45% of blacks starting law school graduated and passed the bar on the first attempt; only 57% became lawyers after multiple bar attempts (compared to 79% and 83% for whites). 61% of blacks actually taking the bar in the BPS passed on their first attempt, compared to 91% of whites.

These are by far the best hard numbers we have on black outcomes – the only ones based on a large national sample of bar results – but they have a couple of weaknesses. They are ten years old now; and they are tainted to a degree by some creaming of the BPS sample. The overall pass rate of all participants in the BPS was 88%; the actual national average bar passage rate at the time was 82.3%.

There have been some encouraging developments for blacks since the BPS. As I noted in Systemic Analysis, the credentials gap between white and black students narrowed a little between the time of the BPS study and recent years. Fewer blacks thus entered law school with LSATs under 140.

On the other hand, following an argument laid out by Jeff Stake, it may be that law schools have widened their intra-school gaps between the races by focusing more on numbers (especially the LSAT) in admissions (Stake 2006, pp. 309-313). If intra-school race gaps expand and professors teach to the median student, then the "mismatch effect" I hypothesize would be worsened.

So what has actually happened? In the BPS sample, the black graduation rate was about 80.7% (19.3% of blacks did not graduate). The only source I know of for national figures on graduation rates is the ABA. The ABA maintains enrollment and graduation statistics on its website. The ABA data does not track individuals to determine graduation rates; it simply reports enrollment by race in each year of law school, and the number of J.D.s awarded by race. 

If we assume that black students entering in Year 0 will mostly graduate in Year 3, but some will graduate in Year 4 and a few in Year 5, we can estimate graduation rates for particular cohorts. Using that method, I estimated that 80% of the black students entering law school in 1991 earned J.D. degrees – a figure very close to the BPS number. 

Using similar methods for more recent cohorts, I estimate current black graduation rates are 78-79%.  This slight drop – which is paralleled by a slight drop in white graduation rates – may simply reflect the addition of several new schools and a ten percent overall increase in enrollment over the past ten years; the new schools probably tend to be low-prestige and have higher flunk-out rates.

Determining what has happened to black bar passage rates is much harder; the states generally don’t even keep bar passage data by race, much less disclose it. We do know that the national first-time bar passage rate was 75% in 2004. We also have very good bar passage data by race for California, and more fragmentary data from Texas and Colorado. What do these tell us?

Pass rate for first-time bar takers

	California, 2003-05 (ABA schools only)	Texas, 2004	Colorado, 1990-2004
All races	67.3%	81%	81.5%
Blacks	39.9%	53%	52.8%

In all three of these states, the first-time bar passage rate for blacks is about 28 points below the overall rate. Simple interpolation would suggest that if the national first-time bar passage rate is 75%, the national black first-time bar passage rate is close to 47% – unless these three states are somehow unrepresentative of national patterns.

(Recall, BTW, that in Systemic Analysis I argue that about half the black-white gap in bar passage is due to preferences and concomitant mismatch effects. Many studies show that, controlling for credentials and law school grades, blacks do not underperform on the bar.)

If the black first-time bar passage rate is only 47%, that is a large decline from the 61% rate observed in the BPS study. (But recall that the BPS figures were probably somewhat higher than the true 1994 numbers, and partly reflect a real decline over the past decade) Again, a substantial decline in black bar passage rates over the past decade is consistent with the available data.  In Colorado, for example, the overall bar passage rate fell from 76% in 1990-94 to 70% in 2000-04; for blacks, it fell from 42.8% to 34.5% – a larger absolute and relative decline.

Let’s assemble the data so far.  If 78% of black law students in the Class of 2004 successfully graduated from law school, 93% of black graduates took the bar (estimated from BPS data), and only 47% of those who do take the bar pass on the first attempt, then only 34% of the overall cohort graduated and passed the bar on the first attempt. The corresponding numbers for whites, using the same estimation procedures, are 90% graduation, 94% take the bar, and 78% pass on the first attempt – that is, 66% of whites starting law school graduate and pass the bar on their first attempt.

Using the BPS, we can estimate how a particular success rate for a cohort of first-time bar takers translates into ultimate bar passage. Again, the method is crude, but the point of this exercise is to gain some initial insight into where things stand now. A first-time bar pass rate of 47% (my guess of the 2004 black rate) translates into an ultimate pass rate of around 65%. (That is, cohorts in the BPS that had first-time bar passage rates in the 45-50% had ultimate bar passage rates about 18 points higher.) A first-time bar pass rate of 78% (my guess of the 2004 white rate) translates, by the same method, into an ultimate pass rate of around 90%.  This yields the following totals:

	BPS (1994 era)		2004 Era (my estimates)
	Whites	Blacks	Whites	Blacks
% of entering law students who graduate	92%	81%	90%	78%
% of graduates who take the bar	94%	93%	94%	93%
% of bar takers who pass on first attempt	91%	61%	78%	47%
% of bar takers who ultimately pass	96.5%	78%	90%	65%
% of entering law students who graduate and pass bar on first attempt	78.7%	45.1%	66%	34%
% of entering law students who ultimately become lawyers	82.7%	57.1%	76%	47%

Clearly, both whites and blacks are having worse outcomes today than in the BPS – mostly because of the decline in bar passage rates. But what about relative outcomes? In one sense, blacks are doing relatively better; the absolute declines for blacks and whites on first-time bar passage are similar, so the ratio of black-to-white failure rates has fallen from around four to around three.

On the other hand, relative chances of success for blacks have fallen much more sharply than for whites. The proportion of blacks graduating and passing the bar on the first attempt has fallen something like one-fourth ((45.1-34)/45.1) and the proportion of the black cohort becoming lawyers has fallen something more than one-sixth ((57.1-47)/57.1); the comparable declines for whites are one-sixth and one-twelfth.

These are seat-of-the-pants estimates, based on limited available data. But they leave no doubt that the mere passage of time has not cured the problem of racial disparities in legal education. The need to understand the causes of those disparities is more urgent than ever.

Comments

[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):

http://www.lagriffedulion.f2s.com/gap.htm

[Corey]

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.

[Corey]

"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