pageok
pageok
pageok
Some Data on Law School Applicants and Matriculants:

For applicants to law school for Fall 2007, the average "Caucasian/White" applicant to law school had an LSAT score of 154.9, and an undergraduate GPA of 3.32. The average "Black/African American" applicant had a LSAT score of 143.7 and an undergraduate GPA of 2.96. The standard deviation for LSAT scores is 9.97.

For law school matriculants in Fall 2007, the averages were 158/3.40 for whites, and 150/3.16 for African Americans.

Asian Americans had very similar scores to whites, on average, and Hispanics' scores were very close to midpoint between the scores of whites/Asians and African Americans.

What's the point? For this post, I don't have a specific point. I think it's obvious that one could use this data, if one were so inclined, to argue that law school affirmative action that insists on something approaching proportional representation in each law school for different ethnic groups is doomed to fail, and one could also use this data, if one were so inclined, to argue that law school affirmative action is absolutely necessary.

But the debates one hears and reads over these issues often seem to me to occur in a data vacuum, so I thought it was a contribution just to provide it.

neurodoc:
How predictive of law school grades are the LSATs? Obviously, the stronger the correlation between LSATS and grades, the more consequential, and hence problematic, are the differences in scores according to race/ethnicity. I expect they are pretty good predicters, perhaps better than most other exams other than the math portion of the SATs to predict performance in subsequent math courses, physics, or engineering.
2.5.2009 9:47am
TRE:
What does it mean the standard deviation for LSAT scores is 9.97? For the averages cited or for LSAT scores in general?
2.5.2009 9:48am
DavidBernstein (mail):
The standard deviation is for all test takers and all tests taken that year, so far as I can tell from the very limited information I have in front of me.
2.5.2009 9:49am
Jack Black (mail):
I would take this data and say with it: leave spared of "affirmative action" or "legacy admit" criticism any law school applicant whose LSAT score was greater than or equal to 160 and whose undergrad GPA was 3.30 (+/- .10).
2.5.2009 10:00am
arthur:
I don't understand the point of the applicants' statistics. It's a free country, so anyone can apply for anything. The wider difference bweteen the applicant and the matriculant pools means that relatively more unqualified blacks than whites apply and presumably are rejected, but I can't conceive any public policy conclusion from that data point. The matriculants' statistics may have some relevance.

By my reading, the difference between black and white applicants in LSAT scores is not statistically significant (less than two standard deviations), and the difference between black and white matriculants is smaller than the non-significant difference between the applicant groups.
2.5.2009 10:00am
Can't find a good name:
Jack Black: How did you choose the numbers 160/3.30? 160 is higher than the average LSAT for white matriculants, but 3.30 is lower than the average GPA for white applicants.
2.5.2009 10:03am
johnd:
The standard deviation is for each test. The averages have no standard deviation because they are not samples (i.e. they contain all data points, not a mere sample of them).
2.5.2009 10:04am
arthur:
Put another way, if someone used this data to argue in Court that affirmative action exists in law school admissions, Professsor Bernstein woudl favor Rule 11 sanctions on that attorney.
2.5.2009 10:04am
martinned (mail) (www):
Just eyeballing it, it looks like the null hypothesis of different LSAT scores for blacks and whites would have to be rejected at a 5% level of confidence. (t = about 1, means p = about 16%.)
2.5.2009 10:07am
martinned (mail) (www):
Why o why can't I delete my comments???
Just as I pressed post, I realised I should really retake undergrad statistics. Consider my previous comment deleted.
2.5.2009 10:07am
Old33 (mail):
*waiting for Charles Murray to show up and explain...*
2.5.2009 10:08am
Jack Black (mail):
Jack Black: How did you choose the numbers 160/3.30? 160 is higher than the average LSAT for white matriculants, but 3.30 is lower than the average GPA for white applicants.

160 is unassailable for anyone.

3.30 plus .10 is 3.40, the average for white applicants.
3.30 minus .10 is 3.20, which is .04 higher than the average for black applicants.

3.30, +/- .10 plus greater than or equal to 160 would seem unassailable from affirmative action or legacy admit criticism, given the placement of that subset in the broader pool and the fact that LSAT score is weighted more heavily than GPA in admissions calculations.

I have no idea why your reply fixates on skin color.
2.5.2009 10:10am
Frog Leg (mail):
Is this data available online?
2.5.2009 10:10am
Jack Black (mail):
Sorry, meant matriculants, not applicants.
2.5.2009 10:12am
ari8 (mail):
By my reading, the difference between black and white applicants in LSAT scores is not statistically significant (less than two standard deviations), and the difference between black and white matriculants is smaller than the non-significant difference between the applicant groups.
How to lie with statistics!
2.5.2009 10:12am
Matt12 (mail):
What this tells me is that we have way too many law schools in this country who are willing to accept mediocre or worse college students.
2.5.2009 10:20am
Gabriel McCall (mail):

Beyond FYA: Analysis of the Utility of LSAT Scores and UGPA for Predicting Academic Success in Law School.


The major finding is that LSAT score and UGPA in combination are related to cumulative LPA at approximately the same level as they are related to first-year LGPA. A second important finding is that patterns of predictive validity for different ethnic and sex groups do not seem to change regardless of whether the criterion is first-year or cumulative LGPA. However, there is an overall tendency for test scores and undergraduate grades to overpredict law school performance for nonwhite students. Data from the study demonstrate the utility of LSAT scores and UGPAs in the law school admission process beyond the prediction of first-year performance.



So, black applicants have lower scores than whites, and those lower scores overpredict their academic success. (Note that academic success is not synonymous with career success.)

Policy questions are not math questions though: the question of whether law schools should make an extra effort to admit and graduate black students even though they (on average) don't do as well academically, is not susceptible to numerical analysis.
2.5.2009 10:31am
anonymouscommenter:
Too often in these debates over racial diversity on law school admissions, the focus is entirely on racial disparities. Isn't it true that class/socioeconomic background has more to do with LSAT scores &admissions than race? Poor students, no matter their racial background, will struggle with the LSAT, while those of more affluent backgrounds thrive.
2.5.2009 10:42am
Ben P:
A 2.96, seriously?
2.5.2009 10:44am
keypusher64 (mail):
By my reading, the difference between black and white applicants in LSAT scores is not statistically significant (less than two standard deviations), and the difference between black and white matriculants is smaller than the non-significant difference between the applicant groups.

You think any difference less than two standard deviations is not statistically significant? One standard deviation in height for an American male is 2.8 inches. Two standard deviations is 5.6 inches. If Bob is 6'2" and Bill is 5'9", wouldn't you agree that Bob is significantly taller than Bill?

The smaller racial gap between 2007 matriculants than 2007 applicants has two obvious explanations:

(i) law schools were practicing affirmative action in admissions more aggressively for the 2007 class than the 2004 class;

(ii) there was a much higher attrition rate for black law students than white law students, and low LSAT/low GPA black students were disproportionately attrited.

Having read some Sander articles, I expect (ii) is much more likely, but is there any data to show the answer?

Put another way, if someone used this data to argue in Court that affirmative action exists in law school admissions, Professsor Bernstein woudl favor Rule 11 sanctions on that attorney.

Whether affirmative action exists in law school admissions isn't an open question, Arthur.
2.5.2009 10:52am
keypusher64 (mail):
Too often in these debates over racial diversity on law school admissions, the focus is entirely on racial disparities. Isn't it true that class/socioeconomic background has more to do with LSAT scores &admissions than race? Poor students, no matter their racial background, will struggle with the LSAT, while those of more affluent backgrounds thrive.

No, it isn't true. Poor whites score higher than middle-class blacks.
2.5.2009 10:53am
JDS:
Superb analysis of this type of data may be found at the
website of "La Griffe du Lion."
2.5.2009 10:56am
Houston Lawyer:
It would be helpful for someone to plot the LSAT distributions curves for white students and black students over each other. What the plot would show is very few blacks in the high end of the range where selective law schools get their white students.
2.5.2009 11:01am
Roscoe2 (mail):

for 2005-2008, % below for LSAT scaled scored of:


applicants:

155 = 64.2%

144 = 23.1%


matriculants:

158 = 75.1%

150 = 44.4%

Source: Item Response Report Additional Information Document for the Dec. 2008 LSAT
2.5.2009 11:01am
TEvanFisher (mail):
Ben P:

I know someone at a top 25 law school who was accepted with a 2.12 undergrad GPA. Statistical anomalies happen.
2.5.2009 11:03am
NRWO:
Neurodoc,

The predictive validities of cognitive tests (i.e., correlations between the tests and later outcomes), including the LSAT, for many outcomes, including GPA, are reported in a meta-analysis by Kuncel et al. (2007, Standardized tests predict graduate students' success, Science, Vol. 315, P. 1080-1081, see figure).

Extrapolating from the figure, the predictive validity of the LSAT is .53 for 1L GPA, .41 for 1L-3L combined GPA, and .32 for bar passing rate.

Kuncel et al. briefly discuss the usual objections to using cognitive tests for predictive purposes. The objections include the claim that the tests are measuring SES and are biased against minorities.

FYI: Both claims have been rejected in meta-analysis of the SAT (but, as far as I can tell, they have not been addressed for the LSAT): The SAT predicts college GPA even after SES is controlled (r = .47 and .44 for unadjusted and adjusted, respectively, see Sackett et al., 2008, American Psychologist, Vol. 63, P. 215). And the SAT also generally predicts GPA equally well for Blacks and Whites (and when it doesn't, it usually overpredicts Black's GPA).

My guess is that the same pattern holds for the LSAT, which -- like most cognitive tests that predict things we care about -- is probably highly g loaded (like the SAT), and which probably derives its predictive validity through its relation to g (like the SAT).
2.5.2009 11:12am
MarkField (mail):

How predictive of law school grades are the LSATs?


I believe the claim is that LSAT+GPA gives a correlation to first year grades of .4. I'm going off memory here, so someone can check me. This, btw, is a substantial improvement over the correlation when I applied to law school, when it was about .3.

It might be worth asking a couple of other important questions as well. First, GPA isn't very helpful without knowing the person's major. It's much harder to get a higher GPA in Engineering than English. Second, we might wonder why correlation to first year grades is all that important a goal. The purpose of a good law school is not to produce good law students, it's to produce good lawyers. The correlations we should be seeking are those which lead to that outcome. My understanding is that Berkeley Law/Boalt Hall* currently has a project designed to try to identify those correlations.


I know someone at a top 25 law school who was accepted with a 2.12 undergrad GPA. Statistical anomalies happen.


While I don't know of any "anamolies" quite that extreme, there were students in my class with GPA well below 3.0.

*The Law School Formerly Known As Boalt Hall now wishes to be called "Berkeley Law".
2.5.2009 11:14am
MarkField (mail):
And NWRO beat me to the punch with better numbers, it appears.
2.5.2009 11:16am
neurodoc:
By my reading, the difference between black and white applicants in LSAT scores is not statistically significant (less than two standard deviations), and the difference between black and white matriculants is smaller than the non-significant difference between the applicant groups.
I think this comment reflects a misunderstanding of what constitutes "statistical significance."

We are told that the SD for LSAT scores, whether it for any particular group taking the exam or for eveyone taking the exam, is 9.97. It does not follow that a difference in mean scores between two groups must be anywhere near 20 (2 x 9.97) for the result to be "statistically significant." We can be reasonably confident that the observed differences weren't just a chance thing that wouldn't hold up if we re-tested and compared the results of the groups. (To say exactly how certain we can be that it the observed differences are not a matter of mere chance, we would have to know the number of test-takers in each group and the SD for the respective groups.)

The statistics are the easy, straightforward part of this; it's the implications of the data and what, if anything, to do about them that is not easy. Other things need to be taken into account:

http://www.washingtonpost.com/wp-dyn/content/article/
2009/02/01/AR2009020102171.html
2.5.2009 11:20am
NRWO:
Above Gabriel McCall cites evidence that the LSAT is generally not biased against minorities -- and, when such biased is observed, it overpredicts later performance (LSGPA) of minorities. "Overpredicts" means that minorities actually do more poorly in school (have lower LSGPAs) than the test would predict, so that, in general, relying solely on test scores would benefit minorities.

Such evidence is consistent with my prediction about the LSAT and with the evidence I cite about the SAT.
2.5.2009 11:25am
Spartacus (www):
By my reading, the difference between black and white applicants in LSAT scores is not statistically significant (less than two standard deviations), and the difference between black and white matriculants is smaller than the non-significant difference between the applicant groups.

To put a finer point on the meaning of Standard Deviation (SD), which I am certain is known by many on this blog, 2/3 of any data sample lie within one SD of the median, while ~95% lie within two SDs of the median. So, to say anything less than two SDs is not statistically significant is like saying that you are statisticially commonplace unless you are such an outlier than you are in a class that over 90% of the population does not fall into. The SD for the LSAT is essentially 10 and I believe the scores are normed to a median of 150 (correct me if I am wrong on this point). So, a 160 means that you are scoring better than something like 75% of the population, a 170 is better than more than 90% of the population (where population means all LSAT takers from thelast two years). What the data show for matriculants is essentially that the average white matriculant was almost one SD above the mean, scored better than something like 70% of the test takers, while the average black matriculant was simply an average test taker.

What would be really interesting to know, as pointed out in a previous comment, would be the SDs and range for each group. It is probably safe to assume that the SD for matricuants is less than the SD for the overall population (test takers), since the sample is smaller and most of those 130s and 140s don't make it in. If the SD for matriculants was, say 8 or 9, that would mean that the average black matricualnt scores lower than around 70% of white applicants, and that about 70% of black matricualnts scored lower than the average white matriculant. But to make a more accurate statement, the real SDs of both black, white, and all matricualants would be needed. However, it is almost certainly less than 9.97.
2.5.2009 11:46am
AndrewK (mail):
That data suggests that a greater proportion of below-racial-median applicants drop out if they are African American than if they are white.

My main point is that LSAT and grades aren't the only two things to look at here. Look at drop-out rates.

And are grades the only predictor of law school success?
2.5.2009 11:48am
AnonyEng (mail):
Ben P: My 2.97 put me above the mean in my major (engineering, mean 2.7). I attended a top 10 law school on the strength of my LSAT, and did quite well.
2.5.2009 11:49am
Bama 1L:
If the LSAT overpredicts LSGPA of minority students, then that's at best a mixed blessing for the minority students. Merely getting into law school is not much of benefit, particularly if one ends up performing poorly.
2.5.2009 11:50am
Eric Muller (www):
DB says: "For this post, I don't have a specific point."

Oh, David, come on.
2.5.2009 11:58am
Ben P:

Ben P: My 2.97 put me above the mean in my major (engineering, mean 2.7). I attended a top 10 law school on the strength of my LSAT, and did quite well.


I have a couple responses, not all of them are related.

1. And how many of your colleagues were also hard science majors?

If it's anything like the law school classes I knew it's not a big percentage. Poli-Sci is probably a plurality, with other liberal arts degrees filling out a very solid majority.

My general sense from Undergrad was that it was much easier to get a B but harder to get an A in most of the liberal arts classes than in the hard science classes. But getting C's is bad.


2. I'm not discounting your experience, but I went to undergrad at a quote "liberal arts college" but what we were known most for is having an excellent "pre-med" program and graduating large numbers of students to medical schools. Among many students who had Chemistry and Biochemistry majors getting into med school still required a GPA on the sunny side of 3.5 to be comfortable.


3. Some of this may be grade shock that I never really experienced. I had a liberal arts degree with computer science minor in Undergrad and have had top 10% grades in law school.

I understand the hard classes, I had a couple classes in undergrad where few students rose above the C range. (in my case mostly econ classes) But my gut reaction to a GPA that starts with anything less than a 3 is "Bad grade."

4. Finally, I'm pretty sure we're not talking about outlier engineering students here. This is an Average GPA of an entire group of applicants. That means at least theoretically that half of the GPA's of applying students were lower than this number. I find that much harder to understand than an engineer who had a lower GPA but demonstrates it's the result of certain classes by a good LSAT.
2.5.2009 12:10pm
deliotb (mail):
Eric, you don't agree that the response of affirmative action advocates to data like this is "see, affirmative action in admission is absolutely necessary, because if we don't have it, the legal profession, especially at elite levels, will not be 'diverse?'" And do you not agree that this is hardly an irrational or frivolous argument? And if so, isn't it the case that merely presenting this data does not inherently support a pro or anti- affirmative action case? If I'm missing something here, let me know.
2.5.2009 12:11pm
Asher (mail):
143's rather low. That isn't even much better than guessing.
2.5.2009 12:20pm
AnonyEng (mail):
Ben P:

1. Not very many but at least 10% of the class (though many had respectable undergrad GPAs).

2. My undergrad was fairly typically in the 5-10 range for engineering schools. I believe (and could be wrong, but not by much) that a 3.3 put you in the top 10% of the class for my major.

3. Most people share your gut reaction, which is why getting into a few good law schools was a shock at the time, but I did outperform what my GPA or even GPA/LSAT combo would have predicted. I should note that coming out of undergrad, I had plenty of job offers from top engineering companies. I think they understood the harsh curve.

4. I'm sure applicants coming from majors/schools with low gpa averages are a small minority. I just get defensive when people assume low GPA students are all dolts.
2.5.2009 12:22pm
DeezRightWingNutz:
Matriculate means enroll, right? I get the sense that some people here are taking it to mean graduate. Is everyone here who is making comments about drop-out rates making their inferences from the O.P. (which, to my eyes, only deals with applicants to and enrollees at law schools).
2.5.2009 12:28pm
Eric Muller (www):
deliotb, you are missing only one thing, and that is the context of the rest of David's blogging on the subject of affirmative action and law school admissions.
2.5.2009 12:37pm
HoyaBlue:
Agree with what was said by Ben P.

I guess it's fairly obvious which top-25 undergraduate I hail from. And let me tell you, with grade inflation, if someone gets a 2.97, they're either not trying or not qualified for law school (and possibly not even qualified for undergraduate).

You'd have to try pretty hard to get below a B average even if you were to only study around 5 hours a week.

And while I'm not sure that Georgetown is representative of everywhere, anyone with a 2.97 has no business entering law school.
2.5.2009 12:38pm
PubliusFL:
Spartacus: What the data show for matriculants is essentially that the average white matriculant was almost one SD above the mean, scored better than something like 70% of the test takers, while the average black matriculant was simply an average test taker.

If the median is 150 and standard deviation is 9.97, and the average white matriculant gets a 158, a z-table shows that the average white matriculant has actually scored better than about 79% of the test takers.
2.5.2009 12:38pm
Spartacus (www):
Spartacus: If the median is 150 and standard deviation is 9.97, and the average white matriculant gets a 158, a z-table shows that the average white matriculant has actually scored better than about 79% of the test takers.

Thanks, I was using rough estimates off the top of my head, but the general idea of my post remains the same. But this does beg the question of whether these assumptions are correct; perhaps the mean is higher, more like 155? it seems hard to believe that the average white matriculant scores better than 79% of test takers, unless a lot of people take the test and don't get in, or unles there are a lot more minority test takers. More data would definitely be useful, but my main point was that if the 9.97 figure is correct, the white-black score differences are certainly statistaclly significant.
2.5.2009 12:44pm
Spartacus (www):
On second thought, perhaps these data are not too impalusible; the average test taker overall should score better than 50% of test takers, so counting for non-whites and non-matric, an average white matric score better than 79% of test takers is not too implausible. Nor is an average black test taker score below a similar proportion of test takers.
2.5.2009 12:47pm
DavidBernstein (mail):
Whoops, Eric, deliotb is me (David), I put my email name instead of my name. Actually, you miss the point--given what I've posted before, I obviously am not shy about my views, and if I intended to make a specific point with this, I would have.
2.5.2009 12:50pm
ASlyJD (mail):
These figures are quoted in the Kaplan LSAT study guide, using the Oct 1, 2005 test.

Perfect score -- 180
99th percentile -- 172-180
95th -- 167
90th -- 164
80th -- 159
70th -- 156
60th -- 154
50th -- 151
40th -- 149
30th -- 146
20th -- 143
10th -- 139
Lowest possible score -- 120

A cursory glance will note that the percentiles are not distributed evenly.

Furthermore, assuming that these percentile numbers are somewhat stable, the average white student is at the average, while the average black student is near the 20th percentile.
2.5.2009 12:50pm
ASlyJD (mail):
Correction: in 2004 that was the case. In 2007 whites were in the 80th and blacks in the 50th.
2.5.2009 12:51pm
pete (mail) (www):

I guess it's fairly obvious which top-25 undergraduate I hail from. And let me tell you, with grade inflation, if someone gets a 2.97, they're either not trying or not qualified for law school (and possibly not even qualified for undergraduate).


And at my school depending somewhat on your major, it was fairly easy to get a 2.97 and still be qualified for law school. I went to a #1 ranked regional liberal arts school where you had to be fairly smart and studying about 30 hours a week to be getting at least a 3.0. This was especially true for those in the hard science and math classes vs us liberal arts types.

I assume for these LSAT score that the students might be spread out enough to make grade inflation average out, but that is a pretty big assumption.
2.5.2009 12:56pm
loki13 (mail):


I guess it's fairly obvious which top-25 undergraduate I hail from. And let me tell you, with grade inflation, if someone gets a 2.97, they're either not trying or not qualified for law school (and possibly not even qualified for undergraduate).

You'd have to try pretty hard to get below a B average even if you were to only study around 5 hours a week.

And while I'm not sure that Georgetown is representative of everywhere, anyone with a 2.97 has no business entering law school.


Let me address this:
1. While grade inflation is rampant on many campuses, it is not rampant on all campuses. When I went to school (back in the halcyon days) the UGPA average was a hair over 3.0, which meant that performance relative to the average was important; luckkily my small liberal arts school was farily well-known with law schools.

2. While grade inflation does occur, it is not uniform in all majors. While other cite engineering, at my school it was economics. In that department, a C was given to good students (a D or an F was the indication that you should no longer pursue the major), a B to outstanding students, and an A was given with the approximate frequency of a book award- but less. So econ majors (or those who took significant econ. classes) had much lower UGPAs, usually in the mid 2s.

3. I did quite well in law school, despite "not [being] qualified for law school" due to my sub-2.97 gpa. It's called being a splitter- the over 170 LSAT is helpful.

4. You are correct. People with sub-2.97 GPAs should not attend GULC. No one should attend GULC. Georgetown Law is a horrible place filled with horrible people (and the scores of horrible people that transfer in regularly so GULC can keep their pretentious rankings and revenues).

(note- I am being sarcastic about point 4. somewhat.)
2.5.2009 12:59pm
ASlyJD (mail):
I would hold that the increase in test scores is simply a factor of the increasing competitiveness in law school admissions. While yes, under affirmative action policies the schools have strong pressure to increase the number of minority students, they have equal pressure keep the school's ranking high. Thus, even their "less qualified" students have to have better grades than they used to.

"Less qualified" is in quotes because as a former Kaplan test prep teacher, I can tell you that test scores are only somewhat predictive. Certainly, with a 164 LSAT, I didn't expect to earn only a 2.2 GPA. (Which isn't too bad at a law school with a class median of 2.7.)
2.5.2009 1:06pm
dll111:

TEvanFisher (mail):
Ben P:

I know someone at a top 25 law school who was accepted with a 2.12 undergrad GPA. Statistical anomalies happen.


His/her last name Bush or Kennedy, per chance?
2.5.2009 1:08pm
AnonyEng (mail):
HoyaBlue:

I'm sure you'll be shocked to know that not every school has the same mean and not every major within a school has the same mean (also not every school practices grade inflation).

While I'm sure I could have done better in undergrad, I don't think I could have done much better.

However, I did take that low low GPA and 99% LSAT and did very well in law school (I'm not sure if you attend/ed GULC but it was one ranked higher than GULC).
2.5.2009 1:18pm
Aultimer:

MarkField

The purpose of a good law school is not to produce good law students, it's to produce good lawyers.

That's certainly not complete, and may not even be correct. Law schools have many, quite interrelated, purposes including:
1. To produce money for other departments
2. To satisfy law firm demand for qualified help
3. To ensure sufficient lawyers working in a particular jurisdiction
4. To enhance the reputation of lesser departments
5. To provide prospective students with qualifications to obtain desired employment
6. To advance knowledge in the field
2.5.2009 1:25pm
moldbug (mail) (www):
2.5.2009 1:30pm
Paul McKaskle (mail):
Having served on law school admissions committees for decades, I have always been interested in the correlation between law school grades and the LSAT. My first observation is that LSAT useful mainly for determining the number of those admited who will successfully complete law school. Out of a group of 100 applicants with, say, a 165 LSAT, most of them will be successful in law school, a substantial number will do extremely well but a few will do very poorly. Out of a group of 100 with, say, a 150 LSAT, some will be successful, a very small number will do well and a goodly number will do very poorly (and if members of this latter group don't actually flunk out, they will have a poor success rate on the bar examination). A group of 100 with, say, a 145 LSAT will perform even more poorly, but a very small number (one or two percent) will be quite successful in law school. In general a law school uses the LSAT to maximize the number of students it enrolls who will be successful in graduating and becoming successful lawyers--the more students with high LSATs, the larger number of successful graduates.

Of course LSATs are not the only factor used (but I suspect they play a larger role in admissions than most law schools would be willing to admit). Undergraduate gpas can sometimes provide useful information and the combination of LSAT and gpa is slightly more predictive of success than LSAT alone. As some commenters have noted, the undergraduate field of study can make a difference. A number of years of successful employment after college may well make up for a bad ugpa. I know of a number of students who were not interested in their studies in college who were turned on by the study of law and, as a consequence, did quite well.

Sometimes cultural circumstances have some significance. For example my impression (not a scientific analysis) is that law students who emigrated from former Soviet bloc countries as teenagers tend to do better in law school than their LSATs would predict. Also, my impression is that LSAT scores for African-Americans are somewhat less predictive and other aspects of the African-American applicants background may have somewhat more predictive value as to their potential for success.

Finally, I think the odds of success for a law school applicant with a LSAT score in the 140s are very low. Over the years I have known two or three in this category who have not only done well in law school but have had outstanding legal careers. But I have also known far larger number who simply flunked out after the first year. Because of a desire to increase the number of African-Americans in the law profession, almost of those admitted to my law school with LSATs in the 140s were African-Americans and consequently many in this group have simply failed. While I would like to see a larger number of successful African-American lawyers, I also think that the African-American community is done a dis-service when lots of time and money is invested in an educational enterprise with a very poor success rate. (I am referring only to the admission of African-Americans with very low LSATs, most law schools have African-American admittees with higher LSATs and the success ratio for them is much better.) The African-American students with low LSATs who do flunk out have lost perhaps close to $50,000 in tuition and living expenses and, more critical, a year of time that could have been used to pursue other careers in which the odds of success would have been higher.
2.5.2009 1:46pm
CJColucci:
Some years ago, after much study, I determined that roughly 50% of every law school class ends up in the bottom half. This result proved extremely robust: it held true in schools with affirmative action programs and schools without; it held true regardless of the racial composition of the class; it even held true without regard to the racial composition of the bottom half of the class. (Many people I shared these results with seemed to care an awful lot about the racial composition of the bottom half of the class, but I'm not sure why.) I haven't updated the study, but I'd be surprised if anything has changed.
2.5.2009 2:15pm
Tim Torrent (mail) (www):
I've always felt that GPA was a pretty bogus metric cross university or cross undergrad (and LSAC doesn't help this) because some professors curve to 2.0, some to 2.5, others to 3.3, and some professors don't curve at all. I had one professor who failed 75% of a 400 person class, and 13% got Ds. As a core class, you had to get a C, so literally 300+ people had to retake a class at a major university. Another class was ~40 students and only a single student had not dropped the class by the drop deadline. Certain majors (like engineering or pre-med) are ridiculously harder than other majors (business or poli-sci), and many law schools don't weight that at all (I personally talked to admissions deans about it).

And then you get to law school, where professors apply their personal curves which may be higher or lower. Some professors are ball-busting socratic methodists, while others don't care as long as students get the black letter law. Certain law schools curve to 2.5 while others curve to 3.3. Also, lower tier private law schools tend to have lower curves so they can keep students from transferring out (they're still a business).

And then you get out and take the bar, and if you're lucky, someone at a big firm sees you. My stats in undergrad were mediocre for engineering (2.7ish), and my law school stats were also mediocre (3.0, and not in top 40%). I went to a bottom end second tier law school, but then my first job offer's starting salary stomped this national median of $60k -- and I didn't get it because my family or friends of the family. I got it because a partner saw me arguing a motion in court (some states allow law students to get a limited license to practice public interest law pre-bar).

So these numerics may have a correlation to success, but certainly not a strong one. My numerics weren't great, but I'm doing very well now.
2.5.2009 2:21pm
Asher (mail):

And while I'm not sure that Georgetown is representative of everywhere, anyone with a 2.97 has no business entering law school.


If you would, go to lawschoolnumbers.com, particularly noting this, this, and this chart. (Also see their admissions charts on WUSTL, UMinn, UVA.)
2.5.2009 2:59pm
hawkins:

You'd have to try pretty hard to get below a B average even if you were to only study around 5 hours a week.


That's an innumerable number of hours as an undergraduate that are gone forever.


And while I'm not sure that Georgetown is representative of everywhere, anyone with a 2.97 has no business entering law school.


I know many people that excelled in law school after barely making it through undergrad.
2.5.2009 3:15pm
gasman (mail):

By my reading, the difference between black and white applicants in LSAT scores is not statistically significant (less than two standard deviations), and the difference between black and white matriculants is smaller than the non-significant difference between the applicant groups.

SD describes the variation about the mean within a group.
You need to know the number of individual in each group so that you can determine the Standard Error of the Means, by which you can then infer whether the means differ by chance or for some other reason.

Supposing that there were hundreds (or perhaps thousands) of individuals in each group, then the differences are very unlikely due to chance (i.e. p much less than the magical threshold of 0.05). Now, whether a statistical significance is clinically relevant is still up for debate; do those 10 or so points matter in the final product of a matriculated then graduated lawyer.
Statistical illiteracy is rampant.
2.5.2009 3:16pm
MarkField (mail):

That's certainly not complete, and may not even be correct. Law schools have many, quite interrelated, purposes including:
1. To produce money for other departments
2. To satisfy law firm demand for qualified help
3. To ensure sufficient lawyers working in a particular jurisdiction
4. To enhance the reputation of lesser departments
5. To provide prospective students with qualifications to obtain desired employment
6. To advance knowledge in the field


I see 2-6 as derivative of the goal of producing good lawyers. Unless you mean by 6 that law students might advance knowledge in the field. While that's legitimate, I don't think it's common and I doubt that the CA Legislature, say, takes that into account in funding the law schools at UC. I'm not sure 1 is true, though I guess it might be; again, I doubt it plays much of a role in legislative funding decisions.
2.5.2009 3:52pm
H2:

I've always felt that GPA was a pretty bogus metric cross university or cross undergrad (and LSAC doesn't help this) because some professors curve to 2.0, some to 2.5, others to 3.3, and some professors don't curve at all. I had one professor who failed 75% of a 400 person class, and 13% got Ds.


If I am not mistaken, LSAC does help this. I distinctly remember the law school applicant service pulling a report from each person's college. I applied for 2002 and 2003 (first time I did everything wrong, applied on last day, quickly drafted my personal statement, etc.)
I saw this report, included in my file, and it listed my college, my major and the grade breakdown within that major.

This report is supposed to standardize the grade disparities of different colleges. I thought it was an awesome idea. The law schools would now know if a B average was actually the median (usually C average) or in the typically more rigorous engineering schools if a B average was 90%.

I was a little bit above average and I remember that making me happy because I worked full time and most people did not.
2.5.2009 3:56pm
H2:
Oh, one more comment.

I went to an elite LSAT prep class (not the usual Kaplan or Princeton Review) in Westwood, CA.
It was a great prep class because the instructors/owners had done an awesome job teaching the LSAT and also researching what admissions looked at for admissions.

One night in class, the instructor was telling us what LSAT score we would have to aim for in order to just get in, go to an elite, get a scholarship, etc.
This was a sobering talk for most students. After this talk, he casually added, "Oh, for you African American applicants, take # (6-10?) points of of all those scores."
--I am sure the # was closer to 10 because the reaction.

Well there was an audible gasp from the class. I was sitting in the back and remember people looking around and I saw one African American classmate, just smiling, while everyone else looked jealous.

Instructor then said, "And for you hispanics take 1/2# off of those scores."

So, chicken or the egg?...at least for the top African American scorers, who could be "pulling" the AA LSAT average down.
2.5.2009 4:09pm
Hoosier:
TEvanFisher
Ben P:

I know someone at a top 25 law school who was accepted with a 2.12 undergrad GPA. Statistical anomalies happen.


Percival Vanderbilt von Vanderbilt IV? Accepted at Vandy?
2.5.2009 4:43pm
Hoosier:
H2

You're right. I know that students from my almaq mater got a pretty favorab;e "bouce" when apllying to law schools. LSAC "corrects" for differences among institutions. I have to call this suspicious, though. They seem to assign a specific number value for comparison. But what possible data could they collect to make that degree of specificity meaningful?
2.5.2009 4:47pm
Asher (mail):
See my post above. This happens all the time. At Prof. Kerr's school, for example, 75% of applicants with 2.7 to 3.0 GPAs and an LSAT above 173 get in.
2.5.2009 4:54pm
neurodoc:
HoyaBlue, I don't imagine you were a GULC student circa 1991 when a student (Tim McGraw, or something like that, IIRC) working a part-time job in the admissions office undertook to do his own look at the files of minor students to see for himself if they were cut slack, how they performed after admitted, etc. When he shared his "findings," the place exploded and he was lucky to escape alive, though not with a clean file of his own. Dean Ahren was sorely tried by that episode and proved herself as a leader.

AnonyEng, you don't say which law school you attended. Perhaps as you claim it was one ranked more highly than Georgetown, though there weren't all that many ranked higher. But I doubt very much that your school's b-ball team won the NCAA tournament and sent anyone like Patrick Ewing to the NBA. (His wife was in my class and he was there at our graduation, standing out as literally as can be, though quietly to be sure.)
2.5.2009 5:15pm
Greg Q (mail) (www):
Policy questions are not math questions though: the question of whether law schools should make an extra effort to admit and graduate black students even though they (on average) don't do as well academically, is not susceptible to numerical analysis.

That one, however, is easy. If you're a racist, you can favor such a policy. If you don't judge people based upon skin color, then clearly it's a bad idea.
2.5.2009 6:24pm
AUH20:
Something that hasn't been mentioned yet that may or may not be significant...

In the world of LSDAS, a GPA isn't always what it seems. The GPA reported to schools includes all undergraduate work done by the applicant, regardless of when that work was completed. My applications reflected four years of college (no degree) with a GPA of <2, and four more when I graduated with a 4.0 from another school after six years in the workforce.

According to LSDAS, I'm a 2.8 GPA, but I know better. It did, however, take two admissions cycles, multiple rejections/waitlists/deferrals, and a LSAT in the 99th %ile to finally get where I wanted to be (top ten).

IMO, LSDAS would do well to report only grades from the degree granting university, or at the very least, institute something like a ten year cap. If the goal is accurate predictions of future performance, skewing an applicants figures with ancient history is a mistake.
2.5.2009 8:34pm
Lies, Damned Lies...:

not statistically significant (less than two standard deviations)

lol
2.6.2009 1:17am
BABH:
Shall I be very explicit for the doubting Thomases who show little understanding either of statistics or of life's endless variety?

In 1998, I graduated Yale with a GPA of 2. That's a round 2.0, mind you, well into the age of grade inflation. In 2003, I matriculated at Harvard with an LSAT of 178. I now take home over $250,000 a year. I have no family connections to speak of.
2.6.2009 11:23am
neurodoc:
: In 1998, I graduated Yale with a GPA of 2. That's a round 2.0, mind you, well into the age of grade inflation.
A few years after them, but neck and neck with our last prez and the one who competed against him for the job.

$250K as a third-year associate?
2.6.2009 6:26pm

Post as: [Register] [Log In]

Account:
Password:
Remember info?

If you have a comment about spelling, typos, or format errors, please e-mail the poster directly rather than posting a comment.

Comment Policy: We reserve the right to edit or delete comments, and in extreme cases to ban commenters, at our discretion. Comments must be relevant and civil (and, especially, free of name-calling). We think of comment threads like dinner parties at our homes. If you make the party unpleasant for us or for others, we'd rather you went elsewhere. We're happy to see a wide range of viewpoints, but we want all of them to be expressed as politely as possible.

We realize that such a comment policy can never be evenly enforced, because we can't possibly monitor every comment equally well. Hundreds of comments are posted every day here, and we don't read them all. Those we read, we read with different degrees of attention, and in different moods. We try to be fair, but we make no promises.

And remember, it's a big Internet. If you think we were mistaken in removing your post (or, in extreme cases, in removing you) -- or if you prefer a more free-for-all approach -- there are surely plenty of ways you can still get your views out.