Chicago ($75k) v UCLA (Full Ride +5k Stipend) Forum

[Knock]

But the most likely outcome is a very slightly higher class rank at UCLA, because UCLA students are a little bit weaker competition than Chicago students.

Yeah see. See that. That right there. That is wrong.

What, the premise or the conclusion?

If you challenge the premise (UCLA students are a little bit weaker competition than Chicago students), consider that LSAT & UGPA are the best available measures of academic potential and are correlated with intelligence and law school performance. That tells us the strength of the competition. With the best measures we have, UCLA is weaker than Chicago. The difference is very small, but not zero.

If you challenge the conclusion, consider that law school is a mechanism for ranking students according to certain academic abilities, and if you are dropped into a pool of weak students, you are going to be closer to the top than if you are dropped into a pool of stronger students.

I reject both. The LSAT and UGPA are NOT correlated in any statistically significant way with law school performance and LOL are even worse measures of general intelligence. You read these boards. This place is chock full of people with high LSATs and high GPAs that are idiots and full of people with lower scores and lower gpas that are absolutely brilliant. Obviously this isn't a significant sample size either but it provides a counterpoint to your claim. Moreover, Berkeley has a lower LSAT than most of the t14. Does that mean its easier to go to law school at Berkeley? No. Of course not, and if anyone here said "go to berkeley because it has a lower LSAT median than UVA," they would get laughed at. The fact is that neither of those two measures tell us ANYTHING about the strength of the competition at any law school. Go to Brigham Young University - you'll have about as hard of a time placing in the top 10% there as you will in almost any other school. The difference in the top 10% between Michigan and Columbia is almost zero. The different between median at UVA and median at berkeley, even though UVA has a higher LSAT, is negligble. To presume otherwise shows a lack of understand of how law school admissions work, how law schools test their students, and how human intelligence can effectively be measured.

Moreover, everything stated above assumes that doing well on the LSAT and having a high gpa = doing well in law school. Law school is a completely different animal. You can only claim to understand how it works once you're there (and I'm making no such claims of understanding). To do well there, from what 1L's 2L's and 3L's on this board have repeatedly said takes a completely different way of studying than anything you've ever done before. When that happens, every measure of "intelligence" is thrown out the window and the ball park is open for anybody with a certain set of skills to come out on top.

As such, your conclusion is flawed with respect to this example. Because we can't measure what a "weak pool of students is" we cannot conclude that the students at UCLA are weaker than the students at Chicago, and as such it is impossible to say it is easier to place better at one school over the other.

This topic has been beaten to death, but I wanted to add a statistics question. Correct me if i'm wrong, but I was under the impression that while LSAT + GPA have a correlation of .35 (or so, i'm remembering off the top of my head) to 1L grades, these models are not designed for individual predictive power. So while it may be true there is a weak to moderate correlation in general between LSAT + GPA, it's not accurate to make individual predictions based on this correlation.

[Curry]

This topic has been beaten to death, but I wanted to add a statistics question. Correct me if i'm wrong, but I was under the impression that while LSAT + GPA have a correlation of .35 (or so, i'm remembering off the top of my head) to 1L grades, these models are not designed for individual predictive power. So while it may be true there is a weak to moderate correlation in general between LSAT + GPA, it's not accurate to make individual predictions based on this correlation.

I think LSAT alone has a .35 correlation and gpa has a .25 correlation. I don't know what the aggregate is. That being said, you basically tl;dr what I was trying to say.

[voice of reason]

Curry, you wrote that "LSAT and UGPA are NOT correlated in any statistically significant way with law school performance," but this is not true.

LSAT & UGPA are the two strongest statistically significant predictors of performance in law school. Across law schools, the median correlation between law school grades and the LSAT+UGPA combination is about .5. In predictions of human behavior, this is moderately strong.

You observe that one can't count on doing well (or poorly) in law school just because one has high (or low) scores. Of course that's true, because the predictions are far from perfect. Nonetheless, because LSAT & UGPA predict law school performance moderately well in the aggregate, they are decent measures of the level of competition a student would face at any given school.

(Source of the correlation cited above:
http://www.lsac.org/jd/pdfs/LSAT-Score- ... rmance.pdf)

[Curry]

I don't have time to argue this with you sadly. Use the search function. People far more knowledgable on the issue have discussed this time and time again here and the general consensus is that it has almost no effect.

[adt231]

UCLA

[d34d9823]

I don't have time to argue this with you sadly. Use the search function. People far more knowledgable on the issue have discussed this time and time again here and the general consensus is that it has almost no effect.

Have to disagree with you here, curry. 0.47, which is the median correlation for LSAT+UGPA (http://www.lsac.org/jd/pdfs/LSAT-Score- ... rmance.pdf), is impressively strong for a real world correlation like this that has many random inputs. Obviously, the statistical prediction is going to be something like X% higher class rank at UCLA +- 2X% , but if you're maximizing your statistically predicted performance, it's something to take into account.

[spacepenguin]

I don't have time to argue this with you sadly. Use the search function. People far more knowledgable on the issue have discussed this time and time again here and the general consensus is that it has almost no effect.

Have to disagree with you here, curry. 0.47, which is the median correlation for LSAT+UGPA (http://www.lsac.org/jd/pdfs/LSAT-Score- ... rmance.pdf), is impressively strong for a real world correlation like this that has many random inputs. Obviously, the statistical prediction is going to be something like X% higher class rank at UCLA +- 2X% , but if you're maximizing your statistically predicted performance, it's something to take into account.

One also has to realize that that correlation was derived from a span of ALL law schools, including those with larger LSAT bands. When you get into the top 18 schools or so, you're largely dealing with students in the 95th percentile and above, making the LSAT less of a reliable predictor in terms of 1st year grades. That is, on the higher end of the spectrum the margin of error becomes increasingly small in determining the difference between a 169 and 170, or a 170 and a 175; statistically, the 'difference' in intelligence might easily be mitigated by something non-measurable. Also, while a .47 correlation is the BEST model that currently exists, it doesn't make it a great one since only about 22% of the variance of 1L grades is explained by UGPA+LSAT.