LSAT bash Forum

[swfangirl]

1) POOR PEOPLE CAN AFFORD THE IVY LEAGUE DUE TO VERY GENEROUS FINANCIAL AID. In fact, it's often cheaper than their state school alternative.

LOLwut?

Yeah, I know people here who say that when they crunched the numbers, going here was cheaper than going to their state school...because they were given free tuition, health insurance, room, and board, and given funds for travel to/from, books, and personal expenses. And the amount allotted for some of that (travel, food, books, personal expenses) even exceeded what they needed. I don't quite get why people don't believe this.

Because you're talking about one school as if that evidence makes it true everywhere when that is not the case at all.

I don't think it's a stretch what is true for one of 8 is true for all 8 when this specific school is one of the poorest of the entire league. I'm not talking about HYP--they're known to be the most generous by far. The very poor get full rides at all 8 and I say this with absolute certainty.

ETA: This doesn't apply if you're international. This applies domestically.

[taxguy]

Moreover, any test that is as time senstive as the LSAT, can't be that accurate a predictor. After all, law students do get plenty of time for outlines, briefs and final prep.

Don't they ultimately have to perform well on an exam under timed conditions?

I have been a lawyer for many years. For what it's worth, I agree with you. The LSAT is one of the worst standardized tests in existence. I have seen FAR to many outliers, both ways, for me to accept that test as a valid indicator of law school performance or how you will perform in law. I personally know two lawyers who scored at the bottom of their law school class, LSAT wise, yet graduated in the top5%. I know folks who aced the LSAT and did horribly in law school. The corelation statisics mentioned do not show how wide the outliers can be.

Come on, that's like saying that global warming can't be real because it still snows. Predictive validity is about huge statistical trends. Of course there are going to be a lot of outliers.

If the LSAT was so highly corelated with law school performance, the American Bar Association panel studying them wouldn't have recommended that the test become optional due to the "lack of reliability." Also, ask any lawyer. Feel free to ask as many as you want. They will tell you about the huge number of outliers that they have met. This feeling among most lawyers is common that the Lsat has questionalble significance.

[bk1]

If the LSAT was so highly corelated with law school performance, the American Bar Association panel studying them wouldn't have recommended that the test become optional due to the "lack of reliability."

Uhhh, that is not why the ABA considered dropping the LSAT as an accreditation requirement at all.

[Sandro]

Oh come on. Whoop de do some good lawyers did bad on the LSAT, or something. You're forgetting the LSAT isnt used as a tool to predict how good of a lawyer you will be but rather an ADMISSIONS tool to differentiate the hundreds of thousands of law school applications that schools receive each year. The LSAT has more statistical validity in predicting future LAW SCHOOL (the people who use the LSAT) success than GPA. What is so hard to accept about that ?


Take 400 students and put them at an imaginary T-14 school. 100 have 140s - 100 have 150s - 100 have 160s - and 100 have 170s. Identical(standardized, as well) GPAs. Multiply this experiment times 200 law schools. If there is no statistical evidence that on average the higher the LSAT score the higher the class rank/performance I will stand corrected.

[joebloe]

Oh come on. Whoop de do some good lawyers did bad on the LSAT, or something. You're forgetting the LSAT isnt used as a tool to predict how good of a lawyer you will be but rather an ADMISSIONS tool to differentiate the hundreds of thousands of law school applications that schools receive each year. The LSAT has more statistical validity in predicting future LAW SCHOOL (the people who use the LSAT) success than GPA. What is so hard to accept about that ?


Take 400 students and put them at an imaginary T-14 school. 100 have 140s - 100 have 150s - 100 have 160s - and 100 have 170s. Identical(standardized, as well) GPAs. Multiply this experiment times 200 law schools. If there is no statistical evidence that on average the higher the LSAT score the higher the class rank/performance I will stand corrected.

I recall a thread a week or two ago about a talk this LSAC guy gave at UVA. He gave a very similar example, involving a 1L class composed of something like 20 students in each grouping, and what quartile class rank into which the students would be divided. It was interesting, though I must confess that I don't know dick about stats.

[paulinaporizkova]

If the LSAT was so highly corelated with law school performance, the American Bar Association panel studying them wouldn't have recommended that the test become optional due to the "lack of reliability."

Uhhh, that is not why the ABA considered dropping the LSAT as an accreditation requirement at all.

I heard about this. Why was it anyway?

[d34d9823]

If the LSAT was so highly corelated with law school performance, the American Bar Association panel studying them wouldn't have recommended that the test become optional due to the "lack of reliability."

Uhhh, that is not why the ABA considered dropping the LSAT as an accreditation requirement at all.

I heard about this. Why was it anyway?

Because they don't understand statistics.

[bk1]

If the LSAT was so highly corelated with law school performance, the American Bar Association panel studying them wouldn't have recommended that the test become optional due to the "lack of reliability."

Uhhh, that is not why the ABA considered dropping the LSAT as an accreditation requirement at all.

I heard about this. Why was it anyway?

They basically said something like "we should let schools choose how they evaluate applicants and not force them to all use the same thing."

http://abovethelaw.com/2011/01/aba-cons ... aw-school/

[Hannibal]

If the LSAT was so highly corelated with law school performance, the American Bar Association panel studying them wouldn't have recommended that the test become optional due to the "lack of reliability."

Uhhh, that is not why the ABA considered dropping the LSAT as an accreditation requirement at all.

I heard about this. Why was it anyway?

They basically said something like "we should let schools choose how they evaluate applicants and not force them to all use the same thing."

http://abovethelaw.com/2011/01/aba-cons ... aw-school/

This. The LSAT is one of the best, if not the best standardized test correlating to performance in a professional school.

[bk1]

This. The LSAT is one of the best, if not the best standardized test correlating to performance in a professional school.

I would say mainly thanks to the guiding hands behind it as well as the ABA forcing law school curricula to all look like clones of each other.

[sch6les]

.

[delusional]

Oh come on. Whoop de do some good lawyers did bad on the LSAT, or something. You're forgetting the LSAT isnt used as a tool to predict how good of a lawyer you will be but rather an ADMISSIONS tool to differentiate the hundreds of thousands of law school applications that schools receive each year. The LSAT has more statistical validity in predicting future LAW SCHOOL (the people who use the LSAT) success than GPA. What is so hard to accept about that ?


Take 400 students and put them at an imaginary T-14 school. 100 have 140s - 100 have 150s - 100 have 160s - and 100 have 170s. Identical(standardized, as well) GPAs. Multiply this experiment times 200 law schools. If there is no statistical evidence that on average the higher the LSAT score the higher the class rank/performance I will stand corrected.

I recall a thread a week or two ago about a talk this LSAC guy gave at UVA. He gave a very similar example, involving a 1L class composed of something like 20 students in each grouping, and what quartile class rank into which the students would be divided. It was interesting, though I must confess that I don't know dick about stats.

I quoted him in this discussion a few pages ago. The guy was a former dean of UVA and HNIC (can white people say that if they don't spell out each word?) of LSAC, and he made a number of points that are relevant to this thread's interests. He said that when he was dean, they had an index threshold - over X and in-state was accepted, over Y and out of state. He said that he was personally aware of deans of admission that had received ultimatums: Raise the 25% and 75% numbers this year, or you're out. He said that as the dean, he would have a graph made each semester of LSAT/GPA index, correlated to grades - and it was almost generally highly correlated.

He broke down the numbers as follows - 25 people in each score group - 140, 150, 160, 170. According to the statistics, 12 people from the highest LSAT would be in the highest grade quartile, and twelve from the lowest would be in the lowest quartile. Another eight from the highest quartile would be above median, and vice versa for the lowest LSAT quartile. So 80% of the highest LSAT quartile ends up above median.

[northwood]

If the LSAT was taken out the T-14 (and beyond) would consist of solid 4.0's. They would have to find another way to create a barrier to entry. The LSAT is good at doing this if nothing else (even if it doesn't measure LS preformance, it is still a fairly accurate measure of intelligence).

the thing about removing the LSAT, is you need something to even out the differences between majors. If you took an easy major, from an easy school- you should have a higher gpa. That would hurt all of the hard majors out there, and reward those who took mickymouse majors. That wont help the school pick the best and will hurt anyone with a difficult major who decided on law late in their undergrad. now if you were to make the lsat more reading and writing based, and got rid of the games, it may be better, but it will always be imperfect.

[bk1]

Let's not forget that no school actually has LSATs within the entire range. The cluster at each school so small that the predictive nature of it becomes meh.

[JazzOne]

Let's not forget that no school actually has LSATs within the entire range. The cluster at each school so small that the predictive nature of it becomes meh.

But isn't that already reflected in the low correlation? I mean, there is no way to compare GPAs from different law schools. So, it seems to me (not having read any studies) that the only comparisons that can be studied are those among students at the same school. So, the cluster of scores at particular schools is not an additional reason to discount the LSAT. Instead, its an explanation for the low correlation that we're already aware of. Your suggestion would be akin to discounting the predictive value twice for a single phenomenon.

[bk1]

Let's not forget that no school actually has LSATs within the entire range. The cluster at each school so small that the predictive nature of it becomes meh.

But isn't that already reflected in the low correlation? I mean, there is no way to compare GPAs from different law schools. So, it seems to me (not having read any studies) that the only comparisons that can be studied are those among students at the same school. So, the cluster of scores at particular schools is not an additional reason to discount the LSAT. Instead, its an explanation for the low correlation that we're already aware of. Your suggestion would be akin to discounting the predictive value twice for a single phenomenon.

This is a much better wording of what I meant.

The LSAT is also a better predictor for grades of African Americans than whites and I wonder if this is because African Americans, due to URM bump, are more often outliers compared to the overall LSAT range of the school.

[rundoxierun]

Let's not forget that no school actually has LSATs within the entire range. The cluster at each school so small that the predictive nature of it becomes meh.

But isn't that already reflected in the low correlation? I mean, there is no way to compare GPAs from different law schools. So, it seems to me (not having read any studies) that the only comparisons that can be studied are those among students at the same school. So, the cluster of scores at particular schools is not an additional reason to discount the LSAT. Instead, its an explanation for the low correlation that we're already aware of. Your suggestion would be akin to discounting the predictive value twice for a single phenomenon.

Funny thing is that it isnt that low.. what people dont understand is that in any real life study the correlation is always "low". In fact, if it was what people would see as "high" that would make the data really suspicious. Often when correlations are "high" the study is a fake.