Alex Johnson, UVA Professor discusses problems w/ LSAT Forum

[Kevinlomax]

http://www.youtube.com/watch?v=_7_xHsce57c

[Kevinlomax]

Very interesting video... he also states at the end of the video the LSAT may be gone in a few years

sorry if this is a repost

42:00 says LSAT may be gone in 2-3 years

[McAvoy]

2-3 years? I think he may have been wrong (considering this is from 2010...).

[Lightworks]

It may well be flawed, but it's much better than raw GPA.

[TheSpanishMain]

It may well be flawed, but it's much better than raw GPA.

This. At least the LSAT is the same for everyone. I can't imagine the nightmare of trying to come up with a fair way to compare GPAs based on the difficulty of the major, the grade inflation or lack thereof of the school, etc.

[Nucky]

Perhaps a second option to the LSAT, similar to the ACT/SAT is the best alternative.

Agree with GPA sentiment. But as a splitter, my opinion is admittedly biased.

[TigerDude]

according to the video, the LSAT has a better correlation than any other standardized test.

It's worth watching.

[McAvoy]

Agreed, it's well worth your time. Really interesting.

[lawschool22]

VERY interesting and informative.

If anyone is curious, I solved for the "old" UVA weighting formula he describes, and came up with this:

Index = .194175*(LSAT) + 5.82524*(GPA), where LSAT ranges from 120-180 and GPA ranges from .01-4.3.

The range of index should be 40-60. For out of state he said a 55+ would get in.

Obviously this is old, so should not be taken to be currently in use. Although my index using this formula was a 54.39 and I was WL'd, so...

[liammial]

Disagree that LSAT is a better metric than GPA. For someone who went to some obscure college and is 10 years out of school, then sure. But law schools have gotten thousands of applicants from UCLA, or Illinois, or Harvard. They know what typical GPAs are at those schools, they know the majors, they have access to the transcripts. It's really not difficult for an admissions officer to know what a good GPA is. The LSAT is important too, but it's one test, versus a GPA that includes hundreds of evaluations.

[lawschool22]

Disagree that LSAT is a better metric than GPA. For someone who went to some obscure college and is 10 years out of school, then sure. But law schools have gotten thousands of applicants from UCLA, or Illinois, or Harvard. They know what typical GPAs are at those schools, they know the majors, they have access to the transcripts. It's really not difficult for an admissions officer to know what a good GPA is. The LSAT is important too, but it's one test, versus a GPA that includes hundreds of evaluations.

No. Just no. A GPA is made of hundreds of non-standardized evaluations. Even among different majors at the same university you can see wild fluctuations. Did you miss the part where he said LSAT had a predictive correlation of .44 vs .16 for GPA? Or did you just comment without watching?

[ScottRiqui]

Disagree that LSAT is a better metric than GPA. For someone who went to some obscure college and is 10 years out of school, then sure. But law schools have gotten thousands of applicants from UCLA, or Illinois, or Harvard. They know what typical GPAs are at those schools, they know the majors, they have access to the transcripts. It's really not difficult for an admissions officer to know what a good GPA is. The LSAT is important too, but it's one test, versus a GPA that includes hundreds of evaluations.

No. Just no. A GPA is made of hundreds of non-standardized evaluations. Even among different majors at the same university you can see wild fluctuations. Did you miss the part where he said LSAT had a predictive correlation of .44 vs .16 for GPA? Or did you just comment without watching?

I agree with you, but was that 0.16 correlation the result of just taking GPAs as you find them, with no effort whatsoever to control for school/major/etc? It's possible that even some rough weighting of GPA depending on what degree you earned and when and where you earned it could bump that correlation up significantly.

[lawschool22]

Disagree that LSAT is a better metric than GPA. For someone who went to some obscure college and is 10 years out of school, then sure. But law schools have gotten thousands of applicants from UCLA, or Illinois, or Harvard. They know what typical GPAs are at those schools, they know the majors, they have access to the transcripts. It's really not difficult for an admissions officer to know what a good GPA is. The LSAT is important too, but it's one test, versus a GPA that includes hundreds of evaluations.

No. Just no. A GPA is made of hundreds of non-standardized evaluations. Even among different majors at the same university you can see wild fluctuations. Did you miss the part where he said LSAT had a predictive correlation of .44 vs .16 for GPA? Or did you just comment without watching?

I agree with you, but was that 0.16 correlation the result of just taking GPAs as you find them, with no effort whatsoever to control for school/major/etc? It's possible that even some rough weighting of GPA depending on what degree you earned and when and where you earned it could bump that correlation up significantly.

It's possible, but would take a ton of effort, time, etc to do that for every school and every major. That's the entire reason they developed the LSAT, so we have a standard. Of course the UGPA is helpful in broad strokes for identifying things the LSAT may not (task management, drive, ability to function in an academic setting). But seriously the difference between a 3.5 and a 3.8 is minimally predictive of anything due to the wide variation of schools, majors, profs, grading scales, course load, inflation, etc.

[liammial]

Disagree that LSAT is a better metric than GPA. For someone who went to some obscure college and is 10 years out of school, then sure. But law schools have gotten thousands of applicants from UCLA, or Illinois, or Harvard. They know what typical GPAs are at those schools, they know the majors, they have access to the transcripts. It's really not difficult for an admissions officer to know what a good GPA is. The LSAT is important too, but it's one test, versus a GPA that includes hundreds of evaluations.

No. Just no. A GPA is made of hundreds of non-standardized evaluations. Even among different majors at the same university you can see wild fluctuations. Did you miss the part where he said LSAT had a predictive correlation of .44 vs .16 for GPA? Or did you just comment without watching?

Did you miss the part where I said it was easy for law schools to control for things such as undergraduate rigor and major? Or did you just comment without reading?

[lawschool22]

Disagree that LSAT is a better metric than GPA. For someone who went to some obscure college and is 10 years out of school, then sure. But law schools have gotten thousands of applicants from UCLA, or Illinois, or Harvard. They know what typical GPAs are at those schools, they know the majors, they have access to the transcripts. It's really not difficult for an admissions officer to know what a good GPA is. The LSAT is important too, but it's one test, versus a GPA that includes hundreds of evaluations.

No. Just no. A GPA is made of hundreds of non-standardized evaluations. Even among different majors at the same university you can see wild fluctuations. Did you miss the part where he said LSAT had a predictive correlation of .44 vs .16 for GPA? Or did you just comment without watching?

Did you miss the part where I said it was easy for law schools to control for things such as undergraduate rigor and major? Or did you just comment without reading?

Of course I read your post. What makes you think that is easy to control for those things? You said for someone who went to an "obscure college, then sure." There are plenty of applicants who come from so-called "obscure" colleges that an adcom many know little about. Even at the "non-obscure" colleges, I don't think it's so easy to control for GPA among majors, etc. Even within the same class and same major, the professor you get can have a dramatic effect on your GPA.

Also, GPA inflation has changed over time, but at different rates among different schools, whereas the LSAT is scaled across all applicants. Inflation has become an issue, and at some surprising places (Harvard and other private colleges, for example).

I think it's easy to say "oh they can control for that," but if they could they would have done so and not needed to keep using the LSAT. I'm not saying the GPA is useless, but if you're trying to decide which is a more standardized, correlative metric, the LSAT wins hands down.

[liammial]

Disagree that LSAT is a better metric than GPA. For someone who went to some obscure college and is 10 years out of school, then sure. But law schools have gotten thousands of applicants from UCLA, or Illinois, or Harvard. They know what typical GPAs are at those schools, they know the majors, they have access to the transcripts. It's really not difficult for an admissions officer to know what a good GPA is. The LSAT is important too, but it's one test, versus a GPA that includes hundreds of evaluations.

No. Just no. A GPA is made of hundreds of non-standardized evaluations. Even among different majors at the same university you can see wild fluctuations. Did you miss the part where he said LSAT had a predictive correlation of .44 vs .16 for GPA? Or did you just comment without watching?

Did you miss the part where I said it was easy for law schools to control for things such as undergraduate rigor and major? Or did you just comment without reading?

Of course I read your post. What makes you think that is easy to control for those things? You said for someone who went to an "obscure college, then sure." There are plenty of applicants who come from so-called "obscure" colleges that an adcom many know little about. Even at the "non-obscure" colleges, I don't think it's so easy to control for GPA among majors, etc. Even within the same class and same major, the professor you get can have a dramatic effect on your GPA.

Also, GPA inflation has changed over time, but at different rates among different schools, whereas the LSAT is scaled across all applicants. Inflation has become an issue, and at some surprising places (Harvard and other private colleges, for example).

I think it's easy to say "oh they can control for that," but if they could they would have done so and not needed to keep using the LSAT. I'm not saying the GPA is useless, but if you're trying to decide which is a more standardized, correlative metric, the LSAT wins hands down.

I was just imitating your condescending tone. I think that the relative weighting of LSAT and GPA should depend on length of time removed from college and obscurity of college and major. Georgetown Law School knows what an Engineering major from Berkeley's GPA should be around.

The LSAT is obviously more standardized, and far more predictive than any single college exam. But I would guess that GPA correlates more with law school success if you were to look at people from similar majors from similar schools (humanities in Ivies, for instance). I mean, just think about it. You want to know how some random person is going to do in law school. What would be more helpful, his GPA in a well-known major at a well-known college, or his one-shot LSAT score? GPA, hands down.

Both are important. If I a law school sees someone with a 4.0 from Cal Tech, though, they're going to weigh that considerably more than the person's LSAT, as they should.

[Pneumonia]

I mean come on... what is your point?

[lawschool22]

I was just imitating your condescending tone. I think that the relative weighting of LSAT and GPA should depend on length of time removed from college and obscurity of college and major. Georgetown Law School knows what an Engineering major from Berkeley's GPA should be around.

The LSAT is obviously more standardized, and far more predictive than any single college exam. But I would guess that GPA correlates more with law school success if you were to look at people from similar majors from similar schools (humanities in Ivies, for instance). I mean, just think about it. You want to know how some random person is going to do in law school. What would be more helpful, his GPA in a well-known major at a well-known college, or his one-shot LSAT score? GPA, hands down.

Both are important. If I a law school sees someone with a 4.0 from Cal Tech, though, they're going to weigh that considerably more than the person's LSAT, as they should.

You want to know how some random person is going to do in law school, what would be more helpful? The LSAT. The data backs that up. I don't know how else to say it. You can't compare humanities in Ivies (http://www.thecrimson.com/article/2013/ ... on-mode-a/), or engineering at publics, or any other broad category you want to look at. You can't even compare majors within the SAME school, because, as I mentioned, people can take different professors who have different grading tendencies.

Perhaps if you were to look at students at the same school, with the same major, who took the same professors, the same amount of courses each semester, with the same personal life pressures, etc., you might be able to gain more predictive value from GPA than LSAT. But you would have to do that at every school, for every major, for it to be of use to admissions departments. It simply is not feasible, which is why they don't do it, and why we have a standardized test.

And I don't know why you keep mentioning "obscure" colleges or what you mean by that. If you're implying that "obscure" colleges necessarily have more grade inflation, and thus GPA should be weighted lower, you would be wrong about that. This is one of many reasons why there are not broad generalizations we can make about GPAs within majors, groups of schools, etc. There is simply too much variability.

This is why we have the LSAT. You can take the LSAT multiple times, so it's not like the "one shot" vs. "numerous measuring points" argument of the LSAT vs. GPA argument holds water. Also, if we're talking about predictive value, I think the high-stakes single-shot nature of the LSAT much more accurately reflects the way law school is. Every course grade is determined by one test that takes place on one day.

[liammial]

I was just imitating your condescending tone. I think that the relative weighting of LSAT and GPA should depend on length of time removed from college and obscurity of college and major. Georgetown Law School knows what an Engineering major from Berkeley's GPA should be around.

The LSAT is obviously more standardized, and far more predictive than any single college exam. But I would guess that GPA correlates more with law school success if you were to look at people from similar majors from similar schools (humanities in Ivies, for instance). I mean, just think about it. You want to know how some random person is going to do in law school. What would be more helpful, his GPA in a well-known major at a well-known college, or his one-shot LSAT score? GPA, hands down.

Both are important. If I a law school sees someone with a 4.0 from Cal Tech, though, they're going to weigh that considerably more than the person's LSAT, as they should.

You want to know how some random person is going to do in law school, what would be more helpful? The LSAT. The data backs that up. I don't know how else to say it. You can't compare humanities in Ivies (http://www.thecrimson.com/article/2013/ ... on-mode-a/), or engineering at publics, or any other broad category you want to look at. You can't even compare majors within the SAME school, because, as I mentioned, people can take different professors who have different grading tendencies.

Perhaps if you were to look at students at the same school, with the same major, who took the same professors, the same amount of courses each semester, with the same personal life pressures, etc., you might be able to gain more predictive value from GPA than LSAT. But you would have to do that at every school, for every major, for it to be of use to admissions departments. It simply is not feasible, which is why they don't do it, and why we have a standardized test.

And I don't know why you keep mentioning "obscure" colleges or what you mean by that. If you're implying that "obscure" colleges necessarily have more grade inflation, and thus GPA should be weighted lower, you would be wrong about that. This is one of many reasons why there are not broad generalizations we can make about GPAs within majors, groups of schools, etc. There is simply too much variability.

This is why we have the LSAT. You can take the LSAT multiple times, so it's not like the "one shot" vs. "numerous measuring points" argument of the LSAT vs. GPA argument holds water. Also, if we're talking about predictive value, I think the high-stakes single-shot nature of the LSAT much more accurately reflects the way law school is. Every course grade is determined by one test that takes place on one day.

The data doesn't back it up, though. There is no data to suggest that except a random correlated measure that takes literal raw GPA from any major in any school. I know some major idiots who got good GPAs at good schools. But I also know some people with no motivation who studied for the LSAT for a month and got a 170. The dumb kid who studies hard is going to do better in law school than the unmotivated brilliant kid.

Obviously GPA is flawed. I'm not recommending schools create this grandiose GPA weighting system that solves every problem. And, to reply to your other strawman: I'm also not advising schools to differentiate between applicants based on a few points of GPA. But I also don't want schools to differentiate applicants between a point on the LSAT. GPA is flawed as a predictive metric; LSAT is, for most people, more flawed.

"Obscure" colleges/majors refers to those the law school does not know much about, for whatever reason.

The notion that the LSAT is okay as a one-shot deal because law school is a one-shot deal makes little sense. They're both high variance, maybe, that that doesn't mean we should choose one high variance thing to predict another high variance thing. We want to choose the most predictive measure possible. (There is the argument that people study differently for a test that is a one-shot deal than for a regular exam, but I don't know if that's true. I'd bet people do better on one or the other because of effort or because the LSAT happens to test (or not test) particular skills innate in that person.)

[iamgeorgebush]

+1 to LS22. Within my major, there were profs who were known to be easy graders and profs who were known to be hard graders. Some people purposefully took classes with the easy ones to pad their GPAs, while some students didn't do so. It would be extremely difficult and time-consuming for LS admissions officers to figure out which profs are which.

[ScottRiqui]

GPA is flawed as a predictive metric; LSAT is, for most people, more flawed.

Except that the data we have suggests that when it comes to predictive ability, LSAT scores are better than GPA (as GPA is currently used). What's your evidence that the LSAT is "more flawed"?

[vuthy]

Honestly I don't even know how this can be an argument. Grade inflation has almost completely killed the usefulness of GPA. Sure, it's still helpful if you have an applicant with a crappy GPA, because (barring some credible excuse) it means that they were either overmatched or unmotivated in college. But high GPAs are a dime a dozen these days. Something like 50% of all college grades are A or A- now. This is one of the reasons why so many high-GPA students get so frustrated w/ the LSAT. They assume that their 3.7 means that they are top-level smart. But the LSAT doesn't have grade inflation to save them, and it's a serious wake-up call.

Bottom line: neither GPA nor LSAT is a great predictor of law school success, but colleges have essentially killed the GPA as a useful measurement. So between the two, you have to trust the LSAT more.

[lawschool22]

It sounds like he's a reverse splitter from a preftigious UG who didn't want to retake?

[liammial]

+1 to LS22. Within my major, there were profs who were known to be easy graders and profs who were known to be hard graders. Some people purposefully took classes with the easy ones to pad their GPAs, while some students didn't do so. It would be extremely difficult and time-consuming for LS admissions officers to figure out which profs are which.

Many colleges post median grades on transcripts (mine did).

Except that the data we have suggests that when it comes to predictive ability, LSAT scores are better than GPA (as GPA is currently used). What's your evidence that the LSAT is "more flawed"?

As I've said already, obviously GPA is less predictive if you literally take raw GPA without accounting whatsoever for school or major (which is what the study you're referring to uses). It's quite easy to account for those, at least a bit.

Honestly I don't even know how this can be an argument. Grade inflation has almost completely killed the usefulness of GPA. Sure, it's still helpful if you have an applicant with a crappy GPA, because (barring some credible excuse) it means that they were either overmatched or unmotivated in college. But high GPAs are a dime a dozen these days. Something like 50% of all college grades are A or A- now. This is one of the reasons why so many high-GPA students get so frustrated w/ the LSAT. They assume that their 3.7 means that they are top-level smart. But the LSAT doesn't have grade inflation to save them, and it's a serious wake-up call.

Bottom line: neither GPA nor LSAT is a great predictor of law school success, but colleges have essentially killed the GPA as a useful measurement. So between the two, you have to trust the LSAT more.

Law schools have access to: median grades, average GPA for your class, LSAT scores for your class, etc. They have ample ways to account for grade inflation, if it does in fact exist. Further, you're not competing with applicants from 20 years ago--you're competing with current applicants, under the same grade inflation. (Further, I do not believe there is any justification for grade inflation even existing separate from Harvard (and other schools') kids being smarter than they used to be.) Does MIT not have grade inflation? Law schools need only to look at the school's LSAC report (like the already don't know).

Why do you think top 15 schools are wildly overrepresented at top law schools? The kids being brighter--and thus scoring higher on the LSAT--can't account for all of it.

[lawschool22]

+1 to LS22. Within my major, there were profs who were known to be easy graders and profs who were known to be hard graders. Some people purposefully took classes with the easy ones to pad their GPAs, while some students didn't do so. It would be extremely difficult and time-consuming for LS admissions officers to figure out which profs are which.

Many colleges post median grades on transcripts (mine did).

Except that the data we have suggests that when it comes to predictive ability, LSAT scores are better than GPA (as GPA is currently used). What's your evidence that the LSAT is "more flawed"?

As I've said already, obviously GPA is less predictive if you literally take raw GPA without accounting whatsoever for school or major (which is what the study you're referring to uses). It's quite easy to account for those, at least a bit.

Honestly I don't even know how this can be an argument. Grade inflation has almost completely killed the usefulness of GPA. Sure, it's still helpful if you have an applicant with a crappy GPA, because (barring some credible excuse) it means that they were either overmatched or unmotivated in college. But high GPAs are a dime a dozen these days. Something like 50% of all college grades are A or A- now. This is one of the reasons why so many high-GPA students get so frustrated w/ the LSAT. They assume that their 3.7 means that they are top-level smart. But the LSAT doesn't have grade inflation to save them, and it's a serious wake-up call.

Bottom line: neither GPA nor LSAT is a great predictor of law school success, but colleges have essentially killed the GPA as a useful measurement. So between the two, you have to trust the LSAT more.

Law schools have access to: median grades, average GPA for your class, LSAT scores for your class, etc. They have ample ways to account for grade inflation, if it does in fact exist. Further, you're not competing with applicants from 20 years ago--you're competing with current applicants, under the same grade inflation. (Further, I do not believe there is any justification for grade inflation even existing separate from Harvard (and other schools') kids being smarter than they used to be.) Does MIT not have grade inflation? Law schools need only to look at the school's LSAC report (like the already don't know).

Why do you think top 15 schools are wildly overrepresented at top law schools? The kids being brighter--and thus scoring higher on the LSAT--can't account for all of it.

Dear god man, don't you see why a median GPA for your school is not helpful?

Compare the following four students:

UG median GPA: 3.5

Student A: Journalism major, 3.8
Student B: Journalism major, 3.2
Student C: Engineering major, 3.4
Student D: Engineering major, 3.8

Who do you admit? Note that's not rhetorical. Pick one.