Law School Predictor: The Thread"

(Applications Advice, Letters of Recommendation . . . )
User avatar
YCrevolution
Posts: 4714
Joined: Sat Oct 25, 2008 1:25 am

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby YCrevolution » Sun Jul 12, 2009 4:34 pm

..

User avatar
Bronte
Posts: 2128
Joined: Sun Jan 04, 2009 10:44 pm

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby Bronte » Sun Jul 12, 2009 5:26 pm

YCrevolution wrote:They've been updated to 2009-10 formulas (as of Saturday, I've just changed the FAQ page to reflect the update). The formula sheet is available on the LSAC website (it should be near a link to your Academic Summary Report).

If you PM me with your email address, I can send you the two official PDF files of the 2009-10 formulas and 2008-09 formulas.


I believe I've found it (admissionsindex.pdf). They sure don't make it easy. Thanks for the help.

User avatar
Bronte
Posts: 2128
Joined: Sun Jan 04, 2009 10:44 pm

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby Bronte » Sun Jul 12, 2009 6:16 pm

I've noticed you're using an interpolated median value (i.e., halfway between the 25th/75th). I think you describe it as such on the About page. I'm sure this has been discussed/considered, but given that (a) the medians are the most important value and (b) the distributions tend to be left skewed, this might not be ideal.

Of course, for some reason the medians are hard to come by, but have you thought about trying to use actual medians instead?

User avatar
YCrevolution
Posts: 4714
Joined: Sat Oct 25, 2008 1:25 am

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby YCrevolution » Sun Jul 12, 2009 6:29 pm

..

User avatar
Bronte
Posts: 2128
Joined: Sun Jan 04, 2009 10:44 pm

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby Bronte » Sun Jul 12, 2009 6:43 pm

YCrevolution wrote:I could for Canadian schools, because, oddly enough, they don't seem to publish 25%/75% data. In the end, I'll probably just use estimates for the Canadian schools.

With the way the predictions are set up right now, anything but the most up-to-date median information could start yielding some weird predictions. I haven't really felt motivated enough to try and hunt down the median information at every ABA school, and, for the time being, I'm relatively confident that the predictions wouldn't be too much improved with the medians instead of the interpolated median value.

On the other hand, if somebody would like to do that for me (get the 07-08 cycle median data/Class of 2011 from each school's website*), I'd be happy to incorporate it into the predictions (and give credit to whoever puts the list together).

*That's really the only way. Some sites claim to have a comprehensive listing of median values, but they often just do what I do (average the 25% and 75% numbers).


I'm going to put together a personal spreadsheet for the schools I'm looking at (T20 and lower T14), mainly just for fun. I already, of course, have a pretty clear picture of my chances given the available resources, including yours. Anyway, I'll let you know how big the difference is and maybe we could collaborate on putting it together, like you said. I already did Vandy:

Interpolated median index: 3.367
Reported median index: 3.417

Fairly significant, I suppose.

User avatar
YCrevolution
Posts: 4714
Joined: Sat Oct 25, 2008 1:25 am

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby YCrevolution » Sun Jul 12, 2009 6:52 pm

..

User avatar
Bronte
Posts: 2128
Joined: Sun Jan 04, 2009 10:44 pm

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby Bronte » Sun Jul 12, 2009 7:02 pm

YCrevolution wrote:I'll look into a couple of test cases myself, too. It's more of a matter that it's going to be a pain to find and a pain to update, and to my knowledge, nobody verifies the veracity of median LSAT/GPA claims.


Right, I understand that the legwork is the main nuisance. It's true that the medians rather than the 25th/75th are the USNWR rankings driver, correct? (And, thus, the most important figure to the schools themselves.) If so, it seems absurd that these are not officially reported. Anyway, I'll scrape together the medians for the top 20 and calculate the average percentage difference between the interpolated and reported medians.

User avatar
YCrevolution
Posts: 4714
Joined: Sat Oct 25, 2008 1:25 am

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby YCrevolution » Sun Jul 12, 2009 7:19 pm

..

User avatar
Bronte
Posts: 2128
Joined: Sun Jan 04, 2009 10:44 pm

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby Bronte » Sun Jul 12, 2009 7:38 pm

YCrevolution wrote:I think USNWR might use interpolated median values, too, although I can't find information to confirm this one way or the other. I have a premium subscription (so I can update LSP), and USNWR doesn't list median data, just 25/75 data.


I've heard it said on TLS that the 50th drives the rankings, but only the 25th/75th are reported. Would be interesting to get confirmation. There's a thread circulating around with the 2008-2009 medians all listed (viewtopic.php?f=2&t=78458). This is also, I think, where the fact regarding the medians driving the rankings was asserted.

Another question (and feel free, of course, to scram when you're bored with me): For the schools that don't publish their index formula, what was your method?

Edit: so far, the medians listed in that thread match the medians listed at the UCLA and Vandy websites, which in turn list 25/75 that match those listed on the TLS rankings page.

Oe.Maas
Posts: 15
Joined: Mon Nov 10, 2008 4:51 pm

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby Oe.Maas » Sun Jul 12, 2009 7:48 pm

YCrevolution wrote:I could for Canadian schools, because, oddly enough, they don't seem to publish 25%/75% data. In the end, I'll probably just use estimates for the Canadian schools.


The medians are self-reported and each school has its own way of calculating GPA. Some will drop your worst 10 credits, some will take your best 2 years, some 3, etc.

User avatar
YCrevolution
Posts: 4714
Joined: Sat Oct 25, 2008 1:25 am

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby YCrevolution » Sun Jul 12, 2009 7:50 pm

..

User avatar
YCrevolution
Posts: 4714
Joined: Sat Oct 25, 2008 1:25 am

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby YCrevolution » Sun Jul 12, 2009 7:54 pm

..

Oe.Maas
Posts: 15
Joined: Mon Nov 10, 2008 4:51 pm

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby Oe.Maas » Sun Jul 12, 2009 8:04 pm

YCrevolution wrote:I know they've got some odd ways of calculating GPA; I was looking through it the other day, and realized Canadians always liked to make things more complicated :D . On the bright side, there's apparently at least one Canadian law school where GPA > LSAT (Dalhousie?).


I think that is the case for a couple places, especially at civil law schools, where the LSAT isn't even required. Not sure about Dalhousie though. Overall there is much less pressure to get 170+ on the LSAT. Even at Toronto, which is the hardest school to get into numberswise, the median there is 167 or something like that.

It's a lot more unpredictable compared to American schools I think. A sociology PhD w/ a 143 LSAT got into McGill a couple years ago. :twisted:

User avatar
Bronte
Posts: 2128
Joined: Sun Jan 04, 2009 10:44 pm

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby Bronte » Sun Jul 12, 2009 8:06 pm

YCrevolution wrote:Linear regression using 25/interpolated median/75 data. Specifically, the interpolated median is weighted double of either percentile (50% interpolated median, 25% 25-percentile, 25% 75-percentile), except for maybe a couple schools where I just used the 25/75 data (I think usually when there was a large gap between 25/75 numbers). In some cases, I made further adjustments based on LSN data or changed the constant to something I like.


I had to stare at this for a minute. Let me see if I understand. You put three data points into a bivariate regression analysis, those data points being (LSAT, GPA) at the 25th/50th(est)/75th. Then you used the regression equation to deduce the constants? I guess I'm getting stuck at this point at y = mx + b. Doesn't this only give us one multiplication factor?

Edit: err, I suppose it can be rewritten as b = mx + ky?

User avatar
YCrevolution
Posts: 4714
Joined: Sat Oct 25, 2008 1:25 am

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby YCrevolution » Sun Jul 12, 2009 8:48 pm

..

User avatar
YCrevolution
Posts: 4714
Joined: Sat Oct 25, 2008 1:25 am

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby YCrevolution » Sun Jul 12, 2009 8:50 pm

..

User avatar
Bronte
Posts: 2128
Joined: Sun Jan 04, 2009 10:44 pm

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby Bronte » Sun Jul 12, 2009 9:09 pm

YCrevolution wrote:Yes, which is what I originally did in Excel, but I don't think that solved the problem that it was making the GPA a dependent variable (based on the LSAT).


Right. Thinking about it now, linear regression (the way I'm thinking of it) doesn't quite make sense because it essentially wants to predict one's GPA based on an LSAT or vice versa. On the other hand, an admission index formula seeks instead to assign relative weights to two variables and output a single score based on their values. You probably thought up something different (i.e., other than linear regression) at the time. I'll tinker around with it. One option, of course, would be to simply use another schools formula.

User avatar
T14
Posts: 121
Joined: Thu Dec 04, 2008 10:55 pm

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby T14 » Sun Jul 12, 2009 9:20 pm

YCrevolution wrote:I know they've got some odd ways of calculating GPA; I was looking through it the other day, and realized Canadians always liked to make things more complicated :D . On the bright side, there's apparently at least one Canadian law school where GPA > LSAT (Dalhousie?).


I know the University of Victoria weights GPA/LSAT 70/30.

Index calculation is (175 x GPA) + (3 x LSAT percentile) = (anything over 925 is an auto-admit)

User avatar
YCrevolution
Posts: 4714
Joined: Sat Oct 25, 2008 1:25 am

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby YCrevolution » Sun Jul 12, 2009 9:57 pm

..

User avatar
YCrevolution
Posts: 4714
Joined: Sat Oct 25, 2008 1:25 am

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby YCrevolution » Mon Jul 13, 2009 11:01 am

..

krispykitten
Posts: 441
Joined: Mon May 25, 2009 6:44 pm

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby krispykitten » Mon Jul 13, 2009 11:07 am

.
Last edited by krispykitten on Thu Jan 20, 2011 5:59 pm, edited 2 times in total.

User avatar
PrayingforHYS
Posts: 303
Joined: Mon Jul 13, 2009 12:37 pm

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby PrayingforHYS » Mon Jul 13, 2009 1:04 pm

I have a seemingly very important question about this predictor and how it doesn't take your UG school into account, however.

For example, I'm about to enter my Junior Year with a 3.95, and while I expect my GPA to actually go up over the next 2 years, I'll be conservative and say it drops to 3.9. Well, I've been prepping for the LSAT this summer (thinking about taking it this Sept) and my PTs have been low 170s to high 170s. Ultimately, my goal is a 175, which I think is very reasonable and doable by the time I complete my studying/doing all the past LSATs.

Now, despite my HS stats (2200+ SAT, 3.85 unweighted GPA at arugably the top public HS in one of the 5 biggest states [yes, they have rankings for those, and no I don't live in Wyoming]), I'm at a relatively small (about 4000 UGs) private, catholic, liberal arts school (not important why).

According to this predictor, a 175/3.9 makes Yale: consider, HLS: Strong, and every other school an admit

Even if I'm a little conservative with my LSAT score too and I say 173/3.9, Yale and HLS are Consider, CLS: Strong, All others = admit

Obviously this doesn't take my UG university into account, and I'm curious as to how much that will factor into my chances. I realize that GPA and LSAT are the two primary factors, but a 175/3.9 at an Ivy should obviously make you an auto admit everywhere but YLS/HLS, but a 175/3.9 at a good, but small, catholic univ? Obviously not the same situation. The 175/3.9 would obviously be a good indicator that I'm intelligent (the 175 more so b/c it's standardized), but how do you guys think my UG insitituion will affect this?

It's not like I'm at East Nowheresville State, but I would think that it's a slight disadvantage.

I'd really appreciate input from anyone.

User avatar
YCrevolution
Posts: 4714
Joined: Sat Oct 25, 2008 1:25 am

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby YCrevolution » Mon Jul 13, 2009 1:12 pm

..

User avatar
Bronte
Posts: 2128
Joined: Sun Jan 04, 2009 10:44 pm

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby Bronte » Mon Jul 13, 2009 2:08 pm

So, from Emory to Berkeley, omitting schools that don't publish formulas, the average difference between [interpolated median] / [perfect index] and ["true" median] / [perfect index] is 0.73%. I'll leave it up to you whether or not that's significant, but the fact that the medians are all published in one location on TLS (viewtopic.php?f=2&t=78458) does lessen the legwork, although the data's veracity may be suspect.

One issue that these medians bring up is that they make the third quartile narrower, making the difference between "strong consider" and "accept" more vague. Overall, however, it does make the predictions more conservative.

To PrayingforHYS: The generally accepted response is that undergrad institution does not make significant difference in admissions. Calculators like lawschoolpredictor.com and lawschoolcalculator.com are strictly based on "hard factors" (LSAT/GPA/Race). Anything soft matters should be taken into account qualitatively on your part and considered as a small plus or minus. Also, in regards to the LSAT, be ambition but don't get your hopes up too high. A 175 is never guarantee, or even a probability, regardless of where your PTs are.

User avatar
PrayingforHYS
Posts: 303
Joined: Mon Jul 13, 2009 12:37 pm

Re: Law School Predictor Version 2.3 - Published July 11, 2009

Postby PrayingforHYS » Mon Jul 13, 2009 4:23 pm

Bronte wrote:So, from Emory to Berkeley, omitting schools that don't publish formulas, the average difference between [interpolated median] / [perfect index] and ["true" median] / [perfect index] is 0.73%. I'll leave it up to you whether or not that's significant, but the fact that the medians are all published in one location on TLS (http://top-law-schools.com/forums/viewt ... =2&t=78458) does lessen the legwork, although the data's veracity may be suspect.

One issue that these medians bring up is that they make the third quartile narrower, making the difference between "strong consider" and "accept" more vague. Overall, however, it does make the predictions more conservative.

To PrayingforHYS: The generally accepted response is that undergrad institution does not make significant difference in admissions. Calculators like lawschoolpredictor.com and lawschoolcalculator.com are strictly based on "hard factors" (LSAT/GPA/Race). Anything soft matters should be taken into account qualitatively on your part and considered as a small plus or minus. Also, in regards to the LSAT, be ambition but don't get your hopes up too high. A 175 is never guarantee, or even a probability, regardless of where your PTs are.


Thanks for the advice. By no means do I expect to get a 175 on the LSAT, but I've always tested very well on standardized tests, and--sadistic as this may sound--i actually find studying for LSATs mildly enjoyable (well, certainly the LG section at least)

The only bible that I've used so far is that for the LGs, and with only 1/3 of my test preparation done (1/4 really since LG is just 1 of the 4 graded sections, obviously), I'm scoring in at least the low 170s very consistently. All I meant was that I'm hoping that when I piece together the other sections (by reading the LR and RC bibles), that I'll be able to attain my target score of 175.

Thank you for the advice on UG institution though -- I appreciate it.




Return to “Law School Admissions Forum”

Who is online

Users browsing this forum: No registered users and 2 guests