question about LSP and splitters

(Applications Advice, Letters of Recommendation . . . )
scottybear
Posts: 29
Joined: Fri Nov 02, 2012 11:47 am

question about LSP and splitters

Postby scottybear » Sun Mar 24, 2013 5:54 pm

So, the common wisdom here is that law school predictor is pretty terrible for splitters. Is it so bad in that it is too generous (says you have a chance where you don't), too strict (says you don't have a chance where you do), or just randomly wrong (i.e. for most numbers, your relative chance of accuracy at any given school on LSP is pretty high, but as a splitter your chance of any given school being right is like 50/50)?

Are there patterns to the incorrectness? for example, I could see it being the case that it might be right on for, say, the top 20, because maybe they get enough applicants with good numbers that they don't need to take splitters (i know some are more splitter-friendly, but this is hypothetical), and then it might be off for the rest of t1 because who KNOWS what they want, and then it starts to get more accurate again in the 2/3/4 tiers, because they will take splitters to help their medians due to fewer applicants, etc?

I feel like the information on LSP, while not as helpful for splitters as for traditional applicants, could still be helpful for splitters if contextualized properly.

I've looked at all other prediction programs I could find, been pouring over LSN, etc, but haven't been able to come up with much due to the small sample sizes, etc. For what it's worth, I'm a pretty extreme splitter, and the weird thing is, for my numbers, LSP does a fairly accurate job at predicting (perhaps not at predicting admission) what LSN might say with the exception of a few schools (two in particular, both known GPA whores. interesting. Of course, LSP inflates my chances there).

I'm currently going over the data again. Anyone else have thoughts on just HOW lps is off? I'll post more info as I find it!

User avatar
Tiago Splitter
Posts: 15492
Joined: Tue Jun 28, 2011 1:20 am

Re: question about LSP and splitters

Postby Tiago Splitter » Sun Mar 24, 2013 6:39 pm

It's consistently overoptimistic for splitters, because it doesn't take GPA floors into account. It provides an index based on your GPA/LSAT combo, but while a 3.3/180 and a 3.7/170 might have the same index the 3.3/180 has a much lower chance to get in at a school that utilizes a 3.5 GPA floor.

scottybear
Posts: 29
Joined: Fri Nov 02, 2012 11:47 am

Re: question about LSP and splitters

Postby scottybear » Sun Mar 24, 2013 9:39 pm

hm.
The question I have is, what numbers are they using? It seems like if they give a "consider" to a school given some numbers, they (hopefully) have evidence to back that up, right?

For example, a 173/2.8 splitter gets a consider at UCLA, but a deny at USC even though UCLA has higher GPA medians than USC. Why would this be showing up, unless they had data to back this up? It seems that if it was just a numbers game, based on medians, rankings, or other abstract data, UCLA should always be harder for a splitter to get into than USC. but LSP doesn't seem to think this is the case.

So either there's some data to support this, or something else is going on. Going by the "how to be a successful splitter" guide, UCLA's GPA floor is 3.4. However, a quick search of splitters on mylsn.info over the last few years shows that for people with 170-180/2.0-3.39 numbers, 42 get in, 35 get denied, and then 131 get waitlisted, which is surprising (these are all non-URM).

Am I missing something, or what does this suggest about the GPA floor? Any insights on this? Even with a 175 and a 2.9 max on mylsn, it shows that 2 non-URM applicants got in and 12 got rejected. This is pretty surprising for a school with a supposed "hard" gpa floor of 3.4

Would anyone mind shedding some more light on this, either on the LSP numbers (i.e. where they get their data, etc) or on the idea of a GPA floor, as I'm clearly missing something there, too?

thanks!

wannabelawstudent
Posts: 2588
Joined: Sat Aug 13, 2011 7:33 pm

Re: question about LSP and splitters

Postby wannabelawstudent » Sun Mar 24, 2013 9:53 pm

Id counter Tiagos point in saying for me, it pretty much said I had less than a 5% chance of getting in anywhere I applied. But I was a mega mega splitter.

User avatar
Tiago Splitter
Posts: 15492
Joined: Tue Jun 28, 2011 1:20 am

Re: question about LSP and splitters

Postby Tiago Splitter » Sun Mar 24, 2013 10:00 pm

Many schools publish indexes, and where they don't LSP builds one for them using past data. The index for a particular school might be something like GPA*100 + LSAT*4. Under that system someone with a 3.2/177 has an index of 1028. Someone with a 3.7/170 has a 1050. The system will then look at all the prior data points from LSN, and see that, for example, 55% of people with an index below 1050 get in and 45% of those below 1028 get in. So both the 3.2/177 and the 3.7/170 get a "consider" from LSP. But if the school generally doesn't dip below the 3.4, the splitter's chances are really a lot worse than the other guy's even though LSP gives them a roughly equal shot. Again, LSP is NOT just comparing you to people with your numbers from LSN, but rather is comparing your index score to all prior applicants from LSN.

Using UCLA as an example, someone right at both medians has roughly the same index score as a splitter with a 2.9/177.

wannabelawstudent wrote:Id counter Tiagos point in saying for me, it pretty much said I had less than a 5% chance of getting in anywhere I applied. But I was a mega mega splitter.

LSP has made some adjustments over time to account for its issues with splitters. But you are still better off just using mylsn to get an idea of your chances.

wannabelawstudent
Posts: 2588
Joined: Sat Aug 13, 2011 7:33 pm

Re: question about LSP and splitters

Postby wannabelawstudent » Sun Mar 24, 2013 10:06 pm

I just used it this cycle, in my case my gpa was so low(<3.0, 167) that I was an auto reject everywhere according to lsp and in my actual cycle I was only rejected in 1 of the 15 places I applied. LSN is a much better resource as are TLS stats imo. Although who knows maybe the rest of my app was just amazing...

scottybear
Posts: 29
Joined: Fri Nov 02, 2012 11:47 am

Re: question about LSP and splitters

Postby scottybear » Sun Mar 24, 2013 10:52 pm

i see. thanks for all the thoughts!

But, for me, this doesn't quite explain why UCLA would have give a consider in any situation where USC would give a deny. If it's strictly numbers, shouldn't UCLA always be tougher than USC given rank, medians, etc?

Would you mind explaining this to me?

Again, thanks so much for your insights. Very helpful!

wannabelawstudent
Posts: 2588
Joined: Sat Aug 13, 2011 7:33 pm

Re: question about LSP and splitters

Postby wannabelawstudent » Sun Mar 24, 2013 10:54 pm

Some schools just place more value on the LSAT than others. Ive gotten into higher ranked schools (18-30) and waitlisted at lower randked (30-40) just based on how they value the LSAT.

Ti Malice
Posts: 1955
Joined: Sat Mar 03, 2012 2:55 am

Re: question about LSP and splitters

Postby Ti Malice » Sun Mar 24, 2013 11:01 pm

scottybear wrote:i see. thanks for all the thoughts!

But, for me, this doesn't quite explain why UCLA would have give a consider in any situation where USC would give a deny. If it's strictly numbers, shouldn't UCLA always be tougher than USC given rank, medians, etc?

Would you mind explaining this to me?

Again, thanks so much for your insights. Very helpful!


USC gives greater relative weight to GPA than UCLA.

LSP is useless in general, but it's worse than useless for splitters. It adds a completely needless and (very poorly executed) level of abstraction to the available data. Just use mylsn.info.

Edit: typo
Last edited by Ti Malice on Mon Mar 25, 2013 1:17 pm, edited 1 time in total.

scottybear
Posts: 29
Joined: Fri Nov 02, 2012 11:47 am

Re: question about LSP and splitters

Postby scottybear » Sun Mar 24, 2013 11:14 pm

right, but if the numbers are strictly indexed in such a way that a 17x and a 2.9 are indexed and given the same score as someone who is at the medians (and thus given the same chance of getting in), doesn't this imply that the weight that an individual school put s on GPA vs LSAT is effectively negated?

Or is it more complex than this?

thanks!

wannabelawstudent
Posts: 2588
Joined: Sat Aug 13, 2011 7:33 pm

Re: question about LSP and splitters

Postby wannabelawstudent » Sun Mar 24, 2013 11:16 pm

More complex than that. Its not a simple index that the school goes by. They're more concerned about yeild and medians.

User avatar
Tiago Splitter
Posts: 15492
Joined: Tue Jun 28, 2011 1:20 am

Re: question about LSP and splitters

Postby Tiago Splitter » Sun Mar 24, 2013 11:20 pm

scottybear wrote:right, but if the numbers are strictly indexed in such a way that a 17x and a 2.9 are indexed and given the same score as someone who is at the medians (and thus given the same chance of getting in), doesn't this imply that the weight that an individual school put s on GPA vs LSAT is effectively negated?

Or is it more complex than this?

thanks!

Each school has its own unique index.

A 2.9/177 gets an index score of 3.292 at UCLA and 3.384 at USC.
A 3.7/168 gets an index score of 3.266 at UCLA and 3.471 at USC.

Think of it as an exchange. Trading 9 LSAT points for .8 GPA points gets you a lot better return from USC.

scottybear
Posts: 29
Joined: Fri Nov 02, 2012 11:47 am

Re: question about LSP and splitters

Postby scottybear » Mon Mar 25, 2013 12:08 am

ah, got it. it's HOW each school creates THEIR index.

I was under the impression that it was a universal, abstract index that was then weighed against a school's "index medians", so to speak. This was my error and not implied by any of your explanations, which were all very helpful.

thanks all! I hope this thread comes in handy for someone else later on :)

I'll update with more info if any comes to light, but this has set my mind at ease.

thanks again!

wannabelawstudent
Posts: 2588
Joined: Sat Aug 13, 2011 7:33 pm

Re: question about LSP and splitters

Postby wannabelawstudent » Mon Mar 25, 2013 12:16 am

Nothing is as abstract as people assume it is.

User avatar
simplycatalina
Posts: 949
Joined: Tue Oct 09, 2012 7:41 pm

Re: question about LSP and splitters

Postby simplycatalina » Mon Mar 25, 2013 9:06 am

wannabelawstudent wrote:More complex than that. Its not a simple index that the school goes by. They're more concerned about yeild and medians.


Yeah it's not accurate for reverse splitters either. It said I would be denied or weakly considered at a couple places I've been accepted to.




Return to “Law School Admissions Forum”

Who is online

Users browsing this forum: bns212, dannyswo, MSNbot Media and 13 guests