Law School Discussion Forums
https://www.top-law-schools.com/forums/
https://www.top-law-schools.com/forums/viewtopic.php?f=9&t=136068
LSP is an algorithm designed to model the raw data from LSN. Some schools have gpa/lsat floors that you can see in LSN but aren't necessarily accounted for in LSP. Are you a splitter? LSP is usually not as useful for splitters/urms as it is for 'normal' applicants. Splitters/urms tend to have unpredictable cycles, and there just isn’t enough data out there to make an algorithm super accurate for splitters/urms.AP-375 wrote:Yeah, I've gone over that, and I get that Predictor is imperfect. But it just seems like there is too big of a discrepancy in my chances to be written off as margin of error.
After a poor showing on the LSAT, I'm suddenly finding myself in the unpleasant world of reverse splitter-dom. That must be the issue that I am encountering.hipstermafia wrote:LSP is an algorithm designed to model the raw data from LSN. Some schools have gpa/lsat floors that you can see in LSN but aren't necessarily accounted for in LSP. Are you a splitter? LSP is usually not as useful for splitters/urms as it is for 'normal' applicants. Splitters/urms tend to have unpredictable cycles, and there just isn’t enough data out there to make an algorithm super accurate for splitters/urms.AP-375 wrote:Yeah, I've gone over that, and I get that Predictor is imperfect. But it just seems like there is too big of a discrepancy in my chances to be written off as margin of error.
I'd say very. While you might have a few people lying here and there, it is pretty easy to see where the hard cutoffs are and it shows you the general acceptance chances for LSAT/GPA combos.JDHopeful wrote:exactly how accurate it LSN, btw?
What do you mean? They are set up by applicant classes. 2009-2010 would be my class (the current 1Ls) and so forth.JDHopeful wrote:nice, thanks... now for the 25/median/75 splits, it says "class of 2009," does that mean the class that entered in 2009, or the class graduating in 2009?
All this is true but it is worth noting that LSN is clearly a biased sample. It is a small pool of applicants and the sorts of people who use the site are likely to have certain tendencies.romothesavior wrote:I'd say very. While you might have a few people lying here and there, it is pretty easy to see where the hard cutoffs are and it shows you the general acceptance chances for LSAT/GPA combos.JDHopeful wrote:exactly how accurate it LSN, btw?
The Real Jack McCoy wrote:All this is true but it is worth noting that LSN is clearly a biased sample. It is a small pool of applicants and the sorts of people who use the site are likely to have certain tendencies.romothesavior wrote:I'd say very. While you might have a few people lying here and there, it is pretty easy to see where the hard cutoffs are and it shows you the general acceptance chances for LSAT/GPA combos.JDHopeful wrote:exactly how accurate it LSN, btw?
The most obvious is that there is going to be a bias towards acceptances. For instance, for a school like Harvard, I think that green area in the upper right corner would have more rejections (proportionally) if it reflected the total population. A lot of people don't like reporting that they did worse than other applicants with similar numbers. Just a theory based on who is most likely to use LSN and who is most likely to report. But it is clear that some people just "disappear" once their cycle starts going bad.
For this reason you'd have to lean towards trusting LSP, which uses admissions indexes, which aren't skewed by faulty data samples, such as in LSN. Or is the data from LSP taken mainly from LSN?The Real Jack McCoy wrote:All this is true but it is worth noting that LSN is clearly a biased sample. It is a small pool of applicants and the sorts of people who use the site are likely to have certain tendencies.romothesavior wrote:I'd say very. While you might have a few people lying here and there, it is pretty easy to see where the hard cutoffs are and it shows you the general acceptance chances for LSAT/GPA combos.JDHopeful wrote:exactly how accurate it LSN, btw?
The most obvious is that there is going to be a bias towards acceptances. For instance, for a school like Harvard, I think that green area in the upper right corner would have more rejections (proportionally) if it reflected the total population. A lot of people don't like reporting that they did worse than other applicants with similar numbers. Just a theory based on who is most likely to use LSN and who is most likely to report. But it is clear that some people just "disappear" once their cycle starts going bad.
LSP uses LSN data.AP-375 wrote:For this reason you'd have to lean towards trusting LSP, which uses admissions indexes, which aren't skewed by faulty data samples, such as in LSN. Or is the data from LSP taken mainly from LSN?The Real Jack McCoy wrote:All this is true but it is worth noting that LSN is clearly a biased sample. It is a small pool of applicants and the sorts of people who use the site are likely to have certain tendencies.romothesavior wrote:I'd say very. While you might have a few people lying here and there, it is pretty easy to see where the hard cutoffs are and it shows you the general acceptance chances for LSAT/GPA combos.JDHopeful wrote:exactly how accurate it LSN, btw?
The most obvious is that there is going to be a bias towards acceptances. For instance, for a school like Harvard, I think that green area in the upper right corner would have more rejections (proportionally) if it reflected the total population. A lot of people don't like reporting that they did worse than other applicants with similar numbers. Just a theory based on who is most likely to use LSN and who is most likely to report. But it is clear that some people just "disappear" once their cycle starts going bad.
Your snark factor pleases me.
red law wants more picsrad law wrote:Your snark factor pleases me.
There was a first set? Sorry I missed it.Ragged wrote:red law wants more picsrad law wrote:Your snark factor pleases me.
I meant the avatar being the first, assuming thats of self.rad law wrote:There was a first set? Sorry I missed it.Ragged wrote:red law wants more picsrad law wrote:Your snark factor pleases me.
I dunno if I buy that. For one thing, it is not that small of a pool. Some schools have 1,000 points of data or more each year. Second of all, most applicants apply to safeties, targets, and reaches, which means they are going to have a mixture of acceptances, waitlists, and rejections. So there may be a bias, but I don't think it is as pronounced as you two make it seem.whymeohgodno wrote:The Real Jack McCoy wrote:All this is true but it is worth noting that LSN is clearly a biased sample. It is a small pool of applicants and the sorts of people who use the site are likely to have certain tendencies.romothesavior wrote:I'd say very. While you might have a few people lying here and there, it is pretty easy to see where the hard cutoffs are and it shows you the general acceptance chances for LSAT/GPA combos.JDHopeful wrote:exactly how accurate it LSN, btw?
The most obvious is that there is going to be a bias towards acceptances. For instance, for a school like Harvard, I think that green area in the upper right corner would have more rejections (proportionally) if it reflected the total population. A lot of people don't like reporting that they did worse than other applicants with similar numbers. Just a theory based on who is most likely to use LSN and who is most likely to report. But it is clear that some people just "disappear" once their cycle starts going bad.
The problem isn't small sample size in itself. The problem is that it isn't a random distribution.romothesavior wrote:
I dunno if I buy that. For one thing, it is not that small of a pool. Some schools have 1,000 points of data or more each year. Second of all, most applicants apply to safeties, targets, and reaches, which means they are going to have a mixture of acceptances, waitlists, and rejections. So there may be a bias, but I don't think it is as pronounced as you two make it seem.
Also, why do people use LSN? Just to get a general idea of how competitive they are and what kind of money they can expect. LSN is GREAT for that, far better than any other resource.
Look at this graph and tell me what the magic LSAT score is: http://washu.lawschoolnumbers.com/stats/0910/ (167)
Look at this one and tell me if this school takes many splitters: http://vanderbilt.lawschoolnumbers.com/stats/0910/ (No)
Does this school even have a 1L class, or did they waitlist and YP everyone and their moms?: http://uva.lawschoolnumbers.com/stats/0910/ (No 1Ls)
LSN always gets knocked for being "biased" and too small of a sample size, but neither of those are really true.
For some schools though where the LSAT+GPA medians cross it's goes from yellow/green to almost all green.The Real Jack McCoy wrote:The problem isn't small sample size in itself. The problem is that it isn't a random distribution.romothesavior wrote:
I dunno if I buy that. For one thing, it is not that small of a pool. Some schools have 1,000 points of data or more each year. Second of all, most applicants apply to safeties, targets, and reaches, which means they are going to have a mixture of acceptances, waitlists, and rejections. So there may be a bias, but I don't think it is as pronounced as you two make it seem.
Also, why do people use LSN? Just to get a general idea of how competitive they are and what kind of money they can expect. LSN is GREAT for that, far better than any other resource.
Look at this graph and tell me what the magic LSAT score is: http://washu.lawschoolnumbers.com/stats/0910/ (167)
Look at this one and tell me if this school takes many splitters: http://vanderbilt.lawschoolnumbers.com/stats/0910/ (No)
Does this school even have a 1L class, or did they waitlist and YP everyone and their moms?: http://uva.lawschoolnumbers.com/stats/0910/ (No 1Ls)
LSN always gets knocked for being "biased" and too small of a sample size, but neither of those are really true.
And it is pretty clear that the sample is, in fact, biased. You only have to look at the LSN medians vs. school medians to see that.
All I can say to this is that my cycle went pretty much exactly as LSN told me it would, and it did the same for almost every other person I know. I think the only time an applicant's experience doesn't match up is when you have something great or something terrible in your background (criminal record, great work experience, etc.) or you had something in your application that blew away the committee, and those are usually the random red dots in the midst of green or the random green in the midst of red.The Real Jack McCoy wrote:The problem isn't small sample size in itself. The problem is that it isn't a random distribution.
And it is pretty clear that the sample is, in fact, biased. You only have to look at the LSN medians vs. school medians to see that.
