Law School apps are down... what does it mean? Forum

[NorCalBruin]

Has anyone tried to estimate the level of self-selection bias on LSN by comparing the LSN stats with the U.S. News/ABA stats? It would take a little time, but it seems pretty straightforward to compare the LSAT and GPA percentiles of the T14 on LSN for the past few years with the percentiles reported in U.S. News. My assumption is that the stats are significantly higher on LSN since it's populated with the more neurotic OCD types among us...

This is an interesting idea. However, I think the only thing you could really get out of it would be the ability to compare LSN's acceptance rates (%) to the ABA's and USNEWS acceptance rates, namely because schools (and the aba and usnews) don't release data on, say, the average LSAT of the people who apply, only the people who matriculate. Yea, some people on LSN do end up revealing where they attend, but it's probably too rare and infrequent to be useful.

[jeremysen]

fellas, LSN is <<<10% of the actual applicant number...


http://www.youtube.com/watch?v=OQSNhk5ICTI

[NorCalBruin]

fellas, LSN is <<<10% of the actual applicant number...

I don't know if the extra "<'s" were supposed to mean "way less than", but, at least for the top schools, LSN contains probably around 5-10% of the applicants for each schools. And even if it's less than that (say 2%, for example), most top schools still have several hundred data points, enough to draw conclusions about trends in admissions/applications. It's extremely obvious, for example, that Emory had an LSAT floor of 166 last year, and a floor of 165 the year before that.

In other words, the amount of people (and thus data) on LSN is more than enough to draw certain conclusions. There are PLENTY of data points, especially for top schools. How self-selecting LSN is, or how prone to error it is, well, that's another matter entirely.

[JeremyPaul]

exactly. the self-report bias and the self-selection bias is what i'm worried about. i've been stressing myself out by looking at LSN's LSAT x GPA scatter plots and then trying to chill myself out thinking that there's an upward bias on who self-selects into posting on LSN. i'm especially preoccupied with stanford, which is probably one of the least valid plots given their low enrollment numbers.

[javancho]

Has anyone tried to estimate the level of self-selection bias on LSN by comparing the LSN stats with the U.S. News/ABA stats? It would take a little time, but it seems pretty straightforward to compare the LSAT and GPA percentiles of the T14 on LSN for the past few years with the percentiles reported in U.S. News. My assumption is that the stats are significantly higher on LSN since it's populated with the more neurotic OCD types among us...

I have run those stats for some schools, and what I have found is consistent from school to school. First, LSNers for each school REPORT a bit higher LSATs and GPAs than the school median (self-selection of the most obsessed applicants you might call it). In return, schools ACCEPTS applicants with higher LSATs and GPAs than the reported medians for each school. The medians of ACCEPTED vs medians of the school is about 2-4 points higher for the LSAT, for instance. Finally, for the applicants accepted that actually decide to ENROLL, the data indicates that their medians are very close to the medians reported by the schools.

So, LSNers are better than the typical applicant, but what really matters is the response of the schools to the LSNers cohorts. With that data, one can easily find the chances of getting in, at least just by using GPA, LSAT, URM status, and assuming the historical trends for each school remain stable.

[NorCalBruin]

Has anyone tried to estimate the level of self-selection bias on LSN by comparing the LSN stats with the U.S. News/ABA stats? It would take a little time, but it seems pretty straightforward to compare the LSAT and GPA percentiles of the T14 on LSN for the past few years with the percentiles reported in U.S. News. My assumption is that the stats are significantly higher on LSN since it's populated with the more neurotic OCD types among us...

I have run those stats for some schools, and what I have found is consistent from school to school. First, LSNers for each school REPORT a bit higher LSATs and GPAs than the school median (self-selection of the most obsessed applicants you might call it). In return, schools ACCEPTS applicants with higher LSATs and GPAs than the reported medians for each school. The medians of ACCEPTED vs medians of the school is about 2-4 points higher for the LSAT, for instance. Finally, for the applicants accepted that actually decide to ENROLL, the data indicates that their medians are very close to the medians reported by the schools.

So, LSNers are better than the typical applicant, but what really matters is the response of the schools to the LSNers cohorts. With that data, one can easily find the chances of getting in, at least just by using GPA, LSAT, URM status, and assuming the historical trends for each school remain stable.

This makes sense. Schools are going to accept a lot more people above their medians, than below their medians because they know that people who get into their "reaches" tend to matriculate much more often than those who applicants who might consider their school a "safety".

[javancho]

I just wanted to add that the google trends data for January came out this morning for LSAC.ORG (http://trends.google.com/websites?q=lsa ... a=N&geo=US) and it corroborates the alexa traffic data (http://www.alexa.com/siteinfo/lsac.org#).

I think these two up-to-date data indicate that LSAC.ORG traffic and very likely applicants are in a steep decline this cycle vs the last cycle, particularly AFTER December 2010. I am going to bet that when this cycle is over the decrease in applicants is going to be much steeper than the -12.5% announced with early December 2010 data.

[NorCalBruin]

I just wanted to add that the google trends data for January came out this morning for LSAC.ORG (http://trends.google.com/websites?q=lsa ... a=N&geo=US) and it corroborates the alexa traffic data (http://www.alexa.com/siteinfo/lsac.org#).

I think these two up-to-date data indicate that LSAC.ORG traffic, and very likely applicants, are in a steep decline this cycle vs the last cycle, particularly AFTER December 2010, and I am going to bet that the decline is going to be much steeper than the -12.5% announced with early December 2010 data.

Great stuff. Even if this doesn't mean anything (and I think it does mean something), it's still cool to see different, inventive methods for understanding this cycle.

For what it's worth, the article was from January 25th and was citing the total number of applicants as of January 14th, not early december.

[FiveSermon]

Hopefully medians drop/remain the same.

[duckmoney]

So for schools that pushed their median up a point last year (vandy, for instance, went 168 --> 169), are they going to be able to maintain it if there are fewer applicants with LSAT scores in that range?

[NorCalBruin]

So for schools that pushed their median up a point last year (vandy, for instance, went 168 --> 169), are they going to be able to maintain it if there are fewer applicants with LSAT scores in that range?

Maybe. I think if we see something around a -12.5% drop in applicants that they will probably be able to maintain their LSAT medians. They might have to sacrifice their GPA median a few .00 points. (Although they will try to avoid this as it is 10% of their ranking). They may use splitters to maintain their medians. They might have to use their scholarships and waitlists more proactively and strategically. But in the end, they will try hard to avoid their medians slipping, especially because their LSAT median constitutes the largest part of the ranking (12.5%) that they have serious control over.

Now if we see a truly huge drop in applicants, like a 20%+ drop (which would be surprising, but not impossible), then yea, I think you'll see schools like Vanderbilt slide back a point.

The important thing to remember is that this is all conjecture. We know that applications were down by mid-january. We now that the number of LSAT takers is down. We know that the number of LSN users appears much lower this year (maybe as much as 35-40%). We know that site traffic at LSAC.org is much lower than it was last year. But that's about all we know. And it's not clear exactly what that all means. What's hard to swallow is that we probably won't get anymore hard numbers until March, and probably later. By that time it won't mean much to speculate because most of our cycles will be over, or nearly over.

So all we can do is hope. We can hope that applications are not just down, but tumbling down into a pit of nothingness. Down so far that we hit below 70,000 applicants. We can hope that the February LSAT is down more than 25% and this cycle doesn't have the second most amount of test takers, but the third or the fourth. If those things happen, good things are bound to happen to us. If not, well then it's time to buckle up.

[red_alertz]

pardon my ignorance, what is a "splitter?"

[kapachino]

The economy is growing enough so that people don't have to resort to going to grad school, or they're wising up and deciding against entering a crowded market.

[kapachino]

pardon my ignorance, what is a "splitter?"

Someone with a low GPA and a high LSAT score.

[helloperson]

fellas, LSN is <<<10% of the actual applicant number...


http://www.youtube.com/watch?v=OQSNhk5ICTI

First off, you only use two less than signs to indicate significantly smaller than.
Second, this is partially false. It's at or greater than 10% for some schools at/near the top 14. For instance, UCLA has 526 total applicants for this years cycle on LSN, and they received around 5800 applications in the 2009-2010 cycle.

Let's keep going and show that LSN over represents those with high LSAT scores... For UCLA, 218 of the applicants on LSN have LSAT numbers at or above UCLAs 75th percentile LSAT (169). That's more than *half* of the sample on LSN that has LSAT numbers above the 75th percentile. UCLAs matriculate rate is right about 1/3. To illustrate a point, let us consider that UCLA grants acceptance to around half of people with LSAT numbers above their 75th percentile (congruent with last year's LSN numbers), and these people matriculate at lower rates than everyone else, around 1/5. So that's 25 people in a class of 526. 131 people in the incoming class have LSAT scores in the 75th percentile, and 25 of them are accounted for (~20%)

So, the numbers on LSN, which are probably not unlike the numbers of many users on this board, are skewed towards the top. If you figure that for people with LSATs above 169, there's about 20% representation on LSN, then it makes sense because: it means around 1000 applicants with numbers at/above 169 are applying to UCLA. There are around 150,000 test takers each year, and around 2.5% of them score at/above 169. That's 3750 people, and instead of figuring that around half of those people apply each year to UCLA, you figure that only around a quarter of them do.

tl; dr: 10% is likely too low, at least for those applicants with high LSAT scores

[robotclubmember]

My prediction:
1) All law schools will still fill their classes easily and still have some left over on the waitlist.
2) The drop in applicants will not be enough to significantly improve anyone's chances of getting into a school if they don't have the numbers.
3) In other words: Everything remains nearly the same, if not the same. I kind of see it akin to the current employment environment--unemployment went down nearly 1% over the past two months but it is still hard as hell to find a job and no one's situation looks any brighter. I guess it all depends on how many of those 10,000+ applicants lost were T14 caliber students or T4/non-ABA credentialed students. If most of that applicant pool is part of the latter, top schools will be left relatively unscathed.

This is not analogous to employment at all. It's hard to find a job because there is a finite amount and they are taken. Law school has a finite number of seats but at the beginning of each cycle, none of them are taken. Supply of seats is the same but demand has fallen for those seats, whereas the demand for jobs is increasing.

[jeremysen]

fellas, LSN is <<<10% of the actual applicant number...


http://www.youtube.com/watch?v=OQSNhk5ICTI

First off, you only use two less than signs to indicate significantly smaller than.
Second, this is partially false. It's at or greater than 10% for some schools at/near the top 14. For instance, UCLA has 526 total applicants for this years cycle on LSN, and they received around 5800 applications in the 2009-2010 cycle.

Dont really need a stats lesson from someone who quotes 526 / 5800 as "at or greater than 10%" and compares LSN info from this (incomplete) year with last year's total # apps.

[jeremysen]

fellas, LSN is <<<10% of the actual applicant number...

I don't know if the extra "<'s" were supposed to mean "way less than", but, at least for the top schools, LSN contains probably around 5-10% of the applicants for each schools. And even if it's less than that (say 2%, for example), most top schools still have several hundred data points, enough to draw conclusions about trends in admissions/applications. It's extremely obvious, for example, that Emory had an LSAT floor of 166 last year, and a floor of 165 the year before that.

In other words, the amount of people (and thus data) on LSN is more than enough to draw certain conclusions. There are PLENTY of data points, especially for top schools. How self-selecting LSN is, or how prone to error it is, well, that's another matter entirely.

You're right on the number of "<"s. That was an exaggeration.

But let's see Emory's 09-10 data to address that "LSAT floor" you're talking about:

LSN: 282 accepts, 241 rejects
Actual: 1149 accepts, 3413 rejects

That's a huge difference, and considering the additional self selection bias, this is really unreliable information. A mere 50 more acceptances of people who have 164s would cast doubt on your LSAT floor - and the fact of the matter is that we have 867 acceptances unaccounted for. (assuming that 282 accepts all report accurate info)

[ilovemulch]

Very true about LSN. But, don't most users at least create the account and list where they intend to apply or are pending? I'm sure there are some that really don't create an account or list any data until the very end but really, I doubt that accounts for much. And it's already mid-Feb basically. All the other statistics match what LSN suggests. Applications are down.

I do think applications are slightly down this year, but I also know that a significant number of LSNers only report themselves after their cycle is over.

yeah, I'm afraid the latter part might be true...I am one of these applicants who created a profile at first then decided to take it down for privacy until the cycle is over, at which point I will probably make a public one.

[robotclubmember]

Post your data or go eat a piece of mulch.

[ilovemulch]

Post your data or go eat a piece of mulch.

i mean , it's in my TLS profile (well besides stats...but that's easily searchable)

[javancho]

fellas, LSN is <<<10% of the actual applicant number...

I don't know if the extra "<'s" were supposed to mean "way less than", but, at least for the top schools, LSN contains probably around 5-10% of the applicants for each schools. And even if it's less than that (say 2%, for example), most top schools still have several hundred data points, enough to draw conclusions about trends in admissions/applications. It's extremely obvious, for example, that Emory had an LSAT floor of 166 last year, and a floor of 165 the year before that.

In other words, the amount of people (and thus data) on LSN is more than enough to draw certain conclusions. There are PLENTY of data points, especially for top schools. How self-selecting LSN is, or how prone to error it is, well, that's another matter entirely.

You're right on the number of "<"s. That was an exaggeration.

But let's see Emory's 09-10 data to address that "LSAT floor" you're talking about:

LSN: 282 accepts, 241 rejects
Actual: 1149 accepts, 3413 rejects

That's a huge difference, and considering the additional self selection bias, this is really unreliable information. A mere 50 more acceptances of people who have 164s would cast doubt on your LSAT floor - and the fact of the matter is that we have 867 acceptances unaccounted for. (assuming that 282 accepts all report accurate info)

Actually, NorCalBruin is totally correct. You don't need all the data points to draw conclusions. With just a few, you have enough data.
Of course the applicants of LSN tend to have better credentials. After all, they are obsessed about law school. What really matters is how the school RESPONDS to the LSN applicant pool. From there you can draw conclusions about admission policies, even if you don't have all the data.

[helloperson]

fellas, LSN is <<<10% of the actual applicant number...


http://www.youtube.com/watch?v=OQSNhk5ICTI

First off, you only use two less than signs to indicate significantly smaller than.
Second, this is partially false. It's at or greater than 10% for some schools at/near the top 14. For instance, UCLA has 526 total applicants for this years cycle on LSN, and they received around 5800 applications in the 2009-2010 cycle.

Dont really need a stats lesson from someone who quotes 526 / 5800 as "at or greater than 10%" and compares LSN info from this (incomplete) year with last year's total # apps.

You didn't really address any of my arguments; you just picked a superficial way to discard them.

Considering the topic of the thread, 526 out of an applicant pool that is apparently down considerably from last years equates to roughly 10% of all applicants. Further, out of those 526, people with very high LSAT scores are over represented; about 20% of all applicants with those numbers are represented within the LSN data.

Way to miss the point, jackass.

[jeremysen]

You didn't really address any of my arguments; you just picked a superficial way to discard them.

Considering the topic of the thread, 526 out of an applicant pool that is apparently down considerably from last years equates to roughly 10% of all applicants. Further, out of those 526, people with very high LSAT scores are over represented; about 20% of all applicants with those numbers are represented within the LSN data.

Way to miss the point, jackass.

I got your point fool - your conclusion is intuitive - an idiot could figure that out by common sense. Obviously if acceptances are over represented, you'd find such results. Waste more of your time, tool.

Now, smartass, address my mine: if a guy is gonna make corrections on proper stats notation, shouldn't he address errors in his assumptions / calculations?
- Comparing this year's unfinished applicant pool to last year's total
- Not considering that about a fourth of this years' is "pending," and a ton on WL
- Possibility that UCLA's 75% LSAT score is the same as, on an extreme, its median +1
--> I can probably list more, but here's the conclusion if you're looking for it: you are an idiot


Also, did you really address my full argument? You limited your point to T14, but my remark is still correct for LSN as a whole. And you certainly glossed over my implicit conclusion.


I'm gonna spell it out because I believe it is now necessary for the denser kids like yourself:

The reason I made my first post on this thread was to speak to people who would come up with dumbass phrases/assertions, such as "XYZ school has an LSAT floor of ____," based on probably useful, but statistically unreliable LSN data.

[helloperson]

Well, you're still wrong, and your insults are somehow worse than your logical reasoning skills.