16.9% Decrease In October Test Takers(Detailed Stats Inside) Forum

[KevinP]

Here's some more (semi)useless analysis.

DATA:
For the 2009-2010 cycle:
June: Mean = 151.68, SD = 10.51
October: Mean = 152.29, SD = 152.29
December: Mean = 150.11, SD = 9.92
February: Mean = 148.07, SD = 10.06

Quoted directly from LSAC's research report for 2003-2004 to 2009-2010: "The overall distribution of LSAT scores has remained fairly constant."

Quoted directly from LSAC'S research report on repeat takers for 2003-2004 to 2009-2010: "On average, of first-time test takers, 86.3% tested in June, 79.6% tested in September/October, 65.9% tested in December, and 66.0% tested in February. The actual percentages and number of first-time and repeat test takers for each administration are provided in Tables A-5a and Table A-5b, respectively."

However, this statement may be misleading because the percentage of first-time test takers has changed noticeably in 2006/2007 due to the ABA requirement that law schools report the highest score; therefore, looking at the percentage of first-time test takers for 2009-2010 will probably yield a more representative sample.

Percentage of first-time test takers:
June 2009: 82.5%
October 2009: 75.9%
December 2009: 59.9%
February 2010: 58.6%

Here's the percentile for a 173 for the various LSAT years:
2003-2006: 99.1%
2006-2009: 99.0%
2007-2010: 99.0%
2008-2011: 99.0%

KP'S ANALYSIS:
So far, the decrease in test takers this cycle has been from a pool that historically has the best scorers and the largest percentage of first-time test takers. Moreover, the percentiles for a 173+ have remained relatively stable. Even though the percentiles are reported in 3-year aggregates, the June 2011 and October 2011 administrations should have a slight impact on the aggregate percentiles if the distribution was completely skewed, but this does not appear to be the case. Hence, I expect this result to be especially significant for schools with large class sizes, particularly in the T6--Harvard, Columbia, and NYU.

Looking at last year's data, a large portion of the decrease in test takers was from the February and December LSAT administrations, which both historically have the worst pool of test takers and the greatest percentage of retakes. Based on a post from someone who attended NYU's orientation, NYU's median LSAT fell to 171. By the looks of LSN (http://nyu.lawschoolnumbers.com/stats/1011/) they were rather aggressive in trying to protect their 172 median.

KP'S CONCLUSION (TL;DR):
As many others have already said, this cycle should be unpredictable, but I do think it should be slightly easier.

[KevinP]

Duke seemed to be really lenient at first but have gotten a bit more strict recently, which is to be expected I suppose. Very small sample sizes, but NYU seems to be a little more lenient on GPA's so far too, and Penn hasn't been any stricter (for ED, anyway).

That's true, it's really hard to predict anything so early in the game.

[Jeffort]

Right! I think that there really isn't a very large discrepancy between what the averages for three years are and what the averages for the individual tests are. If there are some discrepancies, I'm sure they get averaged out over the year. (Lots of 170+ in June, not as many in October). So I don't think my reasoning is "flawed."

Well, it is. Your understanding and interpretation of the statistics involved is flawed, especially regarding how the LSAT is designed and scored.

Assuming that the three year range percentile rank stats of achieved scores holds true for each population of test takers (for each particular administration/particular test form) is a big mistake. It does not, it fluctuates from test to test.

To get the available data and reports LSAC has published about this and many other related issues about the test you would have to go on a deep research mission into the resources area of the LSAC site plus into several hardcore psychometric/math and other academic journals and read hundreds or thousands of pages of complex stuff. There are thousands of pages of available research reports LSAC has published over the years about the test that contain data, analysis and studies about tons of topics regarding the LSAT.

One issue I think you might have is misconception of is how the percentile ranks for each scaled score come about. It is not a forced curve. The percentile rank data is simply the byproduct of how test takers have performed, meaning that it is historical data about the distribution of students performance on the previous three years of tests taken under normal conditions that resulted in a reported score. They do not engineer the test-form scales to force a normal bell curve distribution of achieved scores, it just happens naturally from how people perform. That is why the percentile rank of each given scaled score changes slightly (up or down) from test to test.

In UG classes and many other school classes a bell curve distribution of grades is forced onto particular populations of test takers. Not so with the LSAT. Each scaled score is meant to represent a particular ability/skill/performance level that is comparable to everyone else that took an administered LSAT in the last three years.

[iamrobk]

Duke seemed to be really lenient at first but have gotten a bit more strict recently, which is to be expected I suppose. Very small sample sizes, but NYU seems to be a little more lenient on GPA's so far too, and Penn hasn't been any stricter (for ED, anyway).

That's true, it's really hard to predict anything so early in the game.

Yeah. I do agree with your analysis from your post right above, that it'll definitely be at least a little bit easier (and I do think we're starting to see this a little bit; just look at LSN for long enough and you'll see some people being accepted this cycle that probably wouldn't have been in the last 2-3 years). But in another month or 2 we'll definitely be able to say a lot more imo.

[Jeffort]

Another important thing to keep in mind is that since applicant and applications volumes have declined, admission committees are probably paying more attention to and giving more consideration to soft factors when making admission decisions. It makes sense simply from the standpoint that they don't have to sort through and read as many applications as they did in recent cycles and therefore have more time to read and think about that stuff with each application. I believe that means that having your personal statement, LORs and other stuff nice and polished is more important now and can better help admission chances beyond just the statistical numbers chances.

[mijenks]

Another important thing to keep in mind is that since applicant and applications volumes have declined, admission committees are probably paying more attention to and giving more consideration to soft factors when making admission decisions. It makes sense simply from the standpoint that they don't have to sort through and read as many applications as they did in recent cycles and therefore have more time to read and think about that stuff with each application. I believe that means that having your personal statement, LORs and other stuff nice and polished is more important now and can better help admission chances beyond just the statistical numbers chances.

Or softs are less important because admissions will admit whatever numbers they need to make their medians, regardless of softs. Not sure which way it will end up going . . .

[AreJay711]

Another important thing to keep in mind is that since applicant and applications volumes have declined, admission committees are probably paying more attention to and giving more consideration to soft factors when making admission decisions. It makes sense simply from the standpoint that they don't have to sort through and read as many applications as they did in recent cycles and therefore have more time to read and think about that stuff with each application. I believe that means that having your personal statement, LORs and other stuff nice and polished is more important now and can better help admission chances beyond just the statistical numbers chances.

Or softs are less important because admissions will admit whatever numbers they need to make their medians, regardless of softs. Not sure which way it will end up going . . .

Yeah, schools probably could keep their medians the same by using spliters and lowering their 25th and 75th percentiles.

[Samara]

Another important thing to keep in mind is that since applicant and applications volumes have declined, admission committees are probably paying more attention to and giving more consideration to soft factors when making admission decisions. It makes sense simply from the standpoint that they don't have to sort through and read as many applications as they did in recent cycles and therefore have more time to read and think about that stuff with each application. I believe that means that having your personal statement, LORs and other stuff nice and polished is more important now and can better help admission chances beyond just the statistical numbers chances.

Or softs are less important because admissions will admit whatever numbers they need to make their medians, regardless of softs. Not sure which way it will end up going . . .

Yeah, softs will almost certainly be deemphasized as schools have less leniency to maintain their medians. I actually think this means you should put more work into your application if you are a borderline candidate because you will have to stand out more for your softs to matter.

[WhiteGuy5]

Another important thing to keep in mind is that since applicant and applications volumes have declined, admission committees are probably paying more attention to and giving more consideration to soft factors when making admission decisions. It makes sense simply from the standpoint that they don't have to sort through and read as many applications as they did in recent cycles and therefore have more time to read and think about that stuff with each application. I believe that means that having your personal statement, LORs and other stuff nice and polished is more important now and can better help admission chances beyond just the statistical numbers chances.

Or softs are less important because admissions will admit whatever numbers they need to make their medians, regardless of softs. Not sure which way it will end up going . . .

Correct. Just because the total number of applications are down doesn't mean adcomms have a lesser workload. In fact, it seems like fewer applicants are applying to more schools = more work for adcomms. Quite the opposite!

[Jeffort]

Another important thing to keep in mind is that since applicant and applications volumes have declined, admission committees are probably paying more attention to and giving more consideration to soft factors when making admission decisions. It makes sense simply from the standpoint that they don't have to sort through and read as many applications as they did in recent cycles and therefore have more time to read and think about that stuff with each application. I believe that means that having your personal statement, LORs and other stuff nice and polished is more important now and can better help admission chances beyond just the statistical numbers chances.

Or softs are less important because admissions will admit whatever numbers they need to make their medians, regardless of softs. Not sure which way it will end up going . . .

Correct. Just because the total number of applications are down doesn't mean adcomms have a lesser workload. In fact, it seems like fewer applicants are applying to more schools = more work for adcomms. Quite the opposite!

How so? Please explain your reasoning.

For many years tier 1 schools have consistently been receiving far more applications from numerically qualified applicants than they could admit and have been having to deny admission to many qualified applicants that would have kept their medians up to par because of class size limits.

I really don't want to spend the time on it tonight, but it wouldn't take more than a few hours to tabulate the entering class sizes of all tier 1 law schools from the last cycle and compare the number to the approximate number of people that have achieved LSAT scores in their number ranges and applied during the last few LSAT annual cycles based on LSAC data available in the research reports.

Many people with 175+ (even 180) scores and near or actually perfect 4.0+ UGPA have been denied admission by several tier 1 schools they applied to in the last few cycles. I know of several cases of people with a 175-180 and near perfect GPA that got denied admission by every T14 school they applied to.

Bottom line, top law schools are not going to have to lower their admission standards to build full entering classes. It's the T3 and below schools that are probably going to have to lower the bar to fill the seats.

[ahnhub]

Many people with 175+ (even 180) scores and near or actually perfect 4.0+ UGPA have been denied admission by several tier 1 schools they applied to in the last few cycles. I know of several cases of people with a 175-180 and near perfect GPA that got denied admission by every T14 school they applied to.

What!? Was the only school they applied to Yale?

Or did they only apply to hypothetical fantasy law schools because they thought they were too good for Yale?

[WhiteGuy5]

I know of several cases of people with a 175-180 and near perfect GPA that got denied admission by every T14 school they applied to.

Umm...do you understand how ridiculous this statement sounds?

[180asBreath]

C'mon.

Several 3.9-4.33/175-180's got shut out of the T14?

Oh, "that they applied to". The people probably applied to just H or just Y or just HYS; it's not like they blanketed the T14 and got shut out.

[Jeffort]

Many people with 175+ (even 180) scores and near or actually perfect 4.0+ UGPA have been denied admission by several tier 1 schools they applied to in the last few cycles. I know of several cases of people with a 175-180 and near perfect GPA that got denied admission by every T14 school they applied to.

What!? Was the only school they applied to Yale?

Or did they only apply to hypothetical fantasy law schools because they thought they were too good for Yale?

Oh common, don't be a jackwagon, this is a serious discussion.

[Jeffort]

I know of several cases of people with a 175-180 and near perfect GPA that got denied admission by every T14 school they applied to.

Umm...do you understand how ridiculous this statement sounds?

Sure, it sounds crazy. However, I've been in the LSAT teaching, tutoring and admission consulting biz for over 10 years. I'm not talking out of my arse.

Some of the cases were applicants that were total a-holes that submitted crap a$$ personal statements and LOR's, some had a misconduct report(s) of some type in their records (LSAC or UG school records) and/or something that they had to/were supposed to disclose in applications (i.e. criminal convictions) or a combination of some of these things. Rejections of top numbers applicants by Tier 1 schools happen every year for various reasons.

Lower ranked schools gladly took them and gave some of them $$ though, probably figuring 'Fck it, let the students and the state bar C&F review people deal with the been bad student issues in a few years, these admits will really help our stats.'

Not all of the cases were because of mischief, some were just because of crappy personal statements, LORs, and overall crappy application packages with everything except the numbers. Also, some people with great numbers are rejected by top schools every year because of dishonesty in the admission process (they lied about some stuff and it was detected).

[paradox]

I just joined the site so I am unfamiliar with some of the jargon.

What is a "splitter".

Thanks

[Jeffort]

I just joined the site so I am unfamiliar with some of the jargon.

What is a "splitter".

Thanks

High LSAT score, low UGPA or reversed, low LSAT, high GPA.

[NYCLSATTutor]

When trying to interpret the LSAC statistical data it's important to realize that the LSATs Administered chart, even the annual total for each testing cycle is a count of number of tests administered. People that retake the test in the same testing year are counted twice or three times for the number of times they took the test.

For the last several years there has been an increase of serial test takers that took the test multiple times, especially since it was super easy to get a waiver exception to the three times in two years limit.

For about two years LSAC had a policy that if you didn't cancel or postpone your LSAT registration roughly two weeks before test day you were faced with the choice of either not showing up and getting an absent mark on your CAS report or taking it and either keeping or canceling the score. Once the deadlines to cancel or postpone your registration for the administration passed, no matter what you did, something would show on your CAS score report so many people showed up and took the test even when they knew they weren't ready in order to avoid an absent mark or just to give it a shot and hope to get lucky.

Shortly before the June 2011 LSAT LSAC changed the policy so that you now can withdraw your registration up to midnight the day before the test, nothing will show on your CAS report and it doesn't count towards the 3 times in 2 years limit. Many students that didn't feel prepared or for whatever reason have exercised this option. Also, during the two year period (It might have been three years, but I'm pretty sure it was two) without the withdraw the night before option, many students decided to not show up on test day when they weren't ready or something and went with the absent mark option, thus not getting counted in the LSATs administered data, and then took the test sometime later.

There are plenty of people with two, three, four and even five test administrations noted on their CAS report (score, cancel, absent, etc.) ranging from over the last two years up to the last five years with some that took the LSAT again this testing year.

Serial test takers/retakers have accounted for a significant amount of the tests administered record volumes over the last two-three years.
Here's the current repeater report:
http://www.lsac.org/LSACResources/Data/ ... erData.pdf
33.8 percent of people that took the LSAT last year took it more than once.

This would only have an impact on people who take the test 3 times v. more than 3 times. The data you provided said that only 7% of people take it more than twice and that would likely mean even less would take it more than 3 times (if they could). Even if the 7% number remained the same, those 7% could not, by themselves, account for the 18% or so drop in admission standards. Further, from some anecdotal evidence, it seems highly highly unlikely that the people taking the test more than 3 times are ever scoring in the 170's.

[NYCLSATTutor]

FWIW, outside of Y/S, schools have already been taking the highest score, regardless of what the schools claim. Furthermore, since law school is already primarily a numbers game, softs rarely make or break a candidate.

After looking at LSN, it seems like Duke and GULC are being more lenient GPA-wise for test scores who are in the 170s.

What evidence do you have to support this? This is what I would expect from schools, but anecdotally it seems as if its a mixed bag. In the NYC area there are several schools who are not T14 who say they average who seem to be actually averaging scores (at least they did last year).

[Jeffort]

This would only have an impact on people who take the test 3 times v. more than 3 times. The data you provided said that only 7% of people take it more than twice and that would likely mean even less would take it more than 3 times (if they could). Even if the 7% number remained the same, those 7% could not, by themselves, account for the 18% or so drop in admission standards. Further, from some anecdotal evidence, it seems highly highly unlikely that the people taking the test more than 3 times are ever scoring in the 170's.

My comments and analysis does not just focus on students that took the test 3 or more times, it's partially about all people that have taken the test more than once in the last few years and how that has affected the LSATs administered volume summary. Plus, the retake data chart only covers one testing year.

ITT we are mainly talking about the percentage declines of LSATs administered and applications submitted in the last few years as it relates to admission chances/competition to get into top law schools. Read back in the thread a bit, you are missing pieces of the relevant statistical data in forming your conclusion. You have to look at more than just one or two pieces of the data and put it all into context to get a better view of the full picture.

RE:

could not, by themselves, account for the 18% or so drop in admission standards

Huh? Where did that come from?

[NYCLSATTutor]

This would only have an impact on people who take the test 3 times v. more than 3 times. The data you provided said that only 7% of people take it more than twice and that would likely mean even less would take it more than 3 times (if they could). Even if the 7% number remained the same, those 7% could not, by themselves, account for the 18% or so drop in admission standards. Further, from some anecdotal evidence, it seems highly highly unlikely that the people taking the test more than 3 times are ever scoring in the 170's.

My comments and analysis does not just focus on students that took the test 3 or more times, it's partially about all people that have taken the test more than once in the last few years and how that has affected the LSATs administered volume summary. Plus, the retake data chart only covers one testing year. ITT we are talking about the percentage declines of LSATs administered going back three years. Read back in the thread a bit, you are missing pieces of the relevant statistical data in forming your conclusion. You have to look at more than just one or two pieces of the data to put it into context to get a better view of the full picture.

RE:

could not, by themselves, account for the 18% or so drop in admission standards

Huh? Where did that come from?

You were claiming that the fact that LSAC changed its rules to disallow taking the test more than 3 times would change the stats, yes? How would that change anything regarding students who never bumped up against that ceiling in the first place (i.e. have taken it 3 times already).

Also...yeah I shouldn't post when somewhat sleep-deprived apparently haha. I meant LSAT test-takers, not admission standards. And yes the 18% is an approximate number.

[Jeffort]

You were claiming that the fact that LSAC changed its rules to disallow taking the test more than 3 times would change the stats, yes? How would that change anything regarding students who never bumped up against that ceiling in the first place (i.e. have taken it 3 times already).

Also...yeah I shouldn't post when somewhat sleep-deprived apparently haha. I meant LSAT test-takers, not admission standards. And yes the 18% is an approximate number.

No, that is not what I was addressing if you were referring to the future, but yes, the recent policy changes will likely affect the future stats. I've been talking about the stats regarding the past tests/cycles and how the various LSAC policy changes over the last several years have affected student behavior and in turn the volume data and other related stats.

No worries about your mistake in wording dude, get some sleep, have some eggs or pizza or a good hotdog, and if you have time to kill, re-read some of this stuff if you are interested in stats and stuff.

It's hard to hold all the data in your brain and analyze/put it together all at once without your brain having been trained about statistics and/or having a natural aptitude for such thinking. In UG when I took the hard core statistics classes (the ones everyone hates and doesn't take unless they have to) I loved it and easily aced the quizzes and finals. I looked forward to that class every day. When in LS I took a "quantitative methods and the law" elective course. Only ~15 people enrolled in it and by the deadline to drop the class only 6 remained.

I loved that class, it was one of the most enjoyable ones I took and easily rocked it on the final.
We spent a bunch of class time and homework time analyzing LSAT data, UG GPA data, 1L first year performance data, admissions stats data, etc. The prof used all those data sets to teach some of the advanced statistics and then how they related to law (he had data sets that were/are not publicly available he gave us to analyze). It was pure geek out awesomeness!

One of the things we focused on for a while was the correlation between LSAT scores and first semester 1L grades (the class was when I was a 2L). The professor put up a correlation scatter chart graph of our class group stats with LSAT and first semester 1L grades as the X & Y and others with the same recent data from other law schools. Through the analysis of the multiple data sets we noticed something very interesting, especially as it relates to outliers. In all the data sets we noticed a drop off of grades (in comparison to the class group mean/median/mode grades) with people that had scored close to perfect on the LSAT!

When the proff put up the scatter plot correlation graph of our first semester grades compared to LSAT scores, I noticed a dot way over to the right (on the LSAT X axis) that was below the going diagonally up to the right on the Y axis cluster of dots. Since I had tanked (badly) one final first semester I blurted out "Hey that dot is me!", the professor laughed and said yes, it is.

[barneytrouble]

I loved that class, it was one of the most enjoyable ones I took and easily rocked it on the final.
We spent a bunch of class time and homework time analyzing LSAT data, UG GPA data, 1L first year performance data, admissions stats data, etc. The prof used all those data sets to teach some of the advanced statistics and then how they related to law (he had data sets that were/are not publicly available he gave us to analyze). It was pure geek out awesomeness!

One of the things we focused on for a while was the correlation between LSAT scores and first semester 1L grades (the class was when I was a 2L). The professor put up a correlation scatter chart graph of our class group stats with LSAT and first semester 1L grades as the X & Y and others with the same recent data from other law schools. Through the analysis of the multiple data sets we noticed something very interesting, especially as it relates to outliers. In all the data sets we noticed a drop off of grades (in comparison to the class group mean/median/mode grades) with people that had scored close to perfect on the LSAT!

When the proff put up the scatter plot correlation graph of our first semester grades compared to LSAT scores, I noticed a dot way over to the right (on the LSAT X axis) that was below the going diagonally up to the right on the Y axis cluster of dots. Since I had tanked (badly) one final first semester I blurted out "Hey that dot is me!", the professor laughed and said yes, it is.

Well I think you would expect that wouldn't you... Few people are capable of "earning" 180s on test day. Most super smart people fall somehwere in between the 173-177 band I would imagine. The difference between 177 and 180 is a few lucky breaks here and there. So 180s would be more likely to get 1L grades in line with those who scored 177. Similar to how a 7 foot tall person will likely have a taller than average kid, but still shorter than them. Regressing to the mean and all that.

[Jeffort]

I loved that class, it was one of the most enjoyable ones I took and easily rocked it on the final.
We spent a bunch of class time and homework time analyzing LSAT data, UG GPA data, 1L first year performance data, admissions stats data, etc. The prof used all those data sets to teach some of the advanced statistics and then how they related to law (he had data sets that were/are not publicly available he gave us to analyze). It was pure geek out awesomeness!

One of the things we focused on for a while was the correlation between LSAT scores and first semester 1L grades (the class was when I was a 2L). The professor put up a correlation scatter chart graph of our class group stats with LSAT and first semester 1L grades as the X & Y and others with the same recent data from other law schools. Through the analysis of the multiple data sets we noticed something very interesting, especially as it relates to outliers. In all the data sets we noticed a drop off of grades (in comparison to the class group mean/median/mode grades) with people that had scored close to perfect on the LSAT!

When the proff put up the scatter plot correlation graph of our first semester grades compared to LSAT scores, I noticed a dot way over to the right (on the LSAT X axis) that was below the going diagonally up to the right on the Y axis cluster of dots. Since I had tanked (badly) one final first semester I blurted out "Hey that dot is me!", the professor laughed and said yes, it is.

Well I think you would expect that wouldn't you... Few people are capable of "earning" 180s on test day. Most super smart people fall somehwere in between the 173-177 band I would imagine. The difference between 177 and 180 is a few lucky breaks here and there. So 180s would be more likely to get 1L grades in line with those who scored 177. Similar to how a 7 foot tall person will likely have a taller than average kid, but still shorter than them. Regressing to the mean and all that.

Yep, regression to the mean.

Part of the reason I tanked that first semester 1L final was because I was cocky due to my LSAT score since it was one of the highest in the entering class. I got punished and slapped down to humility by the grade due to that arrogance. I didn't really like the class or get a lot out of the lectures because it was early in the morning (8:20 am start) and I half-a$$ed it (slept through parts of it with my head on the desk in the textbook so it looked like I was reading). I got a spanking when grades were posted and learned my lesson.

[WhiteGuy5]

So...now that we've gotten a nice batch of acceptances/denials. Does anyone see anything out of the ordinary with this cycle?