C/O 2016 median lsat/gpa/class size

(Applications Advice, Letters of Recommendation . . . )
Big Dog
Posts: 1191
Joined: Wed Dec 16, 2009 9:34 pm

Re: C/O 2016 median lsat/gpa/class size

Postby Big Dog » Tue Sep 24, 2013 7:43 pm

it may also indicate that Chicago is just more focused on GPA, and is willing to accept reverse splitters to hold it. Chicago's 50th is tied for highest with Yale @ 3.90. It's 25th is 4th highest, just below those GPA hoes at Boalt. Chicago's 75th is tied for second, two-hundredths under Yale, but significantly higher than CLS and NYU.

Moreover, Chicago's (and Stanford's) smaller class size enable them to pick and choose among high GPAs, particularly in contrast to those large classes in Manhattan.

User avatar
Monochromatic Oeuvre
Posts: 1929
Joined: Fri May 10, 2013 9:40 pm

Re: C/O 2016 median lsat/gpa/class size

Postby Monochromatic Oeuvre » Tue Sep 24, 2013 8:20 pm

Crowing wrote:I don't see a large 25-50 split as necessarily indicative of sketchy behavior.

E.g. In the case of CLS/NYU vs. Chicago, one could say that CLS/NYU bases decisions very heavily on numbers, while Chicago is highly likely to admit applicants above its medians but is less focused on numbers and more on other variables of an applicant if they have lower numbers or are splitters. I'm not saying this is definitely the case, but I do think that quartile splits are by themselves not meaningful enough to draw conclusions about admissions philosophy from.


If you are highly likely to admit applicants above your medians, you are almost by definition highly unlikely to admit applicants below them. The fact that someone below a median is accepted does not make a school "less focused on numbers" w/r/t to that applicant, it just means other factors were significant enough to compensate for such a weakness. As applicants are in competition with one another because admissions is a zero sum game, numbers cannot be more or less important for an applicant merely because of what those numbers are. UChi's non-URM, non-ED acceptance rates last cycle:

At/above both medians: 87%
At/above one median and below another: 50%
Below both medians: 9%

There is not a big mystery going on, and UChi is not a black box. 167, 168 and 169 have a decent shot if they have a 3.9+. 3.5-3.6ish have a shot with a 172-175. They dip WAY below their GPA median for applicants slightly above the LSAT median and vice versa. Under a sliding scale (non-manipulated) applicants who are slightly above one median and well below the other are almost always dinged. Not so with manipulators.

User avatar
rayiner
Posts: 6184
Joined: Thu Dec 11, 2008 11:43 am

Re: C/O 2016 median lsat/gpa/class size

Postby rayiner » Tue Sep 24, 2013 9:41 pm

Monochromatic Oeuvre wrote:A simple test of whether your medians are manipulated is this: Would someone at both medians be a student of average or above-average quality at a school?

For example, at NYU, someone with a 3.7/170 is roughly a student of average quality, if we're measuring that using an intuitive idea of quality or something like the index formula. A 3.75/170 at Northwestern would be a solidly above-average student there, because their medians are such that most of the 3.8+ students have 165-168 and a whole bunch of the 170+ students have 3.4 or lower.

A "sliding scale", like the one CLS uses, is not median manipulation. It is making admissions almost entirely-numbers based, but consideration is only given to meet preset numerical standards on that scale, not specifically on whether someone is above a median.


That doesn't make any sense, because of the way medians work. If you have 3.75/170 medians, and say 100 students, then at least 100 have a 3.75+ and at least 100 have a 170+. That means there is someone in the middle who has a 3.75/170. It's hard to say they're "solidly above average" because half the class has a higher LSAT score than they do, and half has a higher GPA.

User avatar
midwest17
Posts: 1686
Joined: Sat Aug 31, 2013 5:27 pm

Re: C/O 2016 median lsat/gpa/class size

Postby midwest17 » Tue Sep 24, 2013 9:48 pm

rayiner wrote:That doesn't make any sense, because of the way medians work. If you have 3.75/170 medians, and say 100 students, then at least 100 have a 3.75+ and at least 100 have a 170+. That means there is someone in the middle who has a 3.75/170. It's hard to say they're "solidly above average" because half the class has a higher LSAT score than they do, and half has a higher GPA.


No... at least 50 have 3.75+ and 50 have 170+. And there's nothing at all guaranteeing that there exists a student who's sitting at both medians...

User avatar
Monochromatic Oeuvre
Posts: 1929
Joined: Fri May 10, 2013 9:40 pm

Re: C/O 2016 median lsat/gpa/class size

Postby Monochromatic Oeuvre » Tue Sep 24, 2013 9:52 pm

rayiner wrote:
Monochromatic Oeuvre wrote:A simple test of whether your medians are manipulated is this: Would someone at both medians be a student of average or above-average quality at a school?

For example, at NYU, someone with a 3.7/170 is roughly a student of average quality, if we're measuring that using an intuitive idea of quality or something like the index formula. A 3.75/170 at Northwestern would be a solidly above-average student there, because their medians are such that most of the 3.8+ students have 165-168 and a whole bunch of the 170+ students have 3.4 or lower.

A "sliding scale", like the one CLS uses, is not median manipulation. It is making admissions almost entirely-numbers based, but consideration is only given to meet preset numerical standards on that scale, not specifically on whether someone is above a median.


That doesn't make any sense, because of the way medians work. If you have 3.75/170 medians, and say 100 students, then at least 100 have a 3.75+ and at least 100 have a 170+. That means there is someone in the middle who has a 3.75/170. It's hard to say they're "solidly above average" because half the class has a higher LSAT score than they do, and half has a higher GPA.


But schools can--and do--use full rides for exactly the purpose you're describing. The whole gaming thing would fall apart if you didn't have enough scholarship money to attract students above both your medians. But as it stands, you could theoretically have a gamed class with just one student above both medians. Suppose a class like this:

4.0/175 x1
3.9/165 x100
3.0/172 x100

Your medians are 3.9/172, even though the average quality of your students is WELL below the quality of a 3.9/172 student.

User avatar
rayiner
Posts: 6184
Joined: Thu Dec 11, 2008 11:43 am

Re: C/O 2016 median lsat/gpa/class size

Postby rayiner » Tue Sep 24, 2013 10:12 pm

Monochromatic Oeuvre wrote:
rayiner wrote:
Monochromatic Oeuvre wrote:A simple test of whether your medians are manipulated is this: Would someone at both medians be a student of average or above-average quality at a school?

For example, at NYU, someone with a 3.7/170 is roughly a student of average quality, if we're measuring that using an intuitive idea of quality or something like the index formula. A 3.75/170 at Northwestern would be a solidly above-average student there, because their medians are such that most of the 3.8+ students have 165-168 and a whole bunch of the 170+ students have 3.4 or lower.

A "sliding scale", like the one CLS uses, is not median manipulation. It is making admissions almost entirely-numbers based, but consideration is only given to meet preset numerical standards on that scale, not specifically on whether someone is above a median.


That doesn't make any sense, because of the way medians work. If you have 3.75/170 medians, and say 100 students, then at least 100 have a 3.75+ and at least 100 have a 170+. That means there is someone in the middle who has a 3.75/170. It's hard to say they're "solidly above average" because half the class has a higher LSAT score than they do, and half has a higher GPA.


But schools can--and do--use full rides for exactly the purpose you're describing. The whole gaming thing would fall apart if you didn't have enough scholarship money to attract students above both your medians. But as it stands, you could theoretically have a gamed class with just one student above both medians. Suppose a class like this:

4.0/175 x1
3.9/165 x100
3.0/172 x100

Your medians are 3.9/172, even though the average quality of your students is WELL below the quality of a 3.9/172 student.


In practice it doesn't look like that though. You get a bunch of 173/3.4 and 167/3.9 in a school with a median of 170/3.7. There's just not a lot of people at the real extremes, because there aren't a lot of super splitters in the applicant pool.

User avatar
Monochromatic Oeuvre
Posts: 1929
Joined: Fri May 10, 2013 9:40 pm

Re: C/O 2016 median lsat/gpa/class size

Postby Monochromatic Oeuvre » Tue Sep 24, 2013 10:17 pm

rayiner wrote:
Monochromatic Oeuvre wrote:But schools can--and do--use full rides for exactly the purpose you're describing. The whole gaming thing would fall apart if you didn't have enough scholarship money to attract students above both your medians. But as it stands, you could theoretically have a gamed class with just one student above both medians. Suppose a class like this:

4.0/175 x1
3.9/165 x100
3.0/172 x100

Your medians are 3.9/172, even though the average quality of your students is WELL below the quality of a 3.9/172 student.


In practice it doesn't look like that though. You get a bunch of 173/3.4 and 167/3.9 in a school with a median of 170/3.7. There's just not a lot of people at the real extremes, because there aren't a lot of super splitters in the applicant pool.


No, of course it's not quite that extreme. I'm just using a hypothetical to demonstrate a point--that by utilizing splitters and reverse splitters, along with a small number of people both medians, you can have medians that are inflated w/r/t the true quality of your student body, as is the case with Penn, UVA, and Northwestern, to name the worst offenders.

User avatar
mlansky
Posts: 69
Joined: Thu Apr 11, 2013 1:50 pm

Re: C/O 2016 median lsat/gpa/class size

Postby mlansky » Tue Sep 24, 2013 10:29 pm

A few questions, plz disregard if tangential: Why are some schools median gamers and others aren't? Do the tendencies and policies of admissions departments change, and if so, how quickly? Any speculation about whether anything will be different in the upcoming cycle?

User avatar
Crowing
Posts: 2636
Joined: Fri Feb 10, 2012 4:20 pm

Re: C/O 2016 median lsat/gpa/class size

Postby Crowing » Tue Sep 24, 2013 10:41 pm

Monochromatic Oeuvre wrote:
Crowing wrote:I don't see a large 25-50 split as necessarily indicative of sketchy behavior.

E.g. In the case of CLS/NYU vs. Chicago, one could say that CLS/NYU bases decisions very heavily on numbers, while Chicago is highly likely to admit applicants above its medians but is less focused on numbers and more on other variables of an applicant if they have lower numbers or are splitters. I'm not saying this is definitely the case, but I do think that quartile splits are by themselves not meaningful enough to draw conclusions about admissions philosophy from.


If you are highly likely to admit applicants above your medians, you are almost by definition highly unlikely to admit applicants below them. The fact that someone below a median is accepted does not make a school "less focused on numbers" w/r/t to that applicant, it just means other factors were significant enough to compensate for such a weakness. As applicants are in competition with one another because admissions is a zero sum game, numbers cannot be more or less important for an applicant merely because of what those numbers are. UChi's non-URM, non-ED acceptance rates last cycle:

At/above both medians: 87%
At/above one median and below another: 50%
Below both medians: 9%

There is not a big mystery going on, and UChi is not a black box. 167, 168 and 169 have a decent shot if they have a 3.9+. 3.5-3.6ish have a shot with a 172-175. They dip WAY below their GPA median for applicants slightly above the LSAT median and vice versa. Under a sliding scale (non-manipulated) applicants who are slightly above one median and well below the other are almost always dinged. Not so with manipulators.


What are the percentages for those data points for CLS and NYU?

User avatar
Monochromatic Oeuvre
Posts: 1929
Joined: Fri May 10, 2013 9:40 pm

Re: C/O 2016 median lsat/gpa/class size

Postby Monochromatic Oeuvre » Tue Sep 24, 2013 10:49 pm

mlansky wrote:A few questions, plz disregard if tangential: Why are some schools median gamers and others aren't? Do the tendencies and policies of admissions departments change, and if so, how quickly? Any speculation about whether anything will be different in the upcoming cycle?


The gaming phenomenon is relatively new. No school was doing it ten years ago.

It can change. UChi's gaming is quite recent--within the past couple years or so. To my knowledge, no school that formerly gamed has gone back to not gaming.

As for why some schools do it and why some don't: The incentive to game is obvious w/r/t USNWR. Look through the spreadsheet ITT and you'll see that most T1 schools do it in at least some fashion, even if it isn't incredibly blatant. As to those that don't:

-HYS doesn't really need to game, as it really wouldn't help them/matter at all.
-I don't know why CLS and NYU don't; maybe there's actually something non-cynical going on there.
-Berkeley will never do anything it doesn't feel like doing.
-Cornell and UT both don't either, which is extra weird because they both would have A LOT to gain if they did.

User avatar
LSATSCORES2012
Posts: 770
Joined: Mon Jul 02, 2012 2:12 pm

Re: C/O 2016 median lsat/gpa/class size

Postby LSATSCORES2012 » Tue Sep 24, 2013 10:55 pm

So you keep providing examples of gaming in a general sense but not showing that Chicago is doing it... the graphs on the previous page seem to show that they're not doing the splitter/reverse splitter thing of UVA.

Maybe I missed something because I'm not reading this thread very thoroughly.

(Btw, also based on the graphs on the previous page I think that the median might be 170... but hopefully not Reg's 169 :lol: Guess we'll know soon)
Last edited by LSATSCORES2012 on Tue Sep 24, 2013 10:56 pm, edited 1 time in total.

User avatar
Monochromatic Oeuvre
Posts: 1929
Joined: Fri May 10, 2013 9:40 pm

Re: C/O 2016 median lsat/gpa/class size

Postby Monochromatic Oeuvre » Tue Sep 24, 2013 10:56 pm

Crowing wrote:
Monochromatic Oeuvre wrote:
Crowing wrote:I don't see a large 25-50 split as necessarily indicative of sketchy behavior.

E.g. In the case of CLS/NYU vs. Chicago, one could say that CLS/NYU bases decisions very heavily on numbers, while Chicago is highly likely to admit applicants above its medians but is less focused on numbers and more on other variables of an applicant if they have lower numbers or are splitters. I'm not saying this is definitely the case, but I do think that quartile splits are by themselves not meaningful enough to draw conclusions about admissions philosophy from.


If you are highly likely to admit applicants above your medians, you are almost by definition highly unlikely to admit applicants below them. The fact that someone below a median is accepted does not make a school "less focused on numbers" w/r/t to that applicant, it just means other factors were significant enough to compensate for such a weakness. As applicants are in competition with one another because admissions is a zero sum game, numbers cannot be more or less important for an applicant merely because of what those numbers are. UChi's non-URM, non-ED acceptance rates last cycle:

At/above both medians: 87%
At/above one median and below another: 50%
Below both medians: 9%

There is not a big mystery going on, and UChi is not a black box. 167, 168 and 169 have a decent shot if they have a 3.9+. 3.5-3.6ish have a shot with a 172-175. They dip WAY below their GPA median for applicants slightly above the LSAT median and vice versa. Under a sliding scale (non-manipulated) applicants who are slightly above one median and well below the other are almost always dinged. Not so with manipulators.


What are the percentages for those data points for CLS and NYU?


CLS:

At/above both medians: 86%
At/above one median and below another: 45%
Below both medians: 9%

NYU:

At/above both medians: 95%
At/above one median and below another: 62%
Below both medians: 17%

Both may be slightly inflated because both schools dropped medians (and as such comparing last cycle to the last published median will be a little more favorable for the applicants than if the median hadn't dropped).

Mal Reynolds
Posts: 12630
Joined: Mon Oct 03, 2011 12:16 am

Re: C/O 2016 median lsat/gpa/class size

Postby Mal Reynolds » Tue Sep 24, 2013 10:59 pm

RAYINER IS BACK. NEVER LEAVE AGAIN.

User avatar
Monochromatic Oeuvre
Posts: 1929
Joined: Fri May 10, 2013 9:40 pm

Re: C/O 2016 median lsat/gpa/class size

Postby Monochromatic Oeuvre » Tue Sep 24, 2013 11:09 pm

LSATSCORES2012 wrote:So you keep providing examples of gaming in a general sense but not showing that Chicago is doing it... the graphs on the previous page seem to show that they're not doing the splitter/reverse splitter thing of UVA.

Maybe I missed something because I'm not reading this thread very thoroughly.

(Btw, also based on the graphs on the previous page I think that the median might be 170... but hopefully not Reg's 169 :lol: Guess we'll know soon)


I don't know how to show you Chicago is doing it just as I wouldn't know how to show you that UVA is doing it. I can only provide anecdotal evidence (the median much closer to the 75th, higher acceptances rates of those slightly above one median and well below another and whatnot) that I believe suggests it. I don't think they're doing it on quite the level as other schools, but I also don't think there is no element of artificiality in the medians.

User avatar
mlansky
Posts: 69
Joined: Thu Apr 11, 2013 1:50 pm

Re: C/O 2016 median lsat/gpa/class size

Postby mlansky » Tue Sep 24, 2013 11:10 pm

Monochromatic Oeuvre wrote:
mlansky wrote:A few questions, plz disregard if tangential: Why are some schools median gamers and others aren't? Do the tendencies and policies of admissions departments change, and if so, how quickly? Any speculation about whether anything will be different in the upcoming cycle?


The gaming phenomenon is relatively new. No school was doing it ten years ago.

It can change. UChi's gaming is quite recent--within the past couple years or so. To my knowledge, no school that formerly gamed has gone back to not gaming.

As for why some schools do it and why some don't: The incentive to game is obvious w/r/t USNWR. Look through the spreadsheet ITT and you'll see that most T1 schools do it in at least some fashion, even if it isn't incredibly blatant. As to those that don't:

-HYS doesn't really need to game, as it really wouldn't help them/matter at all.
-I don't know why CLS and NYU don't; maybe there's actually something non-cynical going on there.
-Berkeley will never do anything it doesn't feel like doing.
-Cornell and UT both don't either, which is extra weird because they both would have A LOT to gain if they did.


Makes sense, thanks for all the patient explanation. I guess I just have very limited knowledge of the chain of command involved in admissions decisions and the politics thereof. HYS policies make sense, but how certain schools (like Berkeley, for example) develop their own tastes and are stricter on GPA than LSAT or vice-versa really puzzles me. To the extent that it's not due to USNWR (which the individual traits can't be, as USNWR applies the same criteria to all, right?) it must be the effect of a group of people acting within a particular historical tradition, which makes changes in those trends even more puzzling.

I guess what I'm really wondering is, in the case of a school that starts gaming, did someone in charge make that call? Did everyone in the admissions office just decide that they had enough being ranked X and they were going to do something about it? Do you think they talk openly about these things, or is it just an understood policy shift? Where does the pressure come from?

User avatar
LSATSCORES2012
Posts: 770
Joined: Mon Jul 02, 2012 2:12 pm

Re: C/O 2016 median lsat/gpa/class size

Postby LSATSCORES2012 » Tue Sep 24, 2013 11:16 pm

Monochromatic Oeuvre wrote:
LSATSCORES2012 wrote:So you keep providing examples of gaming in a general sense but not showing that Chicago is doing it... the graphs on the previous page seem to show that they're not doing the splitter/reverse splitter thing of UVA.

Maybe I missed something because I'm not reading this thread very thoroughly.

(Btw, also based on the graphs on the previous page I think that the median might be 170... but hopefully not Reg's 169 :lol: Guess we'll know soon)


I don't know how to show you Chicago is doing it just as I wouldn't know how to show you that UVA is doing it. I can only provide anecdotal evidence (the median much closer to the 75th, higher acceptances rates of those slightly above one median and well below another and whatnot) that I believe suggests it. I don't think they're doing it on quite the level as other schools, but I also don't think there is no element of artificiality in the medians.

I mean, if I wanted to show you that UVA was gaming, I'd show you this picture:

Image

You can see two clear lines where anyone at/above 170 LSAT is accepted, and anyone at/above whatever GPA is accepted (I excluded waitlists).

Here's Chicago's:

Image

Here's Columbia's - Col actually seems to have a more defined line in the LSAT direction, but that also might be a function of just having more data points

Image

Here's NYU's:

Image

In these graphs I can see where you might see it a little bit, but I don't see anything anywhere near as definitive as UVA's (and this might, again, be due to the smaller sample size).

Edit: Maybe the fact that there's more green below the medians than at UVA/Chi has a higher GPA cutoff makes it harder to see the clearly defined lines (UVA seems to literally admit zero people below the medians).

User avatar
cotiger
Posts: 1648
Joined: Tue Jul 23, 2013 11:49 pm

Re: C/O 2016 median lsat/gpa/class size

Postby cotiger » Tue Sep 24, 2013 11:24 pm

Monochromatic Oeuvre wrote:
Crowing wrote:
Monochromatic Oeuvre wrote:
Crowing wrote:I don't see a large 25-50 split as necessarily indicative of sketchy behavior.

E.g. In the case of CLS/NYU vs. Chicago, one could say that CLS/NYU bases decisions very heavily on numbers, while Chicago is highly likely to admit applicants above its medians but is less focused on numbers and more on other variables of an applicant if they have lower numbers or are splitters. I'm not saying this is definitely the case, but I do think that quartile splits are by themselves not meaningful enough to draw conclusions about admissions philosophy from.


If you are highly likely to admit applicants above your medians, you are almost by definition highly unlikely to admit applicants below them. The fact that someone below a median is accepted does not make a school "less focused on numbers" w/r/t to that applicant, it just means other factors were significant enough to compensate for such a weakness. As applicants are in competition with one another because admissions is a zero sum game, numbers cannot be more or less important for an applicant merely because of what those numbers are. UChi's non-URM, non-ED acceptance rates last cycle:

At/above both medians: 87%
At/above one median and below another: 50%
Below both medians: 9%

There is not a big mystery going on, and UChi is not a black box. 167, 168 and 169 have a decent shot if they have a 3.9+. 3.5-3.6ish have a shot with a 172-175. They dip WAY below their GPA median for applicants slightly above the LSAT median and vice versa. Under a sliding scale (non-manipulated) applicants who are slightly above one median and well below the other are almost always dinged. Not so with manipulators.


What are the percentages for those data points for CLS and NYU?


CLS:

At/above both medians: 86%
At/above one median and below another: 45%
Below both medians: 9%

NYU:

At/above both medians: 95%
At/above one median and below another: 62%
Below both medians: 17%

Both may be slightly inflated because both schools dropped medians (and as such comparing last cycle to the last published median will be a little more favorable for the applicants than if the median hadn't dropped).


It seems like NYU and Columbia are two of the most likely to accept a GPA below 25% if the LSAT is above median, and NYU and Michigan are (of the data available) two of the most likely to accept an LSAT below 25% if GPA is above median.

Looks to me like NYU might be guilty of this kind of median gaming you're talking about.

https://docs.google.com/spreadsheet/ccc ... ZN2c#gid=0

User avatar
Monochromatic Oeuvre
Posts: 1929
Joined: Fri May 10, 2013 9:40 pm

Re: C/O 2016 median lsat/gpa/class size

Postby Monochromatic Oeuvre » Tue Sep 24, 2013 11:25 pm

mlansky wrote:
Monochromatic Oeuvre wrote:
mlansky wrote:A few questions, plz disregard if tangential: Why are some schools median gamers and others aren't? Do the tendencies and policies of admissions departments change, and if so, how quickly? Any speculation about whether anything will be different in the upcoming cycle?


The gaming phenomenon is relatively new. No school was doing it ten years ago.

It can change. UChi's gaming is quite recent--within the past couple years or so. To my knowledge, no school that formerly gamed has gone back to not gaming.

As for why some schools do it and why some don't: The incentive to game is obvious w/r/t USNWR. Look through the spreadsheet ITT and you'll see that most T1 schools do it in at least some fashion, even if it isn't incredibly blatant. As to those that don't:

-HYS doesn't really need to game, as it really wouldn't help them/matter at all.
-I don't know why CLS and NYU don't; maybe there's actually something non-cynical going on there.
-Berkeley will never do anything it doesn't feel like doing.
-Cornell and UT both don't either, which is extra weird because they both would have A LOT to gain if they did.


Makes sense, thanks for all the patient explanation. I guess I just have very limited knowledge of the chain of command involved in admissions decisions and the politics thereof. HYS policies make sense, but how certain schools (like Berkeley, for example) develop their own tastes and are stricter on GPA than LSAT or vice-versa really puzzles me. To the extent that it's not due to USNWR (which the individual traits can't be, as USNWR applies the same criteria to all, right?) it must be the effect of a group of people acting within a particular historical tradition, which makes changes in those trends even more puzzling.

I guess what I'm really wondering is, in the case of a school that starts gaming, did someone in charge make that call? Did everyone in the admissions office just decide that they had enough being ranked X and they were going to do something about it? Do you think they talk openly about these things, or is it just an understood policy shift? Where does the pressure come from?


Berkeley really is idiosyncratic. Every year they reject a bunch of HYS-bound people for no outwardly visible reason. They're also the only school that specifically asks about your socioeconomic status in their app. Clearly particular things are very important to them. Medians matter, but admissions offices are still comprised of people who want to see certain things.

If a school is gaming (and again, the reasons to do so are obvious), presumably it's the call of the Dean of Admissions, or perhaps someone higher up in the food chain (which is probably where the pressure comes from). If a school is doing it, then yes, everyone in the office knows about it and it's probably explicitly discussed. From what I can gather from written sources, and a few off-the-record conversations with people who work in admissions, medians are the number one concern. This may be a question you want to ask Spivey. He's talked about how offices have spreadsheets updated daily to show the current medians and how safe they are.

User avatar
Lavitz
Posts: 3098
Joined: Mon May 31, 2010 1:39 am

Re: C/O 2016 median lsat/gpa/class size

Postby Lavitz » Tue Sep 24, 2013 11:29 pm

Monochromatic Oeuvre wrote:-Cornell and UT both don't either, which is extra weird because they both would have A LOT to gain if they did.

With regard to Cornell, I think it's mostly because Dean Geiger is old school and liked the way things used to be.

See http://top-law-schools.com/archives/vie ... &p=1888744

[...]
Rarely, a few voices are still raised to protest the corrupting effect the rankings have on academic decision-making. However, regardless of where a school stands publicly or what it does behind the scenes, no one can credibly dispute the fact that the rankings are now fully ingrained as a part of law school operations and culture.

[...]

The rankings take no account of other admission factors. There is nothing about where degrees were earned; nothing about grading standards of schools or programs; nothing about whether applicants ever spent time outside the library as a student. There is also nothing about graduate work or community service or work experience; nothing about communication skills; nothing about leadership, resilience or passion.

[...]

To increase your application volume, you might spend more on recruitment and outreach efforts. You might increase your travel budget or encourage reluctant applicants to apply by waiving your application fees.

To lower the number of offers you make (remember, fewer offers mean higher selectivity), you might try to pre-screen applicants based on an assessment of how likely they are to accept your offer. If they are likely to get into higher-ranked schools, you could make them jump over an extra hurdle or even put them on a waiting list to test their commitment before you'll make them an offer.

[...]

Since the LSAT and GPA medians are the major criteria involved in assessing student selectivity, your focus will trend toward them. You will certainly deemphasize factors like work experience, graduate school, grade trends and letters of recommendation. Soft factors, like leadership or people skills, are very likely to go since they are the hardest and most time-intensive to tease out of an application anyway.

To raise medians you might decide to trade off high LSATs with low GPAs and vice versa. You might also decide to look at undergraduate GPAs in a vacuum, (i.e., ignore contextual factors like where it was earned, relevance or difficulty of the major, etc.). You might reduce the size of your entering class, but make up for it with part-time, transfer, or graduate students who are not counted for rankings purposes.

There are no direct costs to these approaches. In fact, if a school were to focus only on numerical credentials in assessing academic potential, it would actually reduce the costs associated with judging "soft," more subjective criteria.

[...]

In a rankings driven process, need-based financial aid would definitely take a back seat, because it involves a high cost for very little direct, rankings-related return. Merit-based aid, particularly with merit defined as high test scores and GPAs, would become the preferred allocation method. Carefully targeted merit aid can both lower the number of offers a school would have to make and raise the academic credentials of those who decide to accept an offer of admission.

[...]

While law schools and legal education rightly are held to high standards, the U.S. News ranking criteria push schools toward strategies that, over time, tend to deemphasize the subtle but important non-numerical attributes of merit that often define success in the legal profession.

Instead, they favor the diversion of scarce selection resources toward manipulation of the application process and toward devices aimed purely at raising the raw numerical credentials of an entering class. This ends up devaluing motivation, resilience, leadership, judgment and the many other soft factors that often define the difference between the good and the great members of our profession.


Of course, the admissions office is still beholden to the rankings, and there's evidence they've been caving to the pressure, but that doesn't mean they have to like it. So I guess they'd rather not participate in gaming if they can help it.

User avatar
Tiago Splitter
Posts: 15525
Joined: Tue Jun 28, 2011 1:20 am

Re: C/O 2016 median lsat/gpa/class size

Postby Tiago Splitter » Tue Sep 24, 2013 11:34 pm

Anyone looking for a good laugh should read those holier-than-thou quotes and then go check the LSN graph for Cornell.

User avatar
Monochromatic Oeuvre
Posts: 1929
Joined: Fri May 10, 2013 9:40 pm

Re: C/O 2016 median lsat/gpa/class size

Postby Monochromatic Oeuvre » Tue Sep 24, 2013 11:36 pm

cotiger wrote:
Monochromatic Oeuvre wrote:
CLS:

At/above both medians: 86%
At/above one median and below another: 45%
Below both medians: 9%

NYU:

At/above both medians: 95%
At/above one median and below another: 62%
Below both medians: 17%

Both may be slightly inflated because both schools dropped medians (and as such comparing last cycle to the last published median will be a little more favorable for the applicants than if the median hadn't dropped).


It seems like NYU and Columbia are two of the most likely to accept a GPA below 25% if the LSAT is above median, and NYU and Michigan are (of the data available) two of the most likely to accept an LSAT below 25% if GPA is above median.

Looks to me like NYU might be guilty of this kind of median gaming you're talking about.

https://docs.google.com/spreadsheet/ccc ... ZN2c#gid=0


CLS and NYU operate on essentially a sliding scale. This is not the same thing as median gaming.

For example, CLS's admissions standards used to be fairly clearly defined with requisite GPAs based on your LSAT:

170: 3.9
171: 3.8
172: 3.7
173: 3.6
174: 3.5

The GPA could go a little lower if the LSAT was a little higher, but the scale was mostly predictable. With a balanced applicant pool, the acceptances were fairly uniform along that scale, so the medians came out about halfway between the 25th and the 75th. The key thing was that for the most part, someone who was accepted below one median was also well above the other. They weren't accepting 3.8/168 or 3.4/173, for example. NYU does something similar.

User avatar
jingosaur
Posts: 2225
Joined: Fri Jan 04, 2013 10:33 am

Re: C/O 2016 median lsat/gpa/class size

Postby jingosaur » Tue Sep 24, 2013 11:42 pm

Monochromatic Oeuvre wrote:
cotiger wrote:
Monochromatic Oeuvre wrote:
CLS:

At/above both medians: 86%
At/above one median and below another: 45%
Below both medians: 9%

NYU:

At/above both medians: 95%
At/above one median and below another: 62%
Below both medians: 17%

Both may be slightly inflated because both schools dropped medians (and as such comparing last cycle to the last published median will be a little more favorable for the applicants than if the median hadn't dropped).


It seems like NYU and Columbia are two of the most likely to accept a GPA below 25% if the LSAT is above median, and NYU and Michigan are (of the data available) two of the most likely to accept an LSAT below 25% if GPA is above median.

Looks to me like NYU might be guilty of this kind of median gaming you're talking about.

https://docs.google.com/spreadsheet/ccc ... ZN2c#gid=0


CLS and NYU operate on essentially a sliding scale. This is not the same thing as median gaming.

For example, CLS's admissions standards used to be fairly clearly defined with requisite GPAs based on your LSAT:

170: 3.9
171: 3.8
172: 3.7
173: 3.6
174: 3.5

The GPA could go a little lower if the LSAT was a little higher, but the scale was mostly predictable. With a balanced applicant pool, the acceptances were fairly uniform along that scale, so the medians came out about halfway between the 25th and the 75th. The key thing was that for the most part, someone who was accepted below one median was also well above the other. They weren't accepting 3.8/168 or 3.4/173, for example. NYU does something similar.


NYU has some of the most predictable admissions of the T14. I still don't know whether this is a good thing or a bad thing. Since test taking is such an important part of law school and being barred as a lawyer, it kind of makes sense that admissions are based almost solely on testing. But at the same time, I'm more that just what's on a piece of paper, I'm a special snowflake with my own special abilities, and there are just so many things that make me such a wonderful person and you should consider that because my mommy and daddy say that I'm better than everybody else.

User avatar
jbagelboy
Posts: 9652
Joined: Thu Nov 29, 2012 7:57 pm

Re: C/O 2016 median lsat/gpa/class size

Postby jbagelboy » Tue Sep 24, 2013 11:51 pm

LSATSCORES2012 wrote:
Monochromatic Oeuvre wrote:
LSATSCORES2012 wrote:So you keep providing examples of gaming in a general sense but not showing that Chicago is doing it... the graphs on the previous page seem to show that they're not doing the splitter/reverse splitter thing of UVA.

Maybe I missed something because I'm not reading this thread very thoroughly.

(Btw, also based on the graphs on the previous page I think that the median might be 170... but hopefully not Reg's 169 :lol: Guess we'll know soon)


I don't know how to show you Chicago is doing it just as I wouldn't know how to show you that UVA is doing it. I can only provide anecdotal evidence (the median much closer to the 75th, higher acceptances rates of those slightly above one median and well below another and whatnot) that I believe suggests it. I don't think they're doing it on quite the level as other schools, but I also don't think there is no element of artificiality in the medians.

I mean, if I wanted to show you that UVA was gaming, I'd show you this picture:

Image

You can see two clear lines where anyone at/above 170 LSAT is accepted, and anyone at/above whatever GPA is accepted (I excluded waitlists).

Here's Chicago's:

Image

Here's Columbia's - Col actually seems to have a more defined line in the LSAT direction, but that also might be a function of just having more data points

Image

Here's NYU's:

Image

In these graphs I can see where you might see it a little bit, but I don't see anything anywhere near as definitive as UVA's (and this might, again, be due to the smaller sample size).

Edit: Maybe the fact that there's more green below the medians than at UVA/Chi has a higher GPA cutoff makes it harder to see the clearly defined lines (UVA seems to literally admit zero people below the medians).


Since the use of the waitlist YP is key to UVA's admissions policy, and definitely impacts decision making in the 170-172 LSAT range for CCN, I would humbly suggest that it be included as well

User avatar
LSATSCORES2012
Posts: 770
Joined: Mon Jul 02, 2012 2:12 pm

Re: C/O 2016 median lsat/gpa/class size

Postby LSATSCORES2012 » Wed Sep 25, 2013 12:23 am

Since the use of the waitlist YP is key to UVA's admissions policy, and definitely impacts decision making in the 170-172 LSAT range for CCN, I would humbly suggest that it be included as well

I totally agree with you, but I think the primary focus here is on the idea that Chicago is admitting splitters/reverse splitters in a UVA like fashion, not a claim that Chicago is YPing (which I actually am more inclined to buy...)

I probably should have drawn attention to that fact in my post to make it more clear.

The WL's just make it hard to see what's going on with admits/rejects which I THINK is what really matters for the s/rs thing.

The main point of the UVA graph is to show that, ignoring YP (which is what the waitlists represent), if you are above a certain LSAT or GPA you are virtually guaranteed admission, and if you are below both it is virtually impossible to be admitted. That's what the gaming (I think) we're talking about is, so that's what I focused on.

But like I said, I totally agree with you that the YPing is another factor - a significant one, but not so much related to the medians, and I do think that Chicago engaged in that... but not like UVA, in that it's probably not as easy to see because I think it was done through the interviews.

User avatar
jbagelboy
Posts: 9652
Joined: Thu Nov 29, 2012 7:57 pm

Re: C/O 2016 median lsat/gpa/class size

Postby jbagelboy » Wed Sep 25, 2013 12:40 am

LSATSCORES2012 wrote:
Since the use of the waitlist YP is key to UVA's admissions policy, and definitely impacts decision making in the 170-172 LSAT range for CCN, I would humbly suggest that it be included as well

I totally agree with you, but I think the primary focus here is on the idea that Chicago is admitting splitters/reverse splitters in a UVA like fashion, not a claim that Chicago is YPing (which I actually am more inclined to buy...)

I probably should have drawn attention to that fact in my post to make it more clear.

The WL's just make it hard to see what's going on with admits/rejects which I THINK is what really matters for the s/rs thing.

The main point of the UVA graph is to show that, ignoring YP (which is what the waitlists represent), if you are above a certain LSAT or GPA you are virtually guaranteed admission, and if you are below both it is virtually impossible to be admitted. That's what the gaming (I think) we're talking about is, so that's what I focused on.

But like I said, I totally agree with you that the YPing is another factor - a significant one, but not so much related to the medians, and I do think that Chicago engaged in that... but not like UVA, in that it's probably not as easy to see because I think it was done through the interviews.


This is fair, and I appreciate what you're saying here and why you made those graphs.

To Chicago's credit (gasp), I originally believed the interviews to be a more positive sign of selectivity, but given the relatively high rate of interview -> acceptance, especially early on, I agree its purpose is more to weed out the clearly uninterested, hence avoiding admitting unnecessary applications and protecting the yield metric. Columbia and certainly NYU don't seem to attempt this to the same degree

Still, I liked the attempt at personalization, albeit remote, in Chicago's interview process




Return to “Law School Admissions Forum”

Who is online

Users browsing this forum: Bing [Bot], hwwong, smieclux and 7 guests