Employment numbers: % of class v. number of students overall

(Rankings, Profiles, Tuition, Student Life, . . . )
TeufelHunden88
Posts: 22
Joined: Thu Oct 17, 2013 10:14 pm

Employment numbers: % of class v. number of students overall

Postby TeufelHunden88 » Thu Nov 14, 2013 5:25 pm

Basically I'm wondering what is more important to consider:

1. the percentage of a particular school's graduating class that placed in BigLaw and Fed. Clerkships

or

2. the overall number of students placed in BigLaw and Fed. Clerkships

For example, 77.1% of Penn's 2012 class went to BigLaw/Fed. Clerkship, whereas 71.4% of Harvard's 2012 class went BigLaw/Fed. Clerkship. However, that means 208 Penn grads got those jobs and 421 Harvard grads got those jobs. So which should have more influence, the amount of the class or the amount overall?

I'm leaning towards overall... Thoughts?

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

Re: Employment numbers: % of class v. number of students overall

Postby cotiger » Thu Nov 14, 2013 5:45 pm

Percentage. I don't know how this is actually a question.

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

Re: Employment numbers: % of class v. number of students overall

Postby Ti Malice » Thu Nov 14, 2013 5:46 pm

First, Penn is neither Harvard's superior nor its peer. The BigLaw+FedClerk metric isn't as useful for YHS.

Second, generically, why would you care about anything other than percentage? All that matters for you is your chances of having the kind of outcome you want. Try to explain in words why overall count means anything for you.

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

Re: Employment numbers: % of class v. number of students overall

Postby cotiger » Thu Nov 14, 2013 6:10 pm

To be slightly more helpful.. if you're confused about percentages, consider this example:

School X: 500 students, 90% of whom get favorable outcomes
School Y: 200 students, 95% of whom get favorable outcomes

So, 450 School Xers get favorable outcomes vs only 190 School Yers. School X is the clear winner, right? Wrong.

Pick 100 students at random from each school. At school Y, 95 of those 100 have favorable outcomes. At school X, only 90 do.

If you are a student at school Y, 95 times out of 100 you will get a favorable outcome. If you are a student at school X, you will get a favorable outcome only 90 out of 100 times.

Rate statistics are what count.

Does this clear that up for you?

User avatar
Tiago Splitter
Posts: 15513
Joined: Tue Jun 28, 2011 1:20 am

Re: Employment numbers: % of class v. number of students overall

Postby Tiago Splitter » Thu Nov 14, 2013 6:19 pm

If you're a career services official at a large school trying to negotiate his next pay raise you want to focus on the total number of people getting these jobs. If you're a student, you want to focus on the percentage of people getting these jobs.

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

Re: Employment numbers: % of class v. number of students overall

Postby cotiger » Thu Nov 14, 2013 6:25 pm

Tiago Splitter wrote:If you're a career services official at a large school trying to negotiate his next pay raise you want to focus on the total number of people getting these jobs. If you're a student, you want to focus on the percentage of people getting these jobs.


LOL. Cooley, of the legendary 24.7% employment score, got LTFT JD required jobs for 267 C/O 2012 graduates--more than the entire class at sTTTanford! Clearly their CSO is at least #2 in the nation.

TeufelHunden88
Posts: 22
Joined: Thu Oct 17, 2013 10:14 pm

Re: Employment numbers: % of class v. number of students overall

Postby TeufelHunden88 » Thu Nov 14, 2013 6:43 pm

cotiger wrote:To be slightly more helpful.. if you're confused about percentages, consider this example:

School X: 500 students, 90% of whom get favorable outcomes
School Y: 200 students, 95% of whom get favorable outcomes

So, 450 School Xers get favorable outcomes vs only 190 School Yers. School X is the clear winner, right? Wrong.

Pick 100 students at random from each school. At school Y, 95 of those 100 have favorable outcomes. At school X, only 90 do.

If you are a student at school Y, 95 times out of 100 you will get a favorable outcome. If you are a student at school X, you will get a favorable outcome only 90 out of 100 times.

Rate statistics are what count.

Does this clear that up for you?



That does clear it up for me... thanks

User avatar
phillywc
Posts: 3043
Joined: Tue Feb 19, 2013 12:17 am

Re: Employment numbers: % of class v. number of students overall

Postby phillywc » Thu Nov 14, 2013 8:29 pm

I think there is an argument to made that if 2 schools have roughly equal placement percentages, a larger class size is more impressive, but generally I agree with everyone else.

User avatar
hichvichwoh
Posts: 442
Joined: Fri Mar 16, 2012 11:21 am

Re: Employment numbers: % of class v. number of students overall

Postby hichvichwoh » Thu Nov 14, 2013 8:34 pm

phillywc wrote:I think there is an argument to made that if 2 schools have roughly equal placement percentages, a larger class size is more impressive, but generally I agree with everyone else.


if there is, would you care to articulate it? I don't really see how # should matter at all

User avatar
phillywc
Posts: 3043
Joined: Tue Feb 19, 2013 12:17 am

Re: Employment numbers: % of class v. number of students overall

Postby phillywc » Thu Nov 14, 2013 8:42 pm

hichvichwoh wrote:
phillywc wrote:I think there is an argument to made that if 2 schools have roughly equal placement percentages, a larger class size is more impressive, but generally I agree with everyone else.


if there is, would you care to articulate it? I don't really see how # should matter at all

Alumni network/name rec/etc. Plus 95% at 500 is just more impressive than 95% at 200. I'd personally prefer a smaller class size but i can see the argument.

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

Re: Employment numbers: % of class v. number of students overall

Postby cotiger » Thu Nov 14, 2013 8:57 pm

phillywc wrote:
hichvichwoh wrote:
phillywc wrote:I think there is an argument to made that if 2 schools have roughly equal placement percentages, a larger class size is more impressive, but generally I agree with everyone else.


if there is, would you care to articulate it? I don't really see how # should matter at all

Alumni network/name rec/etc. Plus 95% at 500 is just more impressive than 95% at 200. I'd personally prefer a smaller class size but i can see the argument.


But all the increased size of that alumni network did was allow it to place the same percentage of its class as the smaller school. Same overall outcomes.

Take this simplified example. Say school A is twice the size of equal-reputation/placement school B. It would make sense that school A would have twice the alumni network as school B. That doubled alumni network of A is going to want to hire double the number of A grads due to alumni good feelings. So yeah, the larger alumni network got twice the number of kids placed, but they had twice the number that needed to be placed to begin with. The overall percentages are the same. Kids at school A and school B have equal shots at good outcomes.

Rates are the only way to compare outcomes between schools.

The only consideration that the larger alumni network contributes is that it will likely result in having a significant presence in more diverse areas.

User avatar
ScottRiqui
Posts: 3640
Joined: Mon Nov 29, 2010 8:09 pm

Re: Employment numbers: % of class v. number of students overall

Postby ScottRiqui » Thu Nov 14, 2013 9:23 pm

I think the percentages might be more representative and stable year-to-year with a larger class size, though. With a class of 100, it would only take a couple of "coin flips" going the other way for last year's 65% to turn into this year's 60%.

User avatar
Tiago Splitter
Posts: 15513
Joined: Tue Jun 28, 2011 1:20 am

Re: Employment numbers: % of class v. number of students overall

Postby Tiago Splitter » Thu Nov 14, 2013 10:01 pm

ScottRiqui wrote:I think the percentages might be more representative and stable year-to-year with a larger class size, though. With a class of 100, it would only take a couple of "coin flips" going the other way for last year's 65% to turn into this year's 60%.

Yeah if you look at the history going back to 2005 the big swings tend to be by small schools like Cornell and Duke. But all someone has to do is take a look at several years' data to come up with something of an average, which works as long as the big swings are largely the result of self selection (which I believe they are.)

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

Re: Employment numbers: % of class v. number of students overall

Postby cotiger » Thu Nov 14, 2013 10:50 pm

Tiago Splitter wrote:
ScottRiqui wrote:I think the percentages might be more representative and stable year-to-year with a larger class size, though. With a class of 100, it would only take a couple of "coin flips" going the other way for last year's 65% to turn into this year's 60%.

Yeah if you look at the history going back to 2005 the big swings tend to be by small schools like Cornell and Duke. But all someone has to do is take a look at several years' data to come up with something of an average, which works as long as the big swings are largely the result of self selection (which I believe they are.)


Exactly. If the averages are the same, that's all that really matters. Sometimes one does a little better, sometimes a little worse.

For kicks, I calculated the correlation between the biglaw+fedclerk variance over the past three classes of the T14+UT/Vandy/USC/UCLA and the size of their 1L class for C/O 2013. Comes out to be -.20

That's a pretty weak negative correlation, verging on negligible. Tiago, if you have that historical data on a spreadsheet I'd love to check it out.

User avatar
Tiago Splitter
Posts: 15513
Joined: Tue Jun 28, 2011 1:20 am

Re: Employment numbers: % of class v. number of students overall

Postby Tiago Splitter » Thu Nov 14, 2013 10:57 pm

cotiger wrote:
Tiago Splitter wrote:
ScottRiqui wrote:I think the percentages might be more representative and stable year-to-year with a larger class size, though. With a class of 100, it would only take a couple of "coin flips" going the other way for last year's 65% to turn into this year's 60%.

Yeah if you look at the history going back to 2005 the big swings tend to be by small schools like Cornell and Duke. But all someone has to do is take a look at several years' data to come up with something of an average, which works as long as the big swings are largely the result of self selection (which I believe they are.)


Exactly. If the averages are the same, that's all that really matters. Sometimes one does a little better, sometimes a little worse.

For kicks, I calculated the correlation between the biglaw+fedclerk variance over the past three classes of the T14+UT/Vandy/USC/UCLA and the size of their 1L class for C/O 2013. Comes out to be -.20

That's a pretty weak negative correlation, verging on negligible. Tiago, if you have that historical data on a spreadsheet I'd love to check it out.

There's some good 2007-2009 data on this page:

http://www.top-law-schools.com/forums/v ... 1&t=150004

To play devil's advocate, I think an argument can be made that fluctuations could be caused by factors other than self-selection (location, sensitivity to economic changes, etc.) and if so an applicant might seek out a school with a bit more consistency. Ultimately the only thing that really matters is how your school places in August and September before your second year, and things falling apart in a way that aren't systemic are a career killer you want to avoid. But I'd personally be comfortable using averages.

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

Re: Employment numbers: % of class v. number of students overall

Postby cotiger » Thu Nov 14, 2013 11:48 pm

Tiago Splitter wrote:
cotiger wrote:Exactly. If the averages are the same, that's all that really matters. Sometimes one does a little better, sometimes a little worse.

For kicks, I calculated the correlation between the biglaw+fedclerk variance over the past three classes of the T14+UT/Vandy/USC/UCLA and the size of their 1L class for C/O 2013. Comes out to be -.20

That's a pretty weak negative correlation, verging on negligible. Tiago, if you have that historical data on a spreadsheet I'd love to check it out.

There's some good 2007-2009 data on this page:

viewtopic.php?f=1&t=150004

To play devil's advocate, I think an argument can be made that fluctuations could be caused by factors other than self-selection (location, sensitivity to economic changes, etc.) and if so an applicant might seek out a school with a bit more consistency. Ultimately the only thing that really matters is how your school places in August and September before your second year, and things falling apart in a way that aren't systemic are a career killer you want to avoid. But I'd personally be comfortable using averages.


Adding in that data reduces the correlation between size and variance to -.14. That's essentially negligible.

The schools with the three largest variances (i.e. biggest swings) are Cornell (by a large amount), then Chicago, then NYU.

The schools with the three smallest variances (i.e. most consistent) are Stanford, then Harvard and Yale.

YHS are not only the best, they're also the most reliable, which is consistent with the conventional wisdom.

Watchenlaw
Posts: 5
Joined: Mon Nov 25, 2013 5:29 pm

Re: Employment numbers: % of class v. number of students overall

Postby Watchenlaw » Mon Nov 25, 2013 6:17 pm

Let's assume that your goal is only to get some biglaw or clerking gig upon graduation.

Another thing to keep in mind, then, don't just look at the % of students who get jobs. Consider this in conjunction with how difficult it is to get into the school--namely, school rank and average GPA and LSAT data. This will tell you how stiff your competition will be to make it into the top 70% or 80% or whatever% of students who achieve a good outcome. LSAT and GPA are not great predictors of individual law school success, but they will give you some idea of how talanted students will be at a particular school relative to another school. (If you don't think the quality of students--academically at least, which is all that matters for our purposes--is higher at higher-ranked schools, I guess the following will not persuade you; but really dude? Really?.)

For example, if the percentage of people getting good outcomes were the same at Vandy and at Virginia (its not, but let's pretend), then go to Vandy--the percentage of people at both schools would be the same, but your likelihood of making it into the percentage of people getting good outcomes will be much better. All else being equal (and still assuming that the goal is just any biglaw or clerking), Vandy is clearly the better option for you.

Another example: say you are choosing between a significant scholarship at Duke and coming off the waitlist at Columbia. Say 75% of Columbia grads typically succeed while 65% of Duke grads typically succeed. What is the better move, all else being equal? Well, I suggest you have a higher likelihood of excelling at Duke than you do at Columbia, and probably a greater chance of making Duke's 65% than Columbia's 75%. Plus you get money; everybody likes money. Therefore, Duke is the better choice. This is more of less how I chose a lower-T14 over higher-ranked schools and am happy with the outcome (not precisely--I agreed to jump off waitlists at other schools and accept at my school for more scholarship money--but the reasoning was the same).

Also, don't fixate on "What if I don't do well, I will be safer with a Columbia degree than a Duke degree," or whatever. Of course, you could bomb, but you are probably just as likely to do that at either school, in which case, at least it did not cost you as much at Duke. This concern, and probably the analysis above in general, breaks down when you have a truely meaningful gap between the schools. Yes, if you get into Cooley and Texas, you are likely to kill it at Cooley, but the cost/benefit will never pay off in this case. And all else being equal, I'd never choose Iowa over Penn. The aim should be to hit the sweet spot where you are in the top say quarter of accepted applicants, but not head and shoulders over everyone else, in which case you are probably missing out on opportunities by not going to the better school.

There are other factors to consider, of course, including the long term career benefit of having a Columbia degree, but I think that sort of "value" is too abstract in this economy. YHS, and in particular Y, have extra factors that you need to keep in mind--prominantly, people who opt out of biglaw and clerking--that might skew their employment data. This will be important but less true at some other schools, like Berkeley and Michigan.

User avatar
Tiago Splitter
Posts: 15513
Joined: Tue Jun 28, 2011 1:20 am

Re: Employment numbers: % of class v. number of students overall

Postby Tiago Splitter » Mon Nov 25, 2013 6:33 pm

Watchenlaw wrote:Another example: say you are choosing between a significant scholarship at Duke and coming off the waitlist at Columbia. Say 75% of Columbia grads typically succeed while 65% of Duke grads typically succeed. What is the better move, all else being equal? Well, I suggest you have a higher likelihood of excelling at Duke than you do at Columbia, and probably a greater chance of making Duke's 65% than Columbia's 75%. Plus you get money; everybody likes money. Therefore, Duke is the better choice. This is more of less how I chose a lower-T14 over higher-ranked schools and am happy with the outcome (not precisely--I agreed to jump off waitlists at other schools and accept at my school for more scholarship money--but the reasoning was the same).

This is ridiculous when you consider that Columbia's LSAT median is just two points higher than Duke's and Duke's GPA median is actually higher than Columbia's. There are reasons to take Duke over Columbia but thinking you'll place higher at Duke is not one of them.

Watchenlaw wrote:Also, don't fixate on "What if I don't do well, I will be safer with a Columbia degree than a Duke degree," or whatever. Of course, you could bomb, but you are probably just as likely to do that at either school, in which case, at least it did not cost you as much at Duke.

I thought your odds of ending up at any particular point in the class were different?


Watchenlaw wrote:This concern, and probably the analysis above in general, breaks down when you have a truely meaningful gap between the schools. Yes, if you get into Cooley and Texas, you are likely to kill it at Cooley, but the cost/benefit will never pay off in this case.

The funny thing is the analysis actually works when you look at schools with big gaps between them. Schools ranked close to one another provide such a small difference in matriculant quality that thinking you'll do better at one vs. the other just because of miniscule differences in LSAT or GPA numbers is crazy.

Watchenlaw
Posts: 5
Joined: Mon Nov 25, 2013 5:29 pm

Re: Employment numbers: % of class v. number of students overall

Postby Watchenlaw » Mon Nov 25, 2013 9:20 pm

Tiago, you make some good points, and these are just my musings on the topic, but I think my approach has more going for it that you give it credit for.

On your first point, medians are useful but 25th/75th percentiles are more informative. According to this site, Columbia's LSAT percentiles are 170-174 (25th-75th) while Duke's are 166-170, which is significant when you look at what those numbers represent. They mean that 25% of the students who Columbia admits are in the top top 1.79% of all LSAT test takers and 25% are in the top 0.47% of test takers, while the top 75% admitted at Duke are only in the top 5.66% and only 25% are in the top 1.79%. I think this is an significant difference, at the very top end of the scale where even slight increases in scores are extremely difficult. Maybe you dont think that this difference is going to translate into meaningfully more talanted classmates and stiffer competition, which is reasonable, I just disagree.

Even so, Columbia VS. Duke is just a particularly good example for me and probably suggests my approach is more robust than it actually is, which wasn't deliberate--I chose the schools in my examples more or less by random and did not look at the actual numbers for any of the schools I picked--and it will not always turn out like this. For instance, in the other example I picked, Vandy VS. Virgina, the LSATs 25/75 percentiles much closer and the GPAs are no so out of wack that it suggests to me a huge difference in ability (I tend to think GPA is less important, however, so others might disagree). In this case, if Virginia had meaningfully better employment outcomes, all else being equal I would go there. (In my original example, they both have the same employment outcomes, and Vandy is still the better choice in my opinion if that is the case.)

Second: this is a fair point. I guess what I was getting at is that people here sometimes suggest that you should go to the best school you get into because it will have better employment options and so there is more room for error if you don't do well. I think this is an illusion; there may be more employment opportunities at a higher-ranked schools, but the competition to get them is still stiffer, so you really dont have greater security if you don't do as well as you hoped at a top school.

On your third point: I guess we have already established that we disagree about whether small rankings differences can translate into significant differences in the ability of the students at the schools, but I don't think I follow you concerning larger differences. Are you suggesting that the difference in ability between students at Iowa and UPenn (my example) does make up for the difference in employment opportunities? You seem to suggest that in cases of larger gaps "the analysis actually works." Given that the analysis is "determining when your job opportunities/chances actually increase when you go to a lower-ranked school," that is the only conclusion I can glean from what you said, but it is pretty extreme compared to your other two very reasonable criticisms.

User avatar
Tiago Splitter
Posts: 15513
Joined: Tue Jun 28, 2011 1:20 am

Re: Employment numbers: % of class v. number of students overall

Postby Tiago Splitter » Mon Nov 25, 2013 10:33 pm

Watchenlaw wrote:On your third point: I guess we have already established that we disagree about whether small rankings differences can translate into significant differences in the ability of the students at the schools, but I don't think I follow you concerning larger differences. Are you suggesting that the difference in ability between students at Iowa and UPenn (my example) does make up for the difference in employment opportunities? You seem to suggest that in cases of larger gaps "the analysis actually works." Given that the analysis is "determining when your job opportunities/chances actually increase when you go to a lower-ranked school," that is the only conclusion I can glean from what you said, but it is pretty extreme compared to your other two very reasonable criticisms.

To clarify: I think the opportunity to place better within the class only exists when there are significant differences in matriculant qualify. I certainly would expect a median student to do better than median at Iowa. UPenn vs. Cornell? I'd expect the difference to be too small to matter.

Ultimately, I agree that as you go down the US News rankings generally your placement within the class will improve, but I think the differences are too small to make up for the rapid drop off in employment prospects. The only time it might be worth thinking about is if you have particular employment goals that make taking the full ride at the local regional a good idea.




Return to “Choosing a Law School”

Who is online

Users browsing this forum: andpeggy, Baidu [Spider], Mhr, Yahoo [Bot] and 4 guests