Law School Predictor: The Thread"

[OperaSoprano]

I might be misreading the data, but it appears that the people who've self selected to report their numbers have done much better than they "should have." Are we all that much savvier than the average law school applicant?

In particular, I'm thinking of applicants who want to attend schools in their strong consider range. In my experience, if someone is able to demonstrate that he or she really wants in at a strong consider school, that person will very likely be admitted. This is what happened in my own cycle, and I've seen it happen over and over again.

Is there any way to account for "persistence" factors like visiting, interviewing, Why X letters, and LOCIs? IMO these can often trump softs, and at the end of the cycle, when WL decisions need to be made quickly, they can occasionally even trump numbers. Deadatheist's recent admission to Hastings (160/3.63, non URM) is a perfect example. She worked for this, probably even harder than I worked for Fordham, though LSP called her a weak consider. I believe her persistence must have been a contributing factor in her WL admission.

I'm not sure how you could quantify this, but do you think it's credited?

I mean, I'm sure you remember the dramatic mess my cycle was, and I was decently above the median.

The average LSN application would list an applicant (who received a decision)* with a LSAT of 162 and a GPA of 3.47.
*This is different than the average applicant since many applicants applied to multiple schools, and I haven't filtered that out.

Persistence for applicants, particularly waitlisted applicants, probably helps a bit (quite a bit if you're really persistent), but unfortunately I don't have any easy way to account for it. Looking at the info submitted through LSP, some people definitely do far better (and in some cases, far worse) than their numbers would suggest.

Great, I'm pretty average. No huge surprises there. (My mom just yelled at me for referring to my 164 as an LSAT fail. We can't all get a 98th percentile MCAT score, mom.)

I meant that the average applicant (not necessarily an applicant with average numbers) probably knows a lot less about this process than we do. If such knowledge is correctly applied, miracles are very possible.

I know you probably can't incorporate the persistence factor into your spreadsheet, but don't you ever look at the data and wonder what accounts for such surprises? I'm endlessly fascinated and honored when people choose to tell me their stories.

[CyLaw]

Good catch. I really dropped the ball on that one.

It is so bad that I have gotten to the point on TLS where I can't tell if you are being sarcastic or not.

[lhfan]

Incredible job! YCrevolution 2012

[YCrevolution]

..

[Ken]

Thanks for continually improving what was already a great tool and asset to TLS users!

[OperaSoprano]

Everyone loves YCRev!!!

[YCrevolution]

..

[YCrevolution]

..

[jackassjim]

Good catch. I really dropped the ball on that one.

It is so bad that I have gotten to the point on TLS where I can't tell if you are being sarcastic or not.

No, no. I re-read the original post, and you seem to be right: the new test data is not the same as the original one.

[smith8586]

Someday, I hope to figure out why UVA let me in.

Perhaps because you do awesome things like create LSP. I've had more fun playing with this than I care to admit. LSP coupled with the advice I've been given on the board has helped me really re-think my approach to apps. Thanks dude!

[jackassjim]

I'm curious to know how your models would stack up against models that would be purely computed form available data.

For example, you would put all individual entries for Vandy in a dataset, and run a probit regression on those using gpa and lsat as regressors. Then, you would use the coefficients that this would yield to predict the results of hypothetical or future applicants. Maybe bypassing the index formulas and building a strictly data-driven model would be a good idea...

About the pseudo-R2, I don't know how you think you would compute that. Somehow, when I wrote the preceding post, I didn't have index formulas in mind at all, and for some reason I assumed that you had used a technique similar to what I explain in the paragraph above. Most statistical packages (I use R, but Stata does it too) give you the pseudo-R2 automatically when you run a probit. Not sure how this would work in the framework you are using to build LSP

In any case, if you are interested in seeing the different results that the above type of procedure could give us, you can shoot me some data. It should be fairly trivial to run (R is quite good at manipulating and formatting data that is imported from excel).

But really, I have to say, that's amazing work YC! And Cylaw, kudos for collecting all this data!

[CyLaw]

I'm curious to know how your models would stack up against models that would be purely computed form available data.

If all goes to plan this coming semester, I actually plan on doing this with a Machine Learning algorithm for my AI course. Will not be done until December though.

[YCrevolution]

..

[philip.platt]

A closer look...
A possible explanation for the Strong/Weak Consider results is that the chance of being admitted (based on admission index scores and LSP adjustments) should look like a parabolic curve (since it should be a normal distribution), and, based on these accuracy results, a curve with a high and narrow peak in the center (the roughly median applicant).

curve with a high and narrow peak in the center = leptokurtic

you did statistics in a prior life, didn't you? this is really impressive - - props man

[CyLaw]

you did statistics in a prior life, didn't you? this is really impressive - - props man

Agreed, Did you take a lot of stats classes for your Poli Sci degree, or did you learn them just for fun?

[YCrevolution]

..

[OperaSoprano]

Someday, I hope to figure out why UVA let me in.

The chair of the math department at UVA asked to borrow you. He wants you to prove Goldbach's conjecture, but you can probably do that between orientation and the first day of classes.

[jackassjim]

The results of a logistic regression run on the Michigan admissions data from LSN are below

Each extra LSAT point increases the probability of acceptance by almost 18% (exp(0.16416)=1.178403).

I don't quite understand the formulas on your website, so I can't figure out how this compares to what your output gives.

Code: Select all

Coefficients:
               Estimate Std. Error z value Pr(>|z|)    
(Intercept)   -33.76765    3.91254  -8.631  < 2e-16 ***
michigan$LSAT   0.16416    0.02094   7.839 4.56e-15 ***
michigan$GPA    1.59927    0.36019   4.440 8.99e-06 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 

[vanwinkle]

I know you probably can't incorporate the persistence factor into your spreadsheet, but don't you ever look at the data and wonder what accounts for such surprises? I'm endlessly fascinated and honored when people choose to tell me their stories.

Oooh, storytime!

My own prediction calculator gives me a (very) "weak consider" prediction for UVA, even with the new binding ED feature enabled (since I applied to UVA ED). But fortunately for me, I managed to beat my LSP odds. I often try and use this anecdote to illustrate that LSP doesn't always make perfect predictions.

Someday, I hope to figure out why UVA let me in.

I'd love for you to figure out why UVA let me in as well. Your predictor program lists me as a DENY (and only a WEAK CONSIDER if I check URM, which everyone keeps telling me I'm not). I'm waiting for a phone call telling me they've made a mistake and they're taking back the admission offer (and the scholarship money!).

It was a generic application, too, and RD. I didn't even mention UVA in my PS, so OperaSoprano's theory about persistence or pursuit won't explain me.

[YCrevolution]

..

[OperaSoprano]

I know you probably can't incorporate the persistence factor into your spreadsheet, but don't you ever look at the data and wonder what accounts for such surprises? I'm endlessly fascinated and honored when people choose to tell me their stories.

Oooh, storytime!

My own prediction calculator gives me a (very) "weak consider" prediction for UVA, even with the new binding ED feature enabled (since I applied to UVA ED). But fortunately for me, I managed to beat my LSP odds. I often try and use this anecdote to illustrate that LSP doesn't always make perfect predictions.

Someday, I hope to figure out why UVA let me in.

I'd love for you to figure out why UVA let me in as well. Your predictor program lists me as a DENY (and only a WEAK CONSIDER if I check URM, which everyone keeps telling me I'm not). I'm waiting for a phone call telling me they've made a mistake and they're taking back the admission offer (and the scholarship money!).

It was a generic application, too, and RD. I didn't even mention UVA in my PS, so OperaSoprano's theory about persistence or pursuit won't explain me.

Unusual majors unite!

Most people on TLS will scoff at this, but there is something I refer to as the interesting applicant bump. This is also a reasonable descriptor for me. I do NOT have great softs, but, by virtue of my academic background, I was unique, or near unique, in the applicant pool. It should be noted that this cuts both ways. I was convinced that I was toxic until the head of my own admissions office told me otherwise.

It could have been something else in your background; It could have been nearly anything. If you're thrilled, though, UVA did it right.

[YCrevolution]

..

[YCrevolution]

..

[vanwinkle]

Unusual majors unite!

Most people on TLS will scoff at this, but there is something I refer to as the interesting applicant bump. This is also a reasonable descriptor for me. I do NOT have great softs, but, by virtue of my academic background, I was unique, or near unique, in the applicant pool. It should be noted that this cuts both ways. I was convinced that I was toxic until the head of my own admissions office told me otherwise.

It could have been something else in your background; It could have been nearly anything. If you're thrilled, though, UVA did it right.

Oh, I probably had the most interesting application ever. My softs were extreme. The problem is that while I had some awesomely unique things to talk about (a sleeping disorder that was only discovered and treated two years ago), I also had some huge negatives (withdrawing from college three times to avoid failing classes, a semester with two F's, inability to hold down a job for a while, etc.). These are all things that went into my application, too. I just talked about everything, in a "hopefully someone will bet on me in spite of the bad school record" kind of way.

I honestly wasn't expecting T14 schools to be the "betting" type though. I mean, they're bombarded by so many applicants with much stronger GPAs? Maybe it was a "film major bump".

When UVA sent the email asking me to call them (telling me I was admitted, but I didn't know that then), I was convinced it was because I had done something so horribly wrong on my application that they felt the need to tell me about it.

Haha! I had read enough on TLS to know that it was probably an acceptance phone call... But I also kept going, "That can't be right, they're not going to accept me. Maybe I forgot to fill out something on my application."

[OperaSoprano]

Oooh, storytime!

My own prediction calculator gives me a (very) "weak consider" prediction for UVA, even with the new binding ED feature enabled (since I applied to UVA ED). But fortunately for me, I managed to beat my LSP odds. I often try and use this anecdote to illustrate that LSP doesn't always make perfect predictions.

Someday, I hope to figure out why UVA let me in.

I'd love for you to figure out why UVA let me in as well. Your predictor program lists me as a DENY (and only a WEAK CONSIDER if I check URM, which everyone keeps telling me I'm not). I'm waiting for a phone call telling me they've made a mistake and they're taking back the admission offer (and the scholarship money!).

It was a generic application, too, and RD. I didn't even mention UVA in my PS, so OperaSoprano's theory about persistence or pursuit won't explain me.

Unusual majors unite!

Most people on TLS will scoff at this, but there is something I refer to as the interesting applicant bump. This is also a reasonable descriptor for me. I do NOT have great softs, but, by virtue of my academic background, I was unique, or near unique, in the applicant pool. It should be noted that this cuts both ways. I was convinced that I was toxic until the head of my own admissions office told me otherwise.

It could have been something else in your background; It could have been nearly anything. If you're thrilled, though, UVA did it right.

It's probably the best explanation for all three of us.

Thought so. As awesome as LSP is, real life is even better.

Well, except for the economy and stuff.