Law School Stats Analysis Blog Forum

[sinfiery]

Nice! About as expected though the female boost was unexpected. Had a hunch about applying early based on my cycle, good to see the data back it up.

The question was in my mind because of this thread
http://www.top-law-schools.com/forums/v ... 9&t=210221



I'll let you do the honors

[John_rizzy_rawls]

Both really good posts el, keep em coming.

[Lavitz]

Nice! About as expected though the female boost was unexpected. Had a hunch about applying early based on my cycle, good to see the data back it up.

I'm not sure if I'm reading it correctly or not, but isn't Berkeley's female boost below average for the T-14? I thought the first blog post showed that for otherwise identical applicants, females had a 48.6% higher chance of admission, but Berkeley shows a 40.5% boost.

I thought the non-traditional boost was unexpected because the first post said they had no statistically significant effect overall, whereas Berkeley shows a 56.4% boost (80.9% in the second model).

I don't know much about stats so I may be looking at this the wrong way.


I think it'll be interesting to compare the schools against each other and see which school shows the largest boost for each factor. Biggest increase in odds for 1 point LSAT increase, biggest increase in odds for female applicants, etc.

[elterrible78]

Nice! About as expected though the female boost was unexpected. Had a hunch about applying early based on my cycle, good to see the data back it up.

I'm not sure if I'm reading it correctly or not, but isn't Berkeley's female boost below average for the T-14? I thought the first blog post showed that for otherwise identical applicants, females had a 48.6% higher chance of admission, but Berkeley shows a 40.5% boost.

I thought the non-traditional boost was unexpected because the first post said they had no statistically significant effect overall, whereas Berkeley shows a 56.4% boost (80.9% in the second model).

I don't know much about stats so I may be looking at this the wrong way.


I think it'll be interesting to compare the schools against each other and see which school shows the largest boost for each factor. Biggest increase in odds for 1 point LSAT increase, biggest increase in odds for female applicants, etc.

Nah, you're not looking at it wrong at all. I guess I was just surprised because I thought, given the hippy-dippy rap Berkeley sometimes gets, that the boost for females would be bigger. Also, in my mind, I'm comparing it to other schools that I have seen lately that give no boost whatsoever for females, non-trads, etc. And comparing the schools is kind of the longer-term goal I have. I think it'll be interesting to compare within the T14, but also the T14 vs. Non-T14 schools, because from what I'm seeing so far, there do appear to be marked differences.

[Lavitz]

Nah, you're not looking at it wrong at all. I guess I was just surprised because I thought, given the hippy-dippy rap Berkeley sometimes gets, that the boost for females would be bigger.

To be honest, that's what I was expecting too. That's why I thought I would hear something like "the below-average boost for females surprised me" rather than "the increase for female applicants surprised me." I couldn't tell what surprised you about it.

[lhanvt13]

Cool stuff. Thanks!

[elterrible78]

Looking for opinions on this:

When I run regressions on scholarship information, I can get equivalent boosts in scholarship money for each additional LSAT point, URM status, etc. Included is a constant, or "starting point" which is typically a negative number of hundreds of thousands of dollars. I am thinking about including a "calculator" on my school profile pages, where you can punch in your own values, and it'll spit a predicted scholarship amount.

While I think this would be an interesting "toy", there are some serious problems. For one, although it varies by school, the models are not that predictive, which just means there is clearly a lot of stuff at play in scholarship decisions that I am not accounting for (great personal statements, letters of rec, interesting backgrounds, etc). The model I use only accounts for about 20% of the variability for Chicago's scholarships, for example, which means 80% of the scholarship decisions comes down to "other stuff." So just in general, these calculators wouldn't be super predictive (mine, for example, undershot my actual scholarship by $50k...I haven't done it for other schools yet, but it would be interesting to see).

The big problem, though, comes down to user-responsibility. I would include a disclaimer detailing just what problems are inherent, and a heavy dose of "this is just a cool gizmo, do not bank on this!!!" But you know, there are always those people, right? For example, the Chicago calculator would spit out $9,412 at a female applicant with a 155 and a 4.0, yet the chances of that person even being admitted are next to nothing. The last thing I want to do is put ideas in people's heads.

So...yay or nay to this type of thing? Does the little value it provides as an interesting curiosity offset the danger of people misusing?

[Bobnoxious]

Looking for opinions on this:

I am thinking about including a "calculator" on my school profile pages, where you can punch in your own values, and it'll spit a predicted scholarship amount.

While I think this would be an interesting "toy", there are some serious problems. For one, although it varies by school, the models are not that predictive, which just means there is clearly a lot of stuff at play in scholarship decisions that I am not accounting for (great personal statements, letters of rec, interesting backgrounds, etc). The model I use only accounts for about 20% of the variability for Chicago's scholarships, for example, which means 80% of the scholarship decisions comes down to "other stuff." So just in general, these calculators wouldn't be super predictive (mine, for example, undershot my actual scholarship by $50k...I haven't done it for other schools yet, but it would be interesting to see).

Hell YEAH! I'd love to see something like this, but more for the non-T14, especially the stronger regional state schools.

[elterrible78]

Looking for opinions on this:

I am thinking about including a "calculator" on my school profile pages, where you can punch in your own values, and it'll spit a predicted scholarship amount.

While I think this would be an interesting "toy", there are some serious problems. For one, although it varies by school, the models are not that predictive, which just means there is clearly a lot of stuff at play in scholarship decisions that I am not accounting for (great personal statements, letters of rec, interesting backgrounds, etc). The model I use only accounts for about 20% of the variability for Chicago's scholarships, for example, which means 80% of the scholarship decisions comes down to "other stuff." So just in general, these calculators wouldn't be super predictive (mine, for example, undershot my actual scholarship by $50k...I haven't done it for other schools yet, but it would be interesting to see).

Hell YEAH! I'd love to see something like this, but more for the non-T14, especially the stronger regional state schools.

If people want it, I'll do it, and I'll do it for all the schools. Only thing is, I just don't want people using it and saying, "BAM! Gonna get $X at X-School!" I'll put the caveats on there, loud and clear, but I don't want people misled. In the meantime, if you've got a school or two in mind, PM me and I'll send you the info.

[elterrible78]

Profile for the University of Alabama up (remember, I'm going alphabetically unless I get special requests). Also, an unfortunate note on money.

http://www.admissionsbythenumbers.com

[Lavitz]

Profile for the University of Alabama up (remember, I'm going alphabetically unless I get special requests). Also, an unfortunate note on money.

http://www.admissionsbythenumbers.com

I don't see the Alabama profile on the page list--I just see Berkeley there for now.

[elterrible78]

Profile for the University of Alabama up (remember, I'm going alphabetically unless I get special requests). Also, an unfortunate note on money.

http://www.admissionsbythenumbers.com

I don't see the Alabama profile on the page list--I just see Berkeley there for now.

Sumbitch. Thanks, ill take a look when I get where I'm going.

EDIT: Fixed, and thanks again, man.

[domino]

Nice! About as expected though the female boost was unexpected.

I was surprised at how big the boost was too. I'd be curious to see what it is for Y. I think their quota-less admissions leave them with about a 50-50 gender split; if they also receive fewer applications from women, that would mean there is an advantage, but not sure what the mechanism would be given their "faculty do whatever you want" system.

[elterrible78]

Nice! About as expected though the female boost was unexpected.

I was surprised at how big the boost was too. I'd be curious to see what it is for Y. I think their quota-less admissions leave them with about a 50-50 gender split; if they also receive fewer applications from women, that would mean there is an advantage, but not sure what the mechanism would be given their "faculty do whatever you want" system.

Okay, now I'm curious. I'll at least do a basic workup on Y before I hit the rack.

[elterrible78]

Nice! About as expected though the female boost was unexpected.

I was surprised at how big the boost was too. I'd be curious to see what it is for Y. I think their quota-less admissions leave them with about a 50-50 gender split; if they also receive fewer applications from women, that would mean there is an advantage, but not sure what the mechanism would be given their "faculty do whatever you want" system.

Okay, now I'm curious. I'll at least do a basic workup on Y before I hit the rack.

Okay, Yale is up for anyone interested in taking a look. I have UC Davis ready to go, too, but probably won't get around to it until tomorrow. As always, thanks for the support, and I hope you guys find this interesting. Once I start comparing schools, it should be even more useful. Again, I'm wide open to requests...as long as I can figure out how to do it!

[domino]

Nice! About as expected though the female boost was unexpected.

I was surprised at how big the boost was too. I'd be curious to see what it is for Y. I think their quota-less admissions leave them with about a 50-50 gender split; if they also receive fewer applications from women, that would mean there is an advantage, but not sure what the mechanism would be given their "faculty do whatever you want" system.

Okay, now I'm curious. I'll at least do a basic workup on Y before I hit the rack.

Okay, Yale is up for anyone interested in taking a look. I have UC Davis ready to go, too, but probably won't get around to it until tomorrow. As always, thanks for the support, and I hope you guys find this interesting. Once I start comparing schools, it should be even more useful. Again, I'm wide open to requests...as long as I can figure out how to do it!

Yay!

[Instinctive]

Cool stuff, thanks!

[TatNurner]

Tagging to get in on the awesomeness.

[elterrible78]

Tagging to get in on the awesomeness.

Throwing out a special request for Harvard, if possible.

I'm just about there, so that'll be the next profile I do...will try to get it up there by tonight!

[Lavitz]

Tagging to get in on the awesomeness.

Throwing out a special request for Harvard, if possible.

I'm just about there, so that'll be the next profile I do...will try to get it up there by tonight!

This should be interesting.

[californiauser]

OP you should go T14 first then alphabetical

[elterrible78]

OP you should go T14 first then alphabetical

That wouldn't be very egalitarian of me, would it?

Nah, it seems like there's a lot of interest, so I'll probably do those profiles first, I guess. It might take a while anyway. I'm working on HLS as we speak, but some stuff came up, so I'm not sure if it'll get finished tonight. Tomorrow, definitely.

[Micdiddy]

Special request for McGeorge.

[elterrible78]

Special request for McGeorge.

Done. The only factors I could find that correlated with success in admission were a verifiable pulse and a signed master promissory note.

[elterrible78]

Okay, folks, Harvard and Columbia profiles are up, along with a list of projects I have planned for the future (and a request for other ideas). As always, thanks, and I hope you find the info useful.