Law School Predictor: The Thread" Forum

[YCrevolution]

..

[YCrevolution]

..

[Tangerine Gleam]

Thanks! Turns out I just hadn't re-loaded my web browser to reflect the update.

YCrevolution -- thank you so much for making this program. It's an incredible resource, and I hope you don't think we take it for granted. You rule.

[YCrevolution]

..

[cendien]

Hey - saw the discussion on Wash U. I pulled the info on that from this document: --LinkRemoved--

But it does appear as if their program may have ended. I'll take them off the list for now and maybe email their admissions department to get the final word on the program! Thanks for the catch guys.

[catharsis]

YC first let me start by saying how amazing you are. Ive sat down with my family and girlfriend to discuss your predictions given by the calculator and its been really awesome.

I know this has been spoken for already but for my peace of my mind Im at a "STRONG CONSIDER" for Texas. Given past data and your knowledge of how accurate it is and that I apply the earliest day possible in the cycle what is the percentage chance approx. of gaining admittance? I know it may be hard to put a number on it but can you try? Thanks!

[YCrevolution]

..

[YCrevolution]

..

[CyLaw]

If it is still around at the end of this coming cycle, I will definitely fill it out to help.

[CyLaw]

May I also suggest gathering information on what the person thinks of their softs and ask them to list them in a general fashion in the comments. You can then check for a correlation between people who thought they had great softs and people who got into schools that LSP predicts is a deny.

I know that you can't predict everything, but since you are gathering info now, might as well get as much as you can for later use.

[YCrevolution]

..

[SamSeaborn2016]

YC,

I just gotta say the amount of time and care you have put into this is outstanding. Kudos!!!

[OperaSoprano]

Happy to oblige. I had a really odd cycle, though, so I'm not sure how much help my info will be.

[Bronte]

EDIT: I've just included "Softs" with "Comments" on the submission form. Thanks.

If I can make a suggestion, using a numerical rating of one's softs might be more interesting. Obviously, it's still subjective, but a 1-5 numerical scale would make it easier to try to find a correlation. They should consider adding this on LSN, too.

[YCrevolution]

..

[YCrevolution]

..

[cpaige]

Ba-ba-bump for the early morning TLS members.

I've gotten about 20 responses so far that I can use. The more, the merrier, of course.

A new online form is now available (LinkRemoved) at Law School Predictor that asks for your admission information and decisions. By taking a minute to complete and submit this form, you’ll be helping make LSP even more accurate for future applicants. LSP is already cost-free and ad-free, so a small contribution of your time and information is appreciated greatly. The information you submit will not be connected to your real-life identity and submitting your TLS username is optional.

great idea. LSP has been a great tool for me, so i'll definitely add my cycle once it's over.

[pissantvache]

Perhaps for softs you could use a number of binary check boxes (or checkboxes, experience, and prestige). For example, someone with 3 years of work experience at McKinsey or Goldman Sachs probably brings an edge, even if simply because of future employability. You could always look at it through Vault Prestige level as well--v3 is better than v20, etc. That'll allow you to quantify things like work experience (rather than relying on the Fratty Events supervisors); similarly, if people are scholarship recipients, the amount of the award could come into play (on the theory that bigger awards correlate positively to more deserving applicants).

That would, of course, also help to solve the problem of recurring 'what constitutes good softs' threads.

[PrayingforHYS]

hey, great stuff, great contribution

i have a recommendation that I think could cement this as "awesomely amazing"

when you receive something like "consider" or "strong consider", that's nice, but it's also a little broad. I think it would be amazing if it calculated a percentage for your admission (i.e. you have an X% chance of getting into school Y)

I don't know much about designing things like these, but I'd think that it wouldn't be as hard as it sounds to set something like this up. For example, say you're a 174/3.85. At HLS, you're listed as "consider" and your score is 12.108. 25% is 11.730, 50% is 12.036 and 75% is 12.342. Therefore, you're .308 above the 25th percentile, .072 above 50 and .234 below 75. As a result, there has to be some kind of formula that calculates exactly what percentage your scores fall into and lists them accordingly.

For example, 12.342 (75%)-12.036 (50%) = .306 / 24 (the # of spots between 50% and 75%) = .01275

i.e. for each .01275 you are above the 50% marker, you see an increase of 1% in your admission chances (for the purpose of this indicator)

Thus, a 174/3.85 would be 12.108-12.036 (50%) = .072 / .01275 = 5.65 (rounded) -- Thus, you'd have a 55.65% chance at HLS

This isn't genius-level math by any means, and I just wanted to write it out to show my line of thinking...I'm just not sure how you'd program a computer to do this (though I'm sure it's possible with your knowledge)

anyhow, I believe that this would be immensely more helpful than "weak/strong/regular consider"

just my .02 -- great program.

[Bronte]

when you receive something like "consider" or "strong consider", that's nice, but it's also a little broad. I think it would be amazing if it calculated a percentage for your admission (i.e. you have an X% chance of getting into school Y)

Did you know that there's a couple other calculators that already give percentage probabilities? Using the "deny-consider-admit" readout is part of what makes LSP unique. Further, the technique you've described does not provide a probability, but instead an interpolated (i.e., linearly estimated) percentile. It's similar, but, whereas a percentile says "X% of people got in with with numbers at or below your numbers," a probability says "X out of Y people got in with your numbers."

That being said, I added this feature just now to my Excel version of LSP using the TREND() function. The percentiles actually read out very similar to the probabilities produced by LSC, albeit some are much more conservative. This would actually be an interesting feature to add to LSP: an interpolated percentile rank.

[bissey]

The GPA calculator on the law school predictor is off by .01 (.01 higher than it really is).

[YCrevolution]

..

[bissey]

I tried it again and it worked. I retract my statement

The GPA calculator on the law school predictor is off by .01 (.01 higher than it really is).

Any specific combination of numbers/grades/etc.?

If you're crunching a lot of numbers, there's probably going to be a rounding error at some point. There also a hidden 0.0000000001 boost to quality points so that there's not an initial divide-by-zero error (it would update once "Calculate" was clicked, but it looked ugly).

I have replies for the other posts that I'll get to in a few minutes.

[YCrevolution]

..

[YCrevolution]

..