Subscription pricing
Any idea of how good a predictor the 7sage law school admissions predictor is and where it's getting its data from?
0
Any idea of how good a predictor the 7sage law school admissions predictor is and where it's getting its data from?
9 comments
@mstng9878717 of course there's a chance, but I have no reason to think it is.
Still, proceed with caution. I would advise you to add a couple safety and target schools based solely on your numbers, not on the predictor.
@davidbusis895 said:
Hi everyone,
Accurate-ish is a good description. You should crosscheck our model with https://officialguide.lsac.org/release/ugpalsat/ugpalsat.aspx. In most cases, we're pretty similar.
@leahbeuk911 we looked at about 400000 applications from LSN, so the sample size isn't small, although it's far from perfect.
@leahbeuk911 we draw on the same data that mylsn.info draws on, although we have our own statistical model.
In general, I think you should use the model to give you a sense of your odds, but you shouldn't take it literally. I have a feeling that it exaggerates the importance of applying early in the cycle or applying ED for below-median applicants.
Our predictor isn't sophisticated enough to model this, but from an admissions officers’ perspective, ED is a great opportunity to lock in applicants with (1) median-supporting LSAT scores or GPAs, or (2) (less commonly) exceptional diversity. Because some applicants with LSAT scores or GPAs that hit the median will be rejected later in the cycle, we think that ED provides a significant tailwind to candidates who hit at least one of the numbers. (This may also be why our model sees such a strong effect from ED.) On the other hand, we expect that most candidates who have below-median LSAT scores and GPAs will be punted into the RD pool or (less commonly) rejected.
Thanks! Is there any chance the predictor could be over-emphasizing the admissions advantage of being an URM? I am one and I was very surprised at the percentage difference between URM and non-URM.
Hi everyone,
Accurate-ish is a good description. You should crosscheck our model with https://officialguide.lsac.org/release/ugpalsat/ugpalsat.aspx. In most cases, we're pretty similar.
@leahbeuk911 we looked at about 400000 applications from LSN, so the sample size isn't small, although it's far from perfect.
@leahbeuk911 we draw on the same data that mylsn.info draws on, although we have our own statistical model.
In general, I think you should use the model to give you a sense of your odds, but you shouldn't take it literally. I have a feeling that it exaggerates the importance of applying early in the cycle or applying ED for below-median applicants.
Our predictor isn't sophisticated enough to model this, but from an admissions officers’ perspective, ED is a great opportunity to lock in applicants with (1) median-supporting LSAT scores or GPAs, or (2) (less commonly) exceptional diversity. Because some applicants with LSAT scores or GPAs that hit the median will be rejected later in the cycle, we think that ED provides a significant tailwind to candidates who hit at least one of the numbers. (This may also be why our model sees such a strong effect from ED.) On the other hand, we expect that most candidates who have below-median LSAT scores and GPAs will be punted into the RD pool or (less commonly) rejected.
I think it's accurate-ish. I like to use the mylsn.info one a little more, because I like being able to click through and check out the relevant LSN profiles and see scholarship money awarded.
Actually, what I mean is that models are good for making broad generalities. However, you cannot take that broad generality and expect identical outcomes when you make a prediction of a single person (yourself in this instance).
Single person / n=1 refering to the output of a model (a prediction about your outcome) not the input / data upon which the model operates.
@mstng9878717 said:
@ayaladeziree615 said:
I like the 7Sage predictor. Just remember, all statistical models break down when n=1.
what do you mean?
@ayaladeziree615 means models are useless if the data they're based on has small sample size (n=1 sample data point).
@ayaladeziree615 said:
I like the 7Sage predictor. Just remember, all statistical models break down when n=1.
what do you mean?
I like the 7Sage predictor. Just remember, all statistical models break down when n=1.
https://classic.7sage.com/admissions/lesson/affects-chances-getting-law-school/ this explains it all!