Law School Predictor vs. LSAC predictor Forum

[js123]

The numbers I get from using Law School Predictor vs. what I get using LSAC prediction method is completely different. Can anyone explain why this is?

Also any recommendations on other predictors and/or success stories you've had using them would be greatly appreciated.

[georgej]

mylsn.info is a much better predictor because its based on actual admissions outcomes. Therefore the only bias is shows is due to the self-reported nature of the data.

[sandwich]

I'm definitely not interested in defending this too thoroughly, but...

I was under the impression that the LSAC predictor was based on actual admissions outcomes, but had complete data? I had thought that some of the reasons people don't use it as much are because some schools (several t-14s) opt out of that particular predictor, because they give score "bands" (like 0-10%—not really helpful) and because you don't see the sample size (i.e. you can't tell if that 95-100% chance is based on 5,000 data points or 1).

Anyone really should feel free to correct me about this. I only vaguely remember reading this a looong time ago.... back before I really realized how useless even complete data is to such a supersplitter


EDIT: This is definitely not to take away from the other reply. I do think that LSN is one of the best for the majority of the applicant pool, and so is probably the one you should be using.

[cron1834]

The predictors are TTT. LSAC tells me I have a better chance of getting in to NYU than Mich. Wtf? Law School Predictor is slightly more sensible, but LSN is obviously the way to go.

[cotiger]

The predictors do use the stats of the prior year's applicant pool to determine the index formula for each school. However, that means that the predictors assume a linear relationships between LSAT/GPA and acceptance rate (not the case) and ignore the impact of medians, which is why they are poor estimators of future acceptances.

[sandwich]

The predictors do use the stats of the prior year's applicant pool to determine the index formula for each school. However, that means that the predictors assume a linear relationships between LSAT/GPA and acceptance rate, ignoring the impact of medians, which is why they are poor estimators of future acceptances.

This is absolutely correct. I just finished a LONG pm to the op (sorry op!) in which I went into great detail about this. Just for all of you... feel free to amuse yourself by putting in a nice, sensible 3.5/170 in LSP and enjoy your dings from h/y/s/c/c/n, and then put in your 120 LSAT/ 9.0 GPA and enjoy your "admits" at CCN and your "strong considers" at HYS

[js123]

Just to add on to the information, I am a splitter. (Higher LSAT, Lower GPA)

[malleus discentium]

Cf. my thread here with the exact same question. There's a lengthy answer there. I don't know how satisfying I find it but maybe it'll be useful for you.