Harvard Law, Quick Analysis of LSN Database Numbers Forum

[Voxius]

For anyone that's interested, I was curious to see how the self-reported LSN data corresponds with the actual distribution of LSAT scores and GPAs (for anyone who hasn't seen http://www.mylsn.info--check it out; it is another probability calculator but allows you to splice the data by application cycle). Granted, the analysis was fairly crude, but I haven't seen this repeated anywhere. I took the list of self report data from LSN for Harvard (those that applied in 2011 and entered last fall, since that's the most recent data available on HLS's website for comparison), threw it into Excel, and calculated the implied 25%/50%/75% cutoffs for the LSAT and GPA. The results were as follows:

LSAT
75% - 177
50% - 176
25% - 172

GPA
75% - 4.00
50% - 3.90
25% - 3.80

The actual stats listed on HLS's website for that class are:

LSAT
75% - 175
50% - (Median Not Shown)
25% - 170

GPA
75% - 3.95
50% - (Median Not Shown)
25% - 3.77

There are a few different inferences that can be drawn, but generally speaking, it looks as though comparing numbers directly with what's listed on LSN (given self-reporting bias, the likely skew toward more involved students, etc.) would lead to a slightly understated understanding (or probability if using one of the prediction calculators based on LSN) of one's chances.

Thoughts? and/or Does anyone else have ideas for analysis?

[20141023]

.

[LSATSCORES2012]

Very nice first post Keep in mind that those numbers, yours and the ones in Reg's thing, are for the admits, whereas the numbers published by H are for matriculants, or the subsection of those admits who actually choose to attend (and matriculants probably have lesser numbers than admits, overall).

But your conclusion is definitely correct: LSN is not representative. The chart below shows the ratio of the amount of people on LSN with a given LSAT score compared to what would be expected if LSN were representative of all LSAT scores. (Of course, it is probably the case that a lot of people with very low LSAT scores just never end up applying, so perhaps the number of applicants with really low LSAT scores is relatively less than the number of people who get those scores compared to the number of applicants who apply with high scores vs those who get high scores.) It makes sense, too, because if you get a bad LSAT score you're probably less likely to share it. Even more, the people who go on TLS are frequently those who go on LSN and also those who have access to all of TLS's prep resources. (Then again, there are also lots of people who go on TLS and don't post or go on LSN.) In any case, I definitely think you're right, and that it makes sense that you're right.

--ImageRemoved--

[Clearly]

The data are different things! Lsn is admits, website shows attending students. Lsn data is going to be higher because many of those highest students that got into to Harvard on LSN won't be listed in Harvard data because they are at Yale and Stanford, or got full rides to CC