Harvard Law Accepting GRE Scores: An LSAT-Style Logical Fallacy

In the days since Harvard Law announced their decision to start accepting GRE scores, I've wondered if perhaps I was being unfair in my analysis - maybe Harvard does really want more international applicants, engineers, etc. for the sake of diversifying the student body.

And, yes, it's true, the change will lead to a wider and more diverse pool...but I can't shake the nagging feeling that something's fishy going on here.

Because, despite all that:

They haven't adequately made the argument that the GRE is a valid predictor of 1L grades the way the LSAT is!

The study they used to support their claim that the GRE is an equally valid predictor of 1L grades was only based on a sample of current and former HLS students (details here). This is a population likely to do well on a variety of standardized tests and likely to do well in law school - and they didn't study any other group!

To describe it in more formal terms, their argument is strictly correlational within a group of high-achieving, high-aptitude Harvard Law students - there's no control group! As a result, there's no way we can to separate those with only great test scores on each exam from those with only high GPAs and predict each group's 1L grades.

In short, from the information they've released, there's no indication they've made a rigorous attempt to study the GRE's validity as a standardized test independently of their own population.


To me, this suggests they're doing this for more self-interested reasons - rankings, applicant pool size, etc. In fact, I believe HLS hasn't attempted to demonstrate the GRE's predictive validity because they know (or at least suspect) that the GRE isn't an equally valid predictor of 1L grades. My guess is they expect the ABA to allow it anyway because law schools need the applicants. (And, if so, they're probably right about this.)


Why does this all matter? Who cares if the LSAT's a better indicator of 1L grades?

Because while anyone Harvard admits will likely be fine in the end, this change will create a domino effect at other lower-tier schools as they take advantage of the opportunity to expand their class sizes without suffering in the US News rankings.

The LSAT is almost certainly a better indicator of 1L grades - it functions as a barrier to prevent the admission of students likely to flunk out.

Similarly, it's likely a better indicator of students' ability to pass the bar exam - acting as a barrier to prevent the admission of students who might waste 3 years of their lives and over $100,000 in tuition money, yet still not be able to practice law in the end.

Top-tier applicants will likely still opt for the LSAT because they want to show they can ace it, while those who find it more difficult will do their best on the GRE. And many students will likely take both exams just to see which one they have more initial aptitude for. While math-phobia may deter some students from trying the GRE, I can't help wondering whether "Games-phobia" will deter applicants from the LSAT. Only time will tell.