California's move to a two-day bar exam might affect some schools more than others

I was among the first to discuss California's planned move from a three-day bar exam to a two-day bar exam. The first two-day exam will occur in the July 2017 administration.

The old three-day model weighted the Multistate Bar Exam component (the 6-hour multiple choice test) at about 1/3 of the overall score, and the other two days of essays as about 2/3 of the overall score. When the bar studied the issues, it found little difference in assessing aptitute or in scoring between a 1/3-2/3 model and a two-day bar where both sections would be weighted roughly equally (as most states do).

That's true at the macro level. For individual test-takers, of course, that can vary wildly. And even at the school level, we may see somewhat noticeable differences between the MBE scores and the essay scores.

Thanks to a pretty sizeable disclosure from the California bar, we can assess how individual schools fared on the bar, and what their scores would look like if scored under the July 2017 1/2-1/2 model.

This, of course, has many limitations, which I'll start listing here. First, these are the mean scores; they correlate highly with pass rates, but not perfectly. Note that Stanford's mean score blows all other schools out of the water, but its first-time pass rate is only a few percentage points better than others. That means movement up or down in the mean scores would likely improve or worsen the pass rate, but in measures not immediately ascertainable. Second, just because the bar was scored this way in July 2016 does not mean we would expect graduates of these schools to perform similarly in 2017. Indeed, evidence like this would probably drive a change in bar study habits! Graduates would be inclined to focus more attention on the MBE and less attention on the essays, which would change the scores in unknown ways.

The chart at the right shows in red circles what schools' mean scores were this July under the 1/3-2/3 scoring model. The blue circles are what the scores would have been under the 1/2-1/2 model. (Recall that a passing score in California is a 1440.) As you can see, there is almost no difference for most schools. I flagged four schools that might see the biggest changes--San Diego's for the better; and Irvine, San Francisco, and Thomas Jefferson for the worse.

And recall the caveats above--this does not mean it will translate into demonstrable differences in the pass rate, and pass performance is not an indicator of future success. This is particularly school for the three schools I identified that might expect lower means--Irvine is well above the passing score, and San Francisco and Thomas Jefferson are well below it, meaning marginal differences in the mean score would probably affect very few. (For schools closer to the 1440 score, we might expect slightly larger differences, again with the significant caveats listed above about the limited value of using the means.) But it should certainly shift attention in graduate preparation next summer--and whether that changes scores remains to be seen.