The Classroom Crack

The scores kept going up. Run the tape forward. In the late 2020s, professors start comparing notes.

It happens in faculty lounges and at bar conferences. Someone would say it. I’ve got the highest incoming medians in 20 years, but I’m giving more C’s than ever. Someone else would nod. A third person would pull out a phone and show the bar results. Nobody said it plainly, because saying it meant saying something about the students sitting in their classes right now.

What they were all thinking was that the score and the student had come apart.

The inflation chapters explained how medians rise without ability rising. (Median is the middle score in a class.) Retakes add points on average. Accommodations add more, because the system approves almost everyone who applies and never asks why. So a school’s reported median climbs, its ranking improves, and it admits a class that looks stronger on paper than any class before it. The admissions office sends out the press release. The dean gives the speech.

Then those students walk into a contracts class.

Law school doesn’t give time-and-a-half. The Socratic cold call happens once, live, in front of 80 people, and the professor isn’t waiting. The closed-book exam happens once, with no retakes. The skills that matter most in that room are processing speed, holding a complex argument in your head under pressure, and working memory in the moment. Those are exactly what a Socratic class and a 3-hour written exam demand.

So when part of a class carries scores that promised more reasoning ability than the students have, the gap walks into the building with them.

LSAC’s own research showed this. Sweeney et al. (2017) studied 121,378 non-accommodated LSAT takers against 880 extended-time takers with near-identical scores. The extended-time group’s 1L GPA ran about 0.43 standard deviations lower. That’s roughly 3 to 4 points on a 10-point scale. An earlier study found about 20% of extended-time takers landed in the bottom tenth of 1L GPA, against about 10% in the top fifth.¹

Here’s the caveat, stated plainly, because this is an ethics minefield. The studies mix disability status with extra time. They don’t track whether the same students got accommodation on 1L exams. The finding doesn’t prove any one student gamed the system, and this book isn’t saying that. The target is the system. No flag on the score. An approval rate of 98% in 2022-23 and 99% in 2024-25.² And a reason for schools to look away, since a higher accommodated median still counts in the rankings. The problem is a system that can’t tell the difference and doesn’t try, not students with real disabilities.

The gap doesn’t land on accommodated students alone. Curved grading spreads it.

Most law schools grade on a curve, so your grade is measured against the room. When part of the room shows up with inflated credentials, 2 things happen. Students whose scores were lower but real, who tested live with no help, get graded against a room that assumed a higher level of ability, so their real work looks weaker by comparison. And professors set exam difficulty to match a median that doesn’t match the real ability in the room. Exams come out too hard and cause a grading crisis, or too easy and produce grades that can’t tell anyone apart. Either way the signal gets worse.

The result is a middle-of-the-class grade spread that doesn’t tell anyone who can think like a lawyer, which is what grades are for.

Then comes the bar.

Bar passage is the cleanest proof that the inflation is hollow, because it’s the one outcome schools don’t control. Schools control the Socratic method, the curve, and what goes on the diploma, but not whether their graduates pass a standardized exam run by someone else.

At schools whose medians climbed 3 or 4 points over a decade, bar passage did not climb with them. At some schools it held flat. At others it slipped.³

That’s the stress test. If higher LSAT scores really meant higher reasoning ability, bar results would move with them, because reasoning is what the bar tests. When scores go up and bar passage doesn’t, the scores went up for reasons that don’t reflect reasoning. Retakes and accommodations are the candidates.

The NextGen bar exam arrives in July 2026 in 10 jurisdictions, with major states following through 2027 and 2028.⁴ It will sharpen this signal. NextGen tests applied legal reasoning across scenarios, 8 subject areas plus lawyering skills, under standard time, instead of multiple-choice recall of memorized rules. If students from inflated-median schools are really weaker at applied reasoning, NextGen will show it more directly than the old test did.

Schools are watching this nervously, and they should be.

The hiring partners got there first.

Lawyers who make associate-level hiring calls aren’t naive about grade inflation, credential inflation, or the gap between a number and a skill. They’ve been hiring out of law schools for 20 years. They notice when associates from schools with rising medians don’t perform better than they did 5 years ago, and when a first-year who walked in with a 168 can’t build an argument under time pressure that the one who walked in with a 163 can.

At firms that can afford to be picky, the response is a quiet but noticed shift toward law-school GPA.⁵ A 168 from a school with a 166 median looks fine. A 3.7 from that same school, in a class where the curve was set to a compromised cohort (entering class), looks better. GPA isn’t clean either, but it reflects real work in a real law school against real classmates, and firms trust that more than a headline score they’ve started to doubt.

This is the chain the inflation built. AI killed the personal statement as a signal. Schools leaned harder on the one number left, the LSAT. Inflation hollowed the LSAT. Schools started leaning harder on their own grade spreads. At every link, the signal that survives is demonstrated reasoning ability under conditions that can’t be gamed.

Here the story turns from diagnosis to direction.

If you’re an applicant reading this, the chapter isn’t an argument for despair. The students who win in this environment built a real score, one that reflects real reasoning ability developed over real months of hard preparation. They walk into a contracts class able to do what the score said, go on the curve with people who can’t, and build the 1L GPA that firms trust.

That’s Maya. She skipped the shortcuts and built the ability itself, which lasts longer. We’ll get to her in the next chapter.

But first, a line that has to go here. The information gap between a student who can build that score and one who doesn’t know where to start isn’t small, and it doesn’t fall on everyone equally.

Watch: using AI in law school without falling behind

AI Is Changing Law School — Here’s How to Use It Without Falling Behind

This chapter is from Anyone Can Beat the Law School Collapse by Steve Schwartz, LSAT Coach and Founder of LSAT Unplugged. I've been coaching the LSAT since 2005.

Published July 28, 2026.