Is it better for one student to get a job than n students to fail the bar?

A student’s final law school GPA predicts bar passage better than other independent variables. But the relationship isn’t causal: raising the mean GPA of all students does not promote bar passage.  Indeed, some investigators have suggested that the inverse is more likely to be true. When GPA rises for all students, individuals at the bottom of the class aren’t sufficiently signaled that their grades are really and truly bad, and consequently such badly-warned students don’t approach bar study with the requisite degree of seriousness. That is, if a school has a mean of a 3.3 at graduation, the bottom 20% of the class probably has GPA of around a B-.  B- students may well say to themselves “sure, I’m at the bottom of the class, but I’m not a C student!  I’m not in danger of failing the bar!”  But they are.  In this perverse way, raising the mean increases the rate at which weaker students fail the bar, even as overall, grades are positively correlated with passing!

The puzzle deepens. Students often argue that employers focus on mean GPA to the exclusion of class rank.  Given that students are competing with other schools (nationally and regionally), there are race-to-the-bottom pressures on each law school’s curve generated by employment markets.  A school that produces students at the 50th percentile with a 3.5 mean will obtain better employment outcomes than one that produces students at the 50th percentile with a 3.0 mean. All else equal, schools should reduce barriers to employment.  (Of course this result depends on employers indeed acting in the irrational manner described – ignoring or downplaying class rank and focusing on absolute GPA.  This would be very, very difficult to test empirically, though I imagine someone could give it a shot using nifty studies.)

You see the tension, right?  A higher mean simultaneously could boost employment in the middle and higher end of the class while also depressing bar passage at the lower end of the class. These contrasting outcome effects turn on psychological biases resulting from overemphasizing raw grades over percentile rank, but simply providing class rank instead of grades would cause employers to balk. The tension may lead administrators and faculty to an uncomfortable question: when the two conflict, should we privilege bar passage over employment? What is the appropriate calculus? Could we live with one additional student failing the bar if two got a job?

My own view is that the price for bar failure is so high that the number of jobs won in this calculus would have to be unrealistically high.  Consequently lower means are to be preferred to higher ones at some schools.  What do you think?

You may also like...

13 Responses

  1. Andrew Selbst says:

    According to what you’re saying, the student him/herself in an intervening cause of the higher bar failure rate at higher means, but not the reduction of job prospects at lower means. So while the signaling is a real problem, at least it’s one that’s solvable with counter-signaling from the law school, while the job problem is entirely external. This seems to push toward higher means, no?

  2. Dave Hoffman says:

    Isn’t the question which is more effective, “counter-signalling” using non-grade measures with students, or “counter-signalling” using non-grade measures with employers? Wouldn’t employers be more likely to be easily educated?

  3. Andrew Selbst says:

    I suppose that is a form of the question, yes. But what would the signaling to employers look like? “Please use class rank – we have our reasons to keep GPA’s low.” I have to imagine the law schools would be better able to convince the students that they should take this bar thing seriously, than convince the employers that *their* students are better than the others with higher grades, a statement for which there is obvious self-interest.

    Thinking about it another way, the law schools probably are all doing the best counter-signaling they can with the employers right now because employment focus is all the rage and everyone’s looking at employment stats, but may not be worrying as much about the bar failures because former students who failed the bar are never heard from again, and already paid the law school.

  4. If you are correct, the moral of the story is to solve the false conflict by inflating the curve as a general matter … but also require a certain percentage of C-s Ds and Fs to signal real hard to certain students.

  5. Dave Hoffman says:

    Just out of curiosity, how would/do you enforce such a distribution rule?

  6. Josh F. says:

    This assumes there is objectivity in grading and that GPA correlates with intelligence. Rather than just calling it all a mystery, I would take a second look at your very first sentence.

  7. Howard Wasserman says:

    Remove the mean, require a mandatory low-end distribution of D or F (rather than a C-, which still remains too high to get at what Mike is suggesting).

  8. Larry Rosenthal says:

    There is only the “tension” hypothesized in this post between bar passage and employment if employers focus on GPA and not class rank. The claim that employers act in this way is widely repeated by students and faculty, but I have seen very little in the way of empirical evidence to support it. A few years ago, I asked our career service office to look into this matter, and it found very little evidence that more than a tiny handful of employers acted in this manner. Unlike most law professors, I practiced for many years at a supervisory level, and our hiring practices always focused on class rank and not GPA for many reasons, including our awareness that law schools can game GPA. As the post acknowledges, the claim that gives rise to this asserted tension “depends on employers indeed acting in the irrational manner described – ignoring or downplaying class rank and focusing on absolute GPA.” The market presumably would punish this kind of irrationality. In general, I have found that employers have a far more sophisticated understanding of the hiring process that law students or law faculty. This should be unsurprising; they are risking their own money (and clients) on each hire, so we should expect that they approach hiring in a considered fashion. In short, I think it quite unlikely that the “tension” hypothesized in the post actually exists. It may be more likely that increasing prevalence of grade inflation has signaled to employers that high grades — especially at non-elite schools — are probably not a reliable signal of quality.

    Larry Rosenthal
    Chapman University School of Law

  9. Howard Wasserman says:

    Larry: Our experience at this end was very different. We previously had a tough curve with a low mean (2.47-2.67 for 1L classes, 2.67-2.87 for upper-level classes). Our top-5 students routinely were in the 3.3 range and employers identified that as a reason for not hiring. Now, perhaps there was another reason going on, but this is what we were told.

  10. Larry Rosenthal says:


    Your median is is only a little below ours (though our top students have GPAs in the vicinity of 3.6), and without a great deal more information it’s hard to know what to make of your situation. My guess is that the explanation that is being passed along to the faculty is the product of some pretty heavy spinning. I have found that a great deal of that goes on when career services offices are told why their graduates have not been hired — often by persons who actually have no idea why hiring decisions were made — or when career services personnel pass along their own hunches as fact. An explanation for hiring patterns that assumes employer irrationality is surely the least likely to be accurate.


  11. Christine Hurt says:

    To chime in on Andrew and Dave’s conversation on whether it’s more effective to countersignal to low-performing students or employers: Most schools have some sort of academic probation, triggered by a particular GPA. One unintended consequence of raising the mean GPA might be that academic probation is harder to trigger. (This is the case at one school that I know about, and perhaps more.) As long as you reset academic probation, that should be a strong enough countersignal. But, if you don’t reset that trigger, then the B- students wouldn’t consider themselves at-risk.

  12. prometheefeu says:

    You could have a tough letter-grade curve and easier GPA curve. Also, I thought you could just retake the bar. Wouldnt failing the bar exam provide you with the signal that you need to study more and try again?

  13. Anon says:

    At many schools with few requirements past the first year, it is 1L GPA, rather than final law school GPA, that best predicts bar passage. Students can game GPA in non-required courses by avoiding large curved classes (which often overlap with the classes on subjects covered on the bar exam or which have tests that are similar to the types of tests used in the bar exam (e.g., timed essay or multiple choice tests)). If you change the curve, you may get more students taking those courses that help them on the bar, which theoretically would help them both get jobs and pass the bar . Students may still avoid those classes, though, as they seek even higher GPAs to make up for 1L shortfalls. The problem with this strategy is that it is not clear that post 1L GPA affects jobs as much because fewer employers hire on campus after 2L fall interviewing and hiring outside the traditional law school model is likely to be more influenced by alternative factors (e.g., networking etc) since employers aren’t looking at and comparing 30+ resumes at one sitting.