My post below led some commenters to ask whether there's been grade inflation in law school grades. A few related answers.
1. When I was a UCLA Law School student in 1989-92, our curve was 20% As, 40% Bs, and 40% Cs or below (the "below" grades were optional and very rare) in each course. In the mid-90s, we shifted to 20% As, 60% Bs, and 20% Cs or below. Recently, we shifted to a 25-29% As, 41-52% B/B+s, 18-22% B-s, and 5-8% Cs or below for first year classes, and 23-27% As, 50-60% B/B+s, 17-23% B-s, and 0-10% Cs or below for second and third year classes (basically a 3.2 median, slightly below a B+). So our median grades have been increasing, from B- to B to B+ish, and our Cs have been declining.
2. On the other hand, the quality of our incoming students, at least as measured by the LSAT (which to my knowledge has not had any grade inflation of its own), has been increasing, too: In 1998, the first year in which U.S. News & World Report reported the 25th and 75th percentile LSAT scores, UCLA's range was 159-165; this year it was 162-169 (on a 120-180 scale). This is part of a broader trend that is seen even at higher-ranked schools; Columbia was ranked #4 both years, but its LSAT 25-75 range rose from 165-171 to 168-173. Should this justify a corresponding rise in law school grades? I don't know how to answer that.
3. Also, for whatever it's worth, incoming UCLA law students, on average, had an A- average at their undergraduate schools (the 25-75 percentile range reported this year in U.S. News was 3.51-3.82). Back when we gave lots of Cs, lots of students would get their first Cs of their lives at UCLA Law School. Is that the way things should be? Again, I don't know how to answer that.
4. As best I can tell, the increases in our grades have been driven by one main factor: The increases in grades at other schools. We shifted to a B median in the mid-90s because we noticed that most Top 20 schools had a B median. Our B- students were roughly comparable in class rank to B students at peer schools, but they looked worse to employers who weren't that familiar with the UCLA system. (An employer could of course look closely at the descriptions of the grading systems and figure out the difference, but we were afraid that many employers wouldn't look that closely.)
We shifted to a B+ median recently because we noticed that most Top 20 schools had done the same. I'm pretty confident that we were at the trailing edge of the change, not the leading edge. We didn't want to increase our grades beyond what others were doing, but we also didn't want our students to be at a disadvantage. This sort of behavior may be bad in some overall sense. But it is sensible for a school that's trying not to leave its students unfairly disadvantaged. If someone suggested some multi-law-school grading reform, I might endorse it (though I can't speak to any antitrust law questions this might or might not raise). But so long as each school has to make these decisions by itself, I think we did what we had to do.
5. Though I'm not wild about grade inflation, I should note that a B+ median still leaves plenty of gradations between students, especially when one averages together the grades in many classes. If everyone got A+s or As (which is more or less the system at Yale, with what I'm told is roughly 20-30% of each course getting Hs, and the rest getting Ps with the exception of a very few LPs and fails), that might pose more of a problem. But a system with plenty of A+s, As, A-s, B+s, Bs, and B-s, and occasional Cs (with some required in the first year) adequately conveys to employers which students tend to be better and which tend to be worse. And to the extent that such a system causes confusion when employers compare UCLA students from one grading system with UCLA students from another (which tends to be rare, since most students are competing against others who graduate the same year or shortly before or after), maintaining the same median as other schools diminishes confusion when employers compare UCLA students with students from other schools.
6. One possible solution to this problem is to report class rank, something that I'm told virtually no schools systematically do these days (though when it comes to top graduates who are applying for clerkships or teaching jobs, many schools do in fact report informal ranks). But for complex reasons — which may be caused partly by excessive egalitarianism, and partly by a plausible (though not obviously right) concern that the difference between 60th percentile and 40th percentile probably looks bigger than it should, and that a GPA may do a better job of indicating how slight that difference is — there's been little move to return to the class rank system.
Related Posts (on one page):
If not this solution, obviously something has to be done. Schools can only continue inflate so far. There is a ceiling up there that they are rapidly approaching.
As I said, if we were to get to just an A+ and an A, the situation might be worse. But I don't see that happening, except at a school like Yale that knows it can afford to do it. So while I feel some unease about grade inflation, I'm not sure that the problems with it are that obvious.
This makes it a bit of a shock for students who get put on probation or kicked out for getting too many Cs.
Did you see the article about grade deflation at BU. Ann Althouse blogged it (so did I, but I saw it at her site).
Oh, and also, I've heard through hearsay -- at least double, and possible even more remote -- that Yale's percentage of H's is higher, maybe as high as 50%.
I think the problem, if it is exists, is that at least some people assume they know (at least approximately) what a certain grade or GPA means without looking at the actual underlying scheme in detail. So, for example, people will assume certain things about an "A-average student", or a "B-average student", and might even assume the former has done better than the latter without looking at the underlying schemes.
Which just supports Mark's basic point--the best way to prevent such confusion would be to mandate a standard scheme.
I think a curve is eminently sensible to reduce the problem of grade inflation though you'd still have the problem of comparison with schools that do have grade inflation.
A few options:
1) Law firm recruiters are not idiots. With a clear statement of grading policy on each transcript (including each grade's "market share" of that course) one would hope they realize that a 3.8 GPA is meaningless when 75% of the class has it.
2) Law schools can agree with each other on a standard curving policy. I don't see the anti-trust problem provided it's transparent and any law school can join. Get your own legal advice on this one, I'm not a lawyer (yet).
3) Law firms really care about ranking as that's what really tells them where you stand, so publish rankings instead of just GPA's.
Personally I like option 2 and you don't need every law school in the USA to sign on, just the top 30 - 40 schools.
The biggest problem with grade inflation is that students who attend through these statistically large transistions get the shaft - comparing with the next year's class, like with a clerk, or especially where a graduating class has students with varying terms under the new regime - like 4-year night students, or joint JD program students. They have "artificially" deflated rank/averages.
The other problem is that not all systems are transparent - I reviewed the academics of a Georgetown grad recently. The school didn't publish rank, percentile or curve so I couldn't meaningfully evaluate whether the B+ (sans honor designation) was top-half of the class or quite mediocre.
I'm not aware of any other school that uses this system, so it could be difficult to compare 0-100 GPA's with 0-4. But for those who value intermediate gradation, this is probably the way to go.
Different era, different statistics.
Interestingly, though, I have had several legal employers ask me to help explain my grades (eg, "Is this a good grade at your school?") over time. In that sense, it seems to me that even if they could figure all this out independently with enough effort, they aren't willing to put in that effort in all cases.
Which makes sense--at some point the marginal return associated with independently learning more details about each prospective employee's school's grading scheme may be outweighed by the marginal costs associated with learning those details. But insofar as schools could easily reduce those costs (by, say, adopting some standards), it would seem to make sense to do so.
Unless, of course, the schools think their students--or at least enough of their students--benefit from all this. Which is entirely possible.
1. In a few small classes, there are up to 50% Hs given out. But the vast majority of classes feature about 20-30% Hs.
2. Contra the post's parenthetical suggestion, Yale's Hs and Ps are not akin to other schools' A+s and As -- more like As and Bs. As proof, note that getting all Ps at Yale isn't impressive in the least because it's the default. Getting all As at pretty much any school is.
For whatever all that is worth...
This would create a more accurate distribution pattern, and would eliminate math-work for employers.
Sometime in the 90s, when the job market sucked, they changed the system to increase the average GPA.
A: 5%
A-: 10%
B+: 25%
B: 50-55%
B-: 0-5% (usually around the 5% mark, but some professors won't give a B- for a serious effort.)
C and below: 0-5% (as my Corps prof explained it, she'll only give a grade that low if she believes that you would likely be sued for malpractice if you tried to practice in the area.)
This lets you distinguish between the top students and call out the screw-offs, but the big fuzzy middle gets to stay the big fuzzy middle. Which, I think, contributes a lot to NYU's more collegial atmosphere than other law schools. In other words, it's more important to distinguish between 90th and 80th percentile than between 40th and 20th percentile. Grades have other effects than informational content, and the curve should be tailored to the effects desired. NYU's curve does exactly what the school's trying to achieve.
Thankfully, grades shoot up for 2nd and 3rd year students!
That's an interesting "curve". Sounds pretty nice actually.
Ideally, they would publish cumulative class bands - top 10%, top 20%, top quarter, top third, and top half, along with the GPA cutoff for each of these regions. This would minimize the meaningless distinctions (is #75 in the class really better than #78) while also protecting students who fall just short of the cusp (by indicating where the line is drawn) from being too prejudiced by just slightly missing the cutoff.
I think that you're correct about the refusal to allow class ranking being based on an overly sensitive sense of egalitarianism (which, as usual, is cynically discarded when it better serves the interests of the institution, see, e.g., clerkships).
I think the best description of it was stated by Prof. Richard Sander of UCLA when discussing the grading curve change, when it was still just a proposal. One of the students asked him why class ranks weren't published. The following is a paraphrase, but is quite close to what was actually said:
"It is, I suppose, a form of grade socialism. The idea being that, by fuzzying up the distinctions, you can strongly aid the poor students while only slightly harming the good students. However, I suppose that, like often happens with such schemes, you wind up hurting the good students but not really helping the intended beneficiaries."
Speaking as someone who has had to review hundreds of resumes in connection with my (top) firm's recruiting process, I can state from experience that it is not. Top law schools are like Lake Woebegone, where all their students are (or appear to be) above average.
I've given up trying to sort it all out. After all, I have a practice to run. No matter how indispensible new graduates may think themselves to be, I have more pressing concerns (like running a practice) and don't have endless hours to spend trying to discern the hidden meaning of a transcript.
What I'd like to see is a candidate's precise class rank. Since law schools won't give that to me, in the absence of full disclosure, I am forced to improvise. What I'd like to see is a candidate's precise class rank. Since law schools won't give that to me, in the absence of full disclosure, I am forced to improvise. I look at things like law review, clerkships and the ratio of serious classes to absurd guts (e.g., corporate finance v. law and gender) on a candidate's transcript. These improvisations are imperfect. Some excellent people will be overlooked. Some unequiped people will find themselves out of their depth.
The "all our students are smart" justification doesn't fly. So are all our lawyers. That's why we generally limit our recruitment to top schools in the first place. We want to know who the best of the best are, because, experience has shown that the middling top 20 law school graduate will struggle at, e.g., Sullivan &Cromwell, every bit as much as the middling LSAT scorer will struggle at Harvard Law School.
Furthermore, profligatly distributing A+ grades and concealing class rank does a disservice to graduates, leaving them ill-prepared when they rudely learn that such egalitarianism is foreign to big law firms.
Unfortunately, I see no prospect for meaningful change. Since only a small minority can be at the top of their class, the great majority have every incentive to demand a system that obsucures who's who.
Eh? I went to U-Miss law school (hello, TheWagon), a school dangling from the bottom of the 2d tier, and for some reason I thought our transcripts included class rank. I know that I sure as hell had mine on my CV, even though it was nothing too special (top 20%? 25%?).
Particularly for graduates of less-than-stellar schools, self-reporting class rank is a good idea, and I think that transcripts should reflect it to keep graduates honest.
I have seen tremendous grade inflation in the 16 years I have been teaching (mainly not law). It is hard to get data because some schools are very protective of the distributions. My personal observation is that private universities tend to give higher grades.
I agree that 6 grades is enough for differentiation but with a constantly changing scale and differences across schools, GPAs are very hard to interpret without recent specific knowledge.
Consistency across schools would be a great help to people trying to compare candidates. Sure schools get a reputation as easy or hard but that adds unnecessary complexity.
I was involved in one competitive search where if no rank was stated on the resume we assumed they were too low. So for highly competitive jobs ranks already replace GPA.
Some adjustment needs to be made in classes and across years as poor students drop out (dropouts can be given the lowest grade for curving purposes).
Using average standardized tests (SAT or LSAT) to set the grade distribution for each class is one way to standardize.
If there are a large number of stdents in a class, they are likely to have a distribution of ability, and that distribution is likely to be similar from year to year.
If most people in a professional school class don't get it, there is something wrong with the teacher, not the students. In something like law school classes, the depth and complexity of the material is such that the level of understanding and the ability to apply knowledge to complex problems will vary, and good examinations should sort this out. Similarly, writing ability is bound to differ. If the task is mastering a multiplication table, your "what if's" are credible. But they just don't apply to professional school.
If there is payment for service, shouldn’t a student require a certain grade for the work they put in? Or demand a refund? As a prospective employer do you trust a prof (whom the student indirectly pays) to give an accurate accounting of ability? (I’m not serious, but these are legitimate questions.)
More on topic, grade inflation is a big deal, but must be done because others are doing it. I had a well-renowned history prof say in effect, “I will grade such that 90 percent of you will get at least a B, and a few of you will get A’s. You won’t have to work hard for a B, but to get an A you will have to perform as you would expect to in an Ivy league school.”
In my engineering classes, however, the curve was very different from class to class. I’ve had classes where 15% got D’s or F’s and 25% got C’s. (out of 400). I’ve had other classes where the ave was a B, but that was targeted for a score of around 50%. (He would make his test out of 130 or 140 so people who were used to a 95% wouldn’t freak out too much – They’d get an A with a raw score of 75. I’ve taken other classes where only one student got greater than 40% -- happened to be equally a problem of grading (by TA’s) and of teaching. However, in most of my core engineering classes, the curve was decided on a test-by-test basis.
While our class ranks were given (if you asked), many, if not most job placements were done through/to employers who went through the same engineering program.
And in full disclosure,
--B.S. ChE, ’03, minor in History
History GPA – 3.85
Engineering GPA – 2.75
It all depends on the communicative purpose of the grading scheme. If the purpose is to communicate how well the student performed compared to some objective standard, then a strict curve is inappropriate. But if the purpose is to communicate how well the student performed compared to his peers, then a strict curve is in fact appropriate.
As it turns out, I think different people in different contexts have different purposes in mind. Accordingly, and along the lines of what Mike suggested, I always thought the ideal grading scheme would give both the student's "raw score", which would be on some attempt at an objective performance scale, and also their "distribution score", which could just be their percentile rank in the class.
My school doesn't publish grade distributions, but A's are fairly easy to get, B's are the default, and C's are nearly unheard of. My impression of why this is so goes back to a previous commentor's assertion that the 3.0 GPA cutoff for grad students makes that the effective minimum student GPA. This is especially true in my department, where we have a high proportion (greater than 50%) of foreign students. If they don't maintain the 3.0 GPA, they go home.
It seems that there are two pieces of information we want to measure: (1) the student's comprehension of the material presented in the class compared to some objective standard that doesn't vary from year to year or school to school, (2) the student's comprehension of the material relative to the other students in that particular class.
Clearly, some law schools have more smart students than others. After all, law school admissions generally follow the bell curve of the LSAT (which the USNWR rankings roughly follow). Thus, top 20 schools will have more students who will be at the "A level" of the first type mentioned above than compared to a lower tier school. So, in that sense it makes sense for top schools to have a higher curve than lower schools simply because more students in each class will deserve A's based on their comprehension of the subject taught. The challenge, of course, is determining what the objective standard is and measuring it since we don't have standardized exams.
The second measurement--performance relative to other students in the class--has its own challenges. Since top law schools admit top students, one would expect them to do well. So, to measure the differences between students who "always had A's" in undergrad is like trying to zoom in on the A range and break it apart and spread it across the entire A-F scale. So, 98/100 is an A, 96/100 is a B, 94/100 is a C, etc. But, when you do that, you face the challenge of making a meaningful distinction between an essay answer that is worth 95 points versus 94 points. That's a hard thing to do.
Probably the most useful system would be a high curve for top law schools combined with class rank that somehow shows the statistical significance of that rank. (The downside with publishing ranks is that it would seem to increase competition among students and create a less friendly atmosphere.)
But, to bring this back around to law school, none of the native english speakers earning above average scores in those classes would ever dream of scoring as low as 170 on the LSAT. Most would score above 175.
I doubt that a serious double blind statistical test could justify six or even five grade levels in law school classes. Most likely two or three passing grades would be the maximum actual information content these grades are testing. The system described above from Yale (H,P,F) is likely the most honest, informative, and objective system for law school grades and has the additional benefit of discouraging the computation of "averages" among incommensurate entities (such as grades from different professors or different classes entriely).
If the standards of statistical validity required by civil rights regulations of employers who write employment tests (including repeatability, statistical correlation to job performance, and lack of bias correlated with prohibited groupings) do any of the professors here think they could legally continue to assign psuedo-precise grades like "A-" and "C+"?
At first I was convinced that the system was totally random and I shouldn't worry too much about it, but the more I thought about this, the more it made sense: the property professor was a better teacher and the exam was not that difficult; meanwhile, the torts professor was among the worst I had in law school and left everyone in the class confused. Yet, someone looking at my resume would have been mistaken to conclude that I was an ace in torts or that I should not be trusted in dealing with a property-related question.
I'm not sure what the solution is here. Since I was apparently at the top of a low torts curve, it seems that if the torts professor had graded everyone on their knowledge at the end of the class, everyone would have failed, even though the fault was largely hers. In the same vein, if the property professor had ignored the curve, a huge percentage would have gotten an A, even though the reason was that he was a good professor and/or that his exam was too easy.
Ultimately, whether you favor or oppose curve grading may depend on whether you think it is more likely exams will be too hard or too easy, and whether you prefer to put the burden of a bad prof on the student, or give them the benefit of the doubt that if they were smart enought to get into a top school but would still get a C or D, the fault is unlikely to be theirs.
I can't remember where GGLS was on the US News Rankings at the time, but my guess is 4th or 5th tier. At the same point in time, Michigan, which was in the 1st tier, had just raised its curve from a C+ to a B- so that its students could compete with Harvard students for law firm jobs and clerkships. I guess that at some point in the pecking order (maybe the 3rd tier?) the urge to inflate and the urge to deflate cancel each other out and the schools have what used to be called a normal (i.e., a "C") curve.
Can anyone confirm this?
On the other hand, the country is bigger. UCLA may be getting better students simply because the number of seats in law school classes has not kept up with the population increase. If UCLA was getting students in the top 20% 20 years ago, it may be getting students in the top 10% today. Simply because population increased, but class size has not.
Of note, does anyone have any evidence to suggest grade inflation is necessary? Is there a study showing hiring partners often fail to recognize the quality of the law school and a lower GPA in favor of an unknown law student from an average law school with a higher GPA?