How high is high enough?

How high is high enough — How well should my teacher have done on the test?

In thirteen years of test-prep teaching and tutoring I don’t think I’ve ever met a student who didn’t want an instructor who had attained a high score on the test in question. You’re probably not surprised; after all, why would anyone want to be taught by someone who wasn’t good at the stuff himself? But it’s actually not a silly question. “Those who can’t do, teach” is a flippant and slightly inapposite aphorism but it does contain an important grain of truth: teaching and doing are not the same thing. In theory someone might be the greatest, say, LSAT teacher in the world even though he cracks under pressure when actually taking the test himself; a prospective student should, again in theory, care only about getting the best possible instruction. But as I said, that’s the theory; every student I’ve ever met thought that to at least some extent, a high-scoring teacher was a good thing.

But now that I’ve raised the issue, I invite you to think critically about it. How high would you want your tutor to have scored? Is there a minimum below which you’d be uncomfortable? Contrariwise, would you be uncomfortable with a tutor who was a very high scorer, perhaps because you’d worry that he wouldn’t understand your issues with the test? Or does it even matter?

It probably won’t surprise you that I have an opinion about what’s best, and that my opinion is that high scoring is a good thing in a test prep tutor. But my opinion is more nuanced than just high = good. For the bottom line, go here; for the details, read on or just jump to the discussion about the test that matters most to you: LSAT, MCAT, GRE and GMAT, or SAT and ACT.


LSAT

When I started teaching it was with one of the big commercial test prep firms. During my approximately six years with that firm, in addition to teaching classes and individual tutoring students I was also a trainer, meaning I trained and certified new teachers.1 I trained teachers in several subjects, but the one I did most was LSAT so I’ll talk about that first. I learned quite a bit about what it takes to teach the LSAT from training people to do just that, to teach the LSAT. And I also learned how my trainees performed as teachers after I’d trained and certified them.

At the time my company required a 95th percentile score to teach the LSAT.2 95th percentile is right around a score of 167 on the test, so that was supposed to be the cutoff for training. However, now and then a local office would get a bit desperate, or they’d have an LSAT prospect in which they had a lot of confidence, and I’d find a prospective teacher in my training whose best score was below the cutoff. The theory (on their part, not mine) was that they could attain the required score after my training.

I don’t know exactly how many prospective LSAT teachers I’ve had in my trainings over the years, but it’s probably somewhere in the neighborhood of 100. And of those 100 trainees, perhaps twenty or thirty hadn’t scored a 167. Those numbers are large enough to give me a good feeling for what is and is not required to do well as an LSAT teacher. And the answer is…

… in my opinion, a minimum score of 95th percentile (167) was just about right. In all my years of training teachers I never encountered one who had scored higher than 167 but still didn’t understand the test well enough to teach it well (which is far different from being a good teacher, as a lot more goes into that than understanding the test), so 167 is enough. On the other hand, not a single one of those trainees who hadn’t managed 95th percentile scores wound up being an effective teacher. It certainly seemed to me that scoring in that range was a necessary, though not sufficient, condition for teaching well.

Why would this be?

I don’t know for sure but my impression was that while any reasonably smart person could regurgitate the lessons, only those who really understood the test could see beyond the lessons well enough to answer all the questions they’d face in class. More specifically, those scoring in the, say, 90th percentile range seemed to understand the right answers and how to find them but often not to understand the wrong answers, by which I mean why someone might choose them and how to explain how to tell that they were wrong. And to be a good teacher, one needs to be able to do all of those things.

I’m not saying 167 was a magic, absolute cutoff; I’m sure there would be exceptions. But I am saying that the phenomenon was real — a high enough score was necessary for a teacher to be truly good at it — and that the “high” score required seemed to be somewhere close to that 95th percentile cutoff.

Before I move on: You might expect that this 95th percentile thing wouldn’t apply as strongly when the student to be taught is in a much lower scoring range. Suppose someone is trying to move from, for example, 140 to 150 (that’s about 13th percentile to 44th); wouldn’t it be OK for that person’s teacher to be a lower scorer? That might sound like it makes sense by my answer is still a clear no. Remember, I’ve found that only the high scoring teachers understand the test well enough to understand and explain why someone might choose certain wrong answers; to help a student who’s not yet scoring well, that is exactly the skill that’s needed. Frankly, teaching people in the middle-to-highish range is easiest; teaching someone who has a very long way to go, or alternatively, someone who is scratching for those last few points, is the true challenge. That’s why everyone, whether his or her realistic goal is passibility or perfection, should worry about how well his or her teacher has scored.

Keep in mind that this whole discussion has been particular to the LSAT, and I believe that matters. Of all the tests I teach, the LSAT (1) is the only one that has no substantive content (nothing you need to know, like organic chemistry or algebra or whatever) but still manages to be hard; and (2) is by far the most tightly timed — almost everyone is under serious time pressure through the whole thing and it’s possible to get a very good score without doing every question. On tests for which those two things aren’t true, I think what it takes to teach well will vary somewhat; I’ll explain that below. The LSAT’s particular emphasis on speed also explains why scores above 95th percentile aren’t really any better indicators of teaching ability: at the highest levels it’s really all about speed, and there are plenty of teachers who understand the material plenty well enough to teach it but aren’t quite fast enough to ace the test. Meanwhile, the LSAT’s lack of content is why 95th percentile is necessary, while on some other tests 90th is OK; in heavy content-based subjects like, say, organic chemistry, it’s possible for a teacher just to forget something on the actual test, while on the LSAT there’s nothing to forget so an incorrect answer means that the teacher didn’t understand that question.


MCAT

What about the MCAT — is there an MCAT score cutoff below which I think people shouldn’t be teaching?

The answer is a rousing “it depends”. First, it depends on what aspect of the MCAT is being taught. The MCAT contains a whole lot of substance, in several subjects (biology, general and organic chemistry, biochemistry, physics, psychology, and sociology), and in that way it’s very different from the LSAT. I found when training MCAT physics teachers that many of those who became the best teachers had not scored terribly high on the MCAT; in fact, many hadn’t taken the test at all. The reason is fairly simple: MCAT physics isn’t advanced, graduate level stuff but it is tough for the average test-taker, and what was necessary to teach it wasn’t so much a deep knowledge of the MCAT (though that helped) but enough knowledge of physics to explain it well to someone who wasn’t a physics person. Some of my best trainees never did take the MCAT; they tended to be grad students in physics.

I haven’t trained chemistry teachers but I suspect that for g-chem, o-chem, and biochem it’s about the same: A teacher doesn’t need to be great at the MCAT but he does need to be good enough at the subject.

In biology, and probably also psychology and sociology though these two are new additions to the test so I can’t be as sure, we start to move from know the subject to know the test. Most MCAT bio questions are not of the did you memorize the table of hormones type (though these exist); instead, they ask students to read and understand a passage about a complex biological issue (for example, a disease or an experiment) and reason with and beyond the information given. Doing well requires both knowing the underlying biology and understanding how the MCAT works; teaching a student to do well requires exactly the same two things, except at a very high level. I wouldn’t want an MCAT bio teacher who was just a biology geek, nor one who was just good at the MCAT; rather, I’d insist on one who both knew biology extremely well and had scored high, at least 90th percentile, on that portion of the test. I wouldn’t insist on a 15 (old scale) or a 132 (new scale), in part because biology isn’t separated from chemistry and biochem but also because some MCAT passages are so arcane that someone missing a question or two doesn’t mean he’s not capable of teaching everything you’re likely to see. Perhaps a 12 or 13 (old scale; 129 or 130) on biological sciences/biological foundations would be about what I’d look for.3

Finally there’s critical reasoning, the MCAT’s peculiar spin on reading comprehension. MCAT critical reasoning absolutely is a separate species from all other test sections and types, even those such as LSAT reading comprehension that are facially quite similar. I used to train teachers in this subject too, and I found that while LSAT teachers almost always made the jump to MCAT critical reasoning successfully, it wasn’t always easy because the tests are looking for different things and the proper approaches are therefore different. I would insist, if I were choosing a teacher or tutor in this subject, on someone who had scored high — at least 90th percentile for sure, and I’d really like 93rd, which is a 129 on the new test — on that section of the MCAT. (MCAT critical reasoning is its own section with its own score, so this is knowable.)


GRE and GMAT

The GRE is divided into two sections, quantitative (math) and verbal, plus a writing section that most grad schools don’t care much about (though some do; be sure you know about the programs to which you’re applying). If you’re planning on working on only one of these, perhaps because your program cares only about the verbal (which is true for many humanities grad programs) or perhaps because you’re already nailing one section (usually the math) just fine but need work on the other, then you care only about how well your prospective tutor has done in that area. In practice, however, many students care about and therefore should think about both,

In each case I think it’s sort of like LSAT in that you should expect something like a 95th percentile score (about a 165) from your teacher. In the case of verbal this is because anything less means the teacher doesn’t really get the GRE, which has its own idiosyncrasies; in the case of math it’s because while the math on the GRE intimidates non-math people it’s pretty straightforward for a genuinely math-oriented person, so failing to score 95th percentile (which is a 168, pretty close to a perfect 170) suggests that the teacher isn’t very solid in math and may have trouble explaining the more difficult concepts if you have don’t understand the explanation the first time through. I can see fudging on this one a bit, but frankly there are a lot of teachers out there who can score 170 on the math side, so I don’t see why a student should have to settle. (Note that the GRE verbal section is notoriously difficult to get a perfect score on, because of all the vocabulary words one has to know, so scores higher than 165 aren’t really much better in terms of what’s necessary to teach well.)

The GMAT is a similar test to the GRE, and my advice for it is similar. GMAT scores are usually reported as a single number, and I’d advise looking for a 95th percentile total, which is in the 720 or 730 range. It’s better if you can get a breakdown into math and verbal but that’s not usually available; make sure your teacher isn’t much weaker at one part or the other unless you’re not looking for help in the area of weakness.

A surprising fact: the pools of people taking the GRE and the GMAT are different and so the percentiles (which are calculated relative to who takes each test, not to the whole population) mean slightly different things. In the middle of the scoring range I believe they’re about the same, but a very good GRE score is better than a very good GMAT score, even if the percentiles are listed as the same.


SAT and ACT

That’s a lot of talk about the graduate admissions exams; what about the SAT and ACT? This is an easier problem, simply because there are so many SAT and ACT tutors out there. While it probably isn’t necessary that your tutor have scored super high, there simply isn’t any reason to settle for someone who can’t score almost perfectly on it, which any highly experienced, smart tutor should be able to do.

One potential hitch is that many tutors haven’t taken the SAT or ACT in a long time, because adults rarely take it. This isn’t much of a problem, though, if the tutor has GRE or GMAT scores available, because these tests and the appropriate teaching techniques are qualitatively similar; I would expect that anyone who can score 90th percentile on the GRE (both parts) or the GMAT, but who doesn’t focus exclusively on graduate tests, is technically qualified for SAT tutoring. The ACT has additional content — tougher math, plus some science — but most tutors who have been doing ACT work for a while will not have difficulty with this extra substance, so even though it’s not a perfect test I’d apply about the same criteria as for SAT.


Synopsis

My apologies to those who don’t enjoy reading tomes; let me sum it up. If I were hiring a tutor for someone I cared about, here’s what I would insist on in terms of scores:

LSAT — 167 minimum. Do not, do not, do not skimp on this. Higher doesn’t really matter, but lower is a Bad Thing.

MCAT — If you’re looking for help in a particular subject then what you want depends on the subject. In physics and the chemistries, you want an expert in the subject (grad student level is fine but not necessary). In biology, psychology, and sociology, you want someone who knows the subject and the MCAT, so look for at least 90th percentile (about a 128) score in the relevant sections of the MCAT. For critical reasoning4, I would insist on a 129, which is about 93rd percentile, on that section; definitely don’t go below 128. And if you want a teacher for the whole test, you want all of these things, on the official MCAT.

GRE — Ideally 165 verbal, 168 math. You can let these slide a little bit without loss, and it’s fine to get someone good in only one subject of that’s the only one you need to work on.

GMAT — 720 minimum. You absolutely should not have to go below this.

SAT — If your tutor has a recent SAT score, it should be very close to perfect; there’s no reason to settle for less. If (s)he doesn’t, a GRE or GMAT score that meets the above criteria should suffice.

ACT — Same as the SAT in a way, so look for the same as the above, but make sure the tutor also knows the extra content — the ACT requires more math knowledge than the SAT, plus some science.

Two more things to be aware of if you’re taking a commercial class or hiring a tutor from one of the big companies: First, at least two of the largest test prep companies employ teachers who have scored just 90th percentile on the relevant tests. As I’ve explained, I don’t think that’s good enough for most of the tests I’ve been talking about. The test prep companies company try to make it difficult for a customer to “shop” teachers, but if you’ve decided to go the big-company route I suggest you push them on this issue. I wish it didn’t matter, but it does.

Second, some of the test prep firms put teachers in the classroom who have never officially taken the test they’re teaching. They administer unofficial tests for this purpose, proctored and graded as the official tests would be. This is sometimes a reasonable thing to do, I believe, but you should understand that an unofficial test really isn’t the same thing as an official one; if you can, insist on a teacher who has taken (and earned a good score on) an official exam. (Note also that teachers will often decline to tell you their scores if you ask; in my experience; anyone who refuses to tell a student when he took the test or how he did, even when pressed, usually has a good reason for not telling… and the reason is often that he either didn’t take it at all or didn’t get a very impressive score.)

Finally, please understand that I’m not saying that a high score makes a good teacher. There is far more to teaching than just knowing the material in question, and you should look at many factors, not just score. All I’m doing here is explaining the minimum score that I believe indicates sufficient technical knowledge.

If you’re trying to decide what to do about a prep course or tutor, I’d be happy to talk further with you even if you aren’t considering having me help you; get in touch with me through the contact page. And if you’d like to know how I did on these tests you can find that information on my results page.

Notes

  1. I also trained and certified new trainers of new teachers, which is a bit confusing but it’s not an infinite loop; that level is as high as it went.
  2. Prospective teachers didn’t have to have taken the official LSAT, though the company preferred it; otherwise attractive prospects, particularly those who already taught another test type, were allowed to take a full-length practice test in the office. This is a common industry practice.
  3. AAMC hasn’t yet released a full table of MCAT percentiles for the new (2015) MCAT, but so far it looks like 129 on any given subject is 93rd percentile and 130 is 97th; this probably won’t change much.
  4. technically, “critical analysis and reasoning skills”

1 thought on “How high is high enough?”