Average Usmle Scores

This forum made possible through the generous support of SDN members, donors, and sponsors. Thank you.

atropinedoc

Member
7+ Year Member
15+ Year Member
20+ Year Member
Joined
Jul 11, 2003
Messages
59
Reaction score
0
what is an average usmle score? is it 85 or 88 ? I am really confused.

Members don't see this ad.
 
I think 75 is passing for the 2-digit score and 182 for the 3-digit score. The average for the three-digit is usually 200-220 (according to the USMLE site) but I have NO CLUE what the two-digit average is.
 
If we assume that the mean is 200 with an SD of 20, then the highest score one could theoretically get (a 100) would be maybe 280. (4 SD's) So, if 180 is passing, that is a full SD below the average, which tells me that a 75 (if it correlates) is 5 SD's below the maximum...so, the average would be 80. I think that has already been said, but I wanted to see if I could work it out in my own mind. Does that seem weird that the passing grade would be so far below the mean? I thought it was a 70 two digit score, myself, with a three-digit of around 170...but the scores are rising.
 
The average during 2002 was 216, with a standard deviation of 24. You needed a 182 or a seventy-five (two digit score) to pass.
 
OK, I just checked my score report (from August 2002).

It says nothing about average 2 digit scores. I don't think the 2-digit score is considered as important as the three-digit score.

For the record, the mean was 216, SD 21 (ckent must have taken his test really late in the year(?)), SE 6. And yeah, that was for "first time US and Canadian test takers".

No one has been able to explain to me the meaning of the 2-digit score. It seems to be tied directly to specific 3-digit scores, regardless of means and SDs. It makes absolutely no sense to me.
 
Originally posted by BellKicker
OK, I just checked my score report (from August 2002).

It says nothing about average 2 digit scores. I don't think the 2-digit score is considered as important as the three-digit score.

For the record, the mean was 216, SD 21 (ckent must have taken his test really late in the year(?)), SE 6. And yeah, that was for "first time US and Canadian test takers".

No one has been able to explain to me the meaning of the 2-digit score. It seems to be tied directly to specific 3-digit scores, regardless of means and SDs. It makes absolutely no sense to me.

Actually, I took my test on June 24, 2002. I don't understand what the 2-digit score means either, but I've heard that it used to mean something, and then they changed it around so no one understands it anymore.
 
I thought that it was always a way to relate scores from one year to another...i.e. an 85 in 2000 would correlate with an 85 in 2003, at least in terms of how the population did.
 
Originally posted by Idiopathic
I thought that it was always a way to relate scores from one year to another...i.e. an 85 in 2000 would correlate with an 85 in 2003, at least in terms of how the population did.

Right, because a 210 might be an 85 this year, whereas last year it could have been a 212. The two-digit corrects for the statistics (mean and SD).
 
Yeah, that would make sense but that's not how it works. The 2-digit scores are tied to the 3-digit scores. As far as I know, anyway. Like ckent says, it probably used to mean something back in the day and they kept it on for the sake of old school PDs.

It would make so much more sense to have kept the percentile rank on the score reports but they took it off some years ago.

It bugs me.
 
Originally posted by ckent
Actually, I took my test on June 24, 2002.

Frickin' crazy that it would fluctuate that much. Are you sure the SD was 24 (duh, of course you are). I just remember everyone from last summer saying it was 20 in the spring and then 21 in the summer. Maybe the SD is by US/Canadian region (not that that would make any kind of sense whatsoever)

It's 24 this year, as you probably know.
 
Originally posted by BellKicker
Yeah, that would make sense but that's not how it works. The 2-digit scores are tied to the 3-digit scores.

Yes, they are tied, but not irrevocably; the point is as Idiopathic said, for reliability year to year.

From USMLE.org
On the three-digit scale, most scores fall between 160 and 240. The mean score for first-time examinees from accredited medical schools in the United States is in the range of 200 to 220, and the standard deviation is approximately 20. Your score report will include the mean and standard deviation for recent administrations of the Step.

The two-digit score is derived from the three-digit score. It is used in score reporting because some medical licensing authorities have requirements that include language describing a "passing score of 75." The two-digit score is derived in such a way that a score of 75 always corresponds to the minimum passing score.

Derived in such a way - meaning it is flexible.
 
Yeah, but I don't think it's been flexing over the last couple of years. I asked the same questions in the USMLE forum. I think we reached a conclusion that the "lowest" 99 has been fixed at 244 for at least 2 years.

I know that with the mean and SD the "lowest" 99 has gone from top 2% to top 10%.

In other words, a 99 is becoming more and more common. I suppose it's possible the NRMP likes to sit and watch the scores for a few years before they act on them. Maybe next year it'll take a 250 to score a 99, who knows.
 
The three digit score is supposed to be the best representation of your ability and is comparable year to year. It is based on scaled performance. The two digit score is less meaningful as many have said.

The reason the two digit score changes is because it is based on a fractional representation of the total distribution. It is actually a 25 point scale of the overall group that passes the test, with 75 being set at "zero" (lowest pass). If the distribution is more "normal" then the 99 score and the 75 score will be fairly small groups, whereas the midpoint would be at about 87-88.

That being said, it is highly unlikely that the distribution among those with passing scores would ever be normal, and is probably multi-modal. This creates discontinuity in the year-to-year scores and for this reason, they should be ignored.

It is true that some PDs think it is a percentile rank, but that is a mistake. On the other hand, there is not likely to be significant difference in performance between those within 5 points of each other (two-digit score) as far as the rigors of residency are concerned.

Good Luck All.

OtoNerd B)
 
people here (and everywhere else) seem to be really misunderstanding what the 2 and three digit scores mean

2 digit scores are standardized so that the minimum passing score is a 75 and the maximum score is a 100. This approximately correlates to each digit between 75 and 100 being 4 percentiles of those who passed. The 2 digit score is something requested/required by state medical boards for all licensing exams so they can quickly evaluate the score without figuring out each exam's scales.

Scores used to be calculated after a group of students took them and you were given a 2 digit score, a 3 digit score, and a percentile. They used to do it where a large number of students took the exam, then they were all graded and statistics were calculated based upon all the students taking it then. The percentile was where you stood among the other students who took the exam in the block from which your exam was taken and scored. Therefore, each year, the median score would get a 50th percentile, although one year this could be a 205 and the next a 210. So a 240 in 1990 might be a lower percentile than a 235 in 1991. The problem is that a 240 is doing better, but 1991 may have been a "dumber" year so you could have a higher percentile with a lower score.

Now with computerized exams, they do not score your exam relative to the others taking it when you do... it is referenced to historical performance of the test items at which time a "difficulty level" for each item was assigned. This way, a 240 is a 240 no matter when you take it. Also, you do not have to wait for thousands of students to take the exam, do all your statistics, and then calculate the scores... you can calculate the score immediately. The implication of this, however, is that it is impossible to calculate a percentile since you are being scored against an inertia-laden grading curve. Every student could get a 240 now, and every student can fail.

There are many implications of this grading policy, and understanding it (really understanding it, not just saying that 240 is good and my score report says the mean was x with a sd of y) will serve you well.

afc
 
Re: The implication of this, however, is that it is impossible to calculate a percentile since you are being scored against an inertia-laden grading curve. Every student could get a 240 now, and every student can fail.

Hmm... The SAT is standardized against an administration from about forty years ago, with scores being renormalized by about 100 pts out of 1600 in the mid 90s due to the drift downward in student scores over the past half century. Just because the board scores are normalized based on an old administration, doesn't mean that descriptive statistics can't be calculated-- just give students scores at the time they take it, but don't report precentiles until six months later. Or, just report a mean and SD six months later and residency directors can calculate from there.

Also, since the minimum passing score is adjusted periodically, it's not really the case that everyone could fail or everyone could pass. Moreover, with a sample space of 15000 or so students per year, the year-to-year variation is never going to be more than a few points anyway, and the statement that "Every student could get a 240 now or every student could fail" is as technically accurate, but practically worthless as, "Our Sun could spontaneously supernova tomorrow and incinerate us all."
 
While the many eloquent explications proferred in this thread are educational and even enjoyable, the only thing that really matters is how PDs and their ilk view these scores. The several with whom I've spoken about this (my scores will be 6+ years old when I apply due to an intervening PhD - and, man, have things changed in terms of scores since I took Step I - so I've asked :) ) all know that the 2-digit isn't a strict percentage. Most even know that it "probably" isn't a perfect way to make comparisons between applicants. However, ALL have said that they just look at the 2-digit and go from there (meaning that that is all they use for comparisons).

That doesn't mean that the 3-digit score isn't used, but not one of them said they really look at it. It could be that others on their committees use them, as I only asked PDs (and one chair).

Have others talked with PDs/Chairs about this? With no disrespect to those who've explained how our scores are generated, I'm most curious about how the decision makers in other places use Step scores.

P
 
Originally posted by Primate
(my scores will be 6+ years old when I apply due to an intervening PhD - and, man, have things changed in terms of scores since I took Step I - so I've asked :) )

Off the topic Primate, but do you realise that you must take all 3 Steps within 7 years (so if your Step 1 scores will be 6+ years when you apply you'll have to take Step 3 right away)?
 
The 7 year rule depends on the state that you're taking step 3 in (and presumably doing residency in) if you're MD/PhD (and did the degrees simultaneously). Most that I've seen will have some sort of waiver or petition that can be sought if you qualify.
 
I've recalculated (simple math can be anything but) and my scores will be 6 years old when I GRADUATE (barring something dreadful happening during rotations), NOT when I apply. Mea culpa, but good point. I'll have the whole first year to take them.

If that fails, Pelican is right in that most (all?) states let you petition for a waiver if you've done the MD and PhD together. I'm hoping to avoid that, however. :D
 
Top