How is a 260 an 89?

[vsmedic]

---

Members don't see this ad.

I have been out of the loop for a few, and just came back to the experiences thread and am quite shocked. Can someone explain to me how a 249 use to be a 99 but now a 260 is a 89?

 

[AndyRSC]

Two-digit system was revamped last year. No real meaning behind it, just an arbitrary designation. It's no longer reported, so it's essentially meaningless.

 

[ijn]

I have been out of the loop for a few, and just came back to the experiences thread and am quite shocked. Can someone explain to me how a 249 use to be a 99 but now a 260 is a 89?

The NBME was really annoyed at IMGs claiming they had 99th percentile scores from 230s, so they fixed it.

 

[Phloston]

I'm not entirely sure (none of us are), but for some reason, just based on having seen a lot of the new scores, I have the feeling that it's the % correct on the test. This means the three-digit score is the relative one and the two-digit score is the absolute one. That's just my guess, because obviously it's not a percentile either, and if you think about the approximate # of questions people mark per block, it just seems right.

 

[Kaputt]

I'm not entirely sure (none of us are), but for some reason, just based on having seen a lot of the new scores, I have the feeling that it's the % correct on the test. This means the three-digit score is the relative one and the two-digit score is the absolute one. That's just my guess, because obviously it's not a percentile either, and if you think about the approximate # of questions people mark per block, it just seems right.

The two-digit score is not the % correct. It's simply an arbitrarily scaled score like the three-digit score where 75 is passing (and is equivalent to 188). They're phasing it out and residencies don't see it anymore anyway.

 

[AndyRSC]

The two-digit score is not the % correct.

Agreed. The absolutes were tabulated based on NBMEs (not adjusted for % relative to everyone else). They don't match up to the current two-digit system.

 

[Phloston]

Just to point out: neither of you guys know any more than I do. So no one's correct or incorrect here.

 

[AndyRSC]

You've already pointed this out, rather quickly, in your previous post.

You really think a commonplace 248 is 87%, a rare 262 is 90%, and the ridiculously rare 272 is 91% correct?
Again, the NBME % correlation doesn't match up. So while no one here may know, it certainly in no way seems to be correct.

 

[Nellyakgo]

The NBME was really annoyed at IMGs claiming they had 99th percentile scores from 230s, so they fixed it.

If this was the case, then those same IMGs could/would be claiming they are in the 89th percentile with a 260. If the 2 digit doesn't correspond to a percentile, I don't see how this fixes anything.

 

[Kaputt]

Just to point out: neither of you guys know any more than I do. So no one's correct or incorrect here.

No, you are definitely wrong.

http://www.usmle.org/frequently-asked-questions/#scores

The 2-digit score is not a percentile.The 2-digit score is derived from the 3-digit score. It is used in score reporting to meet requirements of some medical licensing authorities that the passing score be reported as 75. The 2-digit score is derived in such a way that a score of 75 always corresponds to the minimum passing score.

It is not a % correct

 

[tco]

You've already pointed this out, rather quickly, in your previous post.

You really think a commonplace 248 is 87%, a rare 262 is 90%, and the ridiculously rare 272 is 91% correct?
Again, the NBME % correlation doesn't match up. So while no one here may know, it certainly in no way seems to be correct.

Define commonplace. All of your terminology is relative. "Commonplace," "rare," and "ridiculously rare," all depend on your population.

I would also like to point out that, in this situation, there are 3 questions per percent. Why would it be so difficult to believe that a few questions could span such a large relative score range, especially when it's extremely unlikely that a test taker is going to score in the mid to high 90's (percentage correct, I mean)?

Just to reiterate what was said by you and others, there's no way to know for sure, so why are you treating this like it's a winnable argument?

 

[tco]

No, you are definitely wrong.

http://www.usmle.org/frequently-asked-questions/#scores



It is not a % correct

I take what I just posted back.

Apparently, it means absolutely nothing. Again.

 

[AndyRSC]

"Commonplace," "rare," and "ridiculously rare," all depend on your population.

The population would be us, people who have reported our scores on SDN. Since we're posting here, on Step I forum, I figured that'd be a given. References are one, two, and having followed scores here for years.

so why are you treating this like it's a winnable argument?

Common sense is so rare it should be considered a superpower. You really think that when that super-smart, super-dedicated student with a photographic memory and encyclopedic knowledge of Step I material comes along and scores in mid 270s, that he only got 91% of the questions right? Whereas the people who score in 240s (a hundred of them for each one in 270s) get something like 87% correct? Too narrow a gap in % between them and a rather arbitrary range, don't you think?

I take what I just posted back.

Bye, genius.

 

[Phloston]

I just had a good laugh at the hostility that developed through this thread.

Let's not forget that we're all teammates here.

 

[tco]

I just had a good laugh at the hostility that developed through this thread.

Let's not forget that we're all teammates here.

It's what happens when some people feel the need to prove their superiority when there is nothing to be superior about.

Common sense is so rare it should be considered a superpower. You really think that when that super-smart, super-dedicated student with a photographic memory and encyclopedic knowledge of Step I material comes along and scores in mid 270s, that he only got 91% of the questions right? Whereas the people who score in 240s (a hundred of them for each one in 270s) get something like 87% correct?

Yes, I think that it's numerically possible. Why does the three digit score have to be on a linear scale? Or even a scale that has a pattern at all? We don't really fully understand what the three digit score means, so how can we use that information to infer what the two digit score means? The link to the NBME website suggests that my original assumption (which is what we're all doing) was incorrect.

Bye, genius.

Bye.

 

[ijn]

If this was the case, then those same IMGs could/would be claiming they are in the 89th percentile with a 260. If the 2 digit doesn't correspond to a percentile, I don't see how this fixes anything.

I was joking. 😛

It's a two digit scaled score. We know for sure it's not a % correct because then everyone would know about the exact number of questions they need to get the score they want. We know for sure it's not a percentile because 260 is 98.5th percentile and not 89th percentile.

 

[Phloston]

We know for sure it's not a percentile because 260 is 98.5th percentile and not 89th percentile.

267 is 2SD above.

 

[rocketbooster]

I like how this statement is completely opposite on how residencies view the Step 1: "Percentiles are not provided in connection with USMLE scores. The calculation and provision of data to be used to rank or make comparisons among examinees is inconsistent with the primary goal of USMLE, which is to provide a series of assessments and recommendations for minimum passing requirements to the state licensing authorities to support decisions about initial licensure."

Too bad that's exact opposite on how residencies use the Step 1! They might as well give out the percentiles since they give out the total # of scores in each 10s category (230-240, 240-250, etc.). Plus, they give the mean and SD which can give you a pretty good approximation of your percentile.

Why not give out the exact percentile then? No idea. It's like their teasing us haha

 

[Kaputt]

267 is 2SD above.

My score report said mean is 222 and SD is 24. So 1SD = 246 and 2SD = 270

From the data in that one thread on the 2009 scores published with charting outcomes, there is a right skew to the data.

Thus, we can't accurately guess percentiles on scores based on SD because it's not normally distributed

Right? Or did I do a big biostats fail

 

[ijn]

267 is 2SD above.

Someone else did the research. They reported the 261+ cohort is in the 98.5th percentile of applicants in the NRMP. The NBME will never post actual percentiles.

 

[Chirurg]

Someone else did the research. They reported the 261+ cohort is in the 98.5th percentile of applicants in the NRMP. The NBME will never post actual percentiles.

this includes reapplicants with low scores, right?
so I'd rather go with Kaputt's score report. (My score report said mean is 222 and SD is 24. So 1SD = 244 and 2SD = 270)

 

[ijn]

this includes reapplicants with low scores, right?
so I'd rather go with Kaputt's score report. (My score report said mean is 222 and SD is 24. So 1SD = 244 and 2SD = 270)

It's the raw NRMP data, so it's everything and anything except for the non-NRMP residencies. Who knows though. Those lower re-applicant scores might be balanced out because the San Fransico match residencies are fields like urology and optho which have averages in the 240s.

 

[gravitywave]

It's the raw NRMP data, so it's everything and anything except for the non-NRMP residencies. Who knows though. Those lower re-applicant scores might be balanced out because the San Fransico match residencies are fields like urology and optho which have averages in the 240s.

that data interpretation you're referring to is flawed, NRMP data doesn't necessarily correlate with how the exam is scored. you can't just throw out all the residency slots that are being held outside the Match, and outcomes doesn't even have all the NRMP data anyway. OTOH, the all-in policy being implemented this year should make the charting outcomes data far more representative in future years.

i agree the two-digit isn't a raw percent correct, but my gut sense is that it's pretty close. similar to shelf scores in that respect.

 

[jatt12]

who cares, about 2 digit score. 3 digit is the BOSS 😎

 

[auburnO5]

who cares, about 2 digit score. 3 digit is the BOSS 😎

For some reason IMG's nut themselves over the 2 digit scores.