Step 1 Score Distribution and Percentiles - REAL DATA

[RapplixGmed]

I summed the distributions of everybody who applied to the match in 2009 to give the distribution of scores and percentiles for step 1, which nbme unhelpfully did not provide for us in our score reports. Just giving a mean and sd is unconvincing for assuming normality so I didn't want to rely on that for calculating percentiles. The lower end of the curve is truncated but it should give you guys a rough idea based on real data. Keep in mind the percentiles are for people applying for the match and not for people taking the test.

---

Members don't see this ad.

 

[DrBowtie]

It looks about "normal" to me especially considering those on the low low end might not have applied for the match.

 

[Lizloveslilies]

Very nice! Looks a little positively skewed. I wonder why there is so much secrecy/ambiguity when it comes to the NBME releasing test statistics and percentiles...

 

[EctopicACTH]

solid work

 

[airmax]

wait, so the average isnt 245 like the sdn self report thread makes it look like?!?

any idea if they calculated this for the 2010 or 2011 match? it should be about the same tho since the nat aves were similar for those 3 years.

 

[ArcGurren]

Solid work, you should submit this to the NBME or to the First Aid authors to see if they can publish it as a rough percentile measurement. It looks normal-ish with a right skew.

 

[lackli]

Solid work!

Question: If you use scores of those who apply to the match, does that include applicants who might be reapplying?

If so does that mean there would be a greater number of scores on the lower end of the spectrum that would skew your percentiles?

Not trying to undermine you; actually the percentiles this way is more useful as it gives a better idea of the level of competition that you'll face during match.

 

[RapplixGmed]

Solid work!

Question: If you use scores of those who apply to the match, does that include applicants who might be reapplying?

If so does that mean there would be a greater number of scores on the lower end of the spectrum that would skew your percentiles?

Not trying to undermine you; actually the percentiles this way is more useful as it gives a better idea of the level of competition that you'll face during match.

Yes, I believe you are exactly right but I think that is already included in the data (not completely sure though). As you see, there are very few people who applied into the match with failing step 1 scores. I'm quite sure most of the people who applied in the 181-190 bin received a passing score. The curve is trucanted below the fail point because those people either didn't apply or retook and received a higher score. If you assume that the majority of retakes still score below average, then what you say is true.

 

[RapplixGmed]

any idea if they calculated this for the 2010 or 2011 match? it should be about the same tho since the nat aves were similar for those 3 years.

Unfortunately, the source where I got this data from doesn't exist for 2010 or 2011.

http://www.nrmp.org/data/index.html
Look for charting outcomes of the match.

Don't know when they'll make a more updated version of this study.

 

[Encephalectomy]

awesome work!

And even though the percentages aren't accurate for the test as a whole, it is perfect for the things that we're concerned about, which is what percentile you are among people applying for residency. Thanks!

 

[Gullain Barre]

Thank you for the analysis. Very interesting data to ponder as I wait for my score.

 

[Torquer]

RapplixGmed where can i access this distribution you've created? doesnt seem to be a link anywhere..

 

[Rothbard]

nm

 

[womp]

Couple of points. If you add up the total number of applicants provided in charting outcomes, the number is 28k-29k. Compare that with the 26-27k in the OP's post.

Also, scores seem to go up 5 points every few years. So for anyone applying for residency in 2013, scores will probably be 5 points higher than shown here.

5 points every few years? That doesn't even make sense since the scores are not raw. The only thing that has gone up is the pass level, but that is probably to account for the fact that hardly anyone gets below a 180, which was probably intended that way but some of the PDs probably complained that those getting the low 180s weren't up to their stuff.

The mean has been around 218-222 for like ever.

 

[WashMe]

Couple of points. If you add up the total number of applicants provided in charting outcomes, the number is 28k-29k. Compare that with the 26-27k in the OP's post.

Also, scores seem to go up 5 points every few years. So for anyone applying for residency in 2013, scores will probably be 5 points higher than shown here.

Not sure what the real relevance of this statement is... there are several possibilities here. Perhaps the data to which you're referring:
(1) Includes people who took step I but didn't apply for the match.
(2) Includes international students; I'm not sure if the OP's data includes IMG's.
(3) Is from a different year or set of years (which does NOT affect the analysis herein).
(4) Includes people who entered the SF match (uro, plastics, optho, maybe something else) and didn't go through ERAS.

Also, the scores for a given residency may go up over time, but the national average has only increased marginally (a few points) in the last 10-15 years and the SD is basically the same. The percentiles in the chart are thus still valid, and completely unrelated to the average score for any given specialty.

 

[The Coon]

Couple of points. If you add up the total number of applicants provided in charting outcomes, the number is 28k-29k. Compare that with the 26-27k in the OP's post.

Also, scores seem to go up 5 points every few years. So for anyone applying for residency in 2013, scores will probably be 5 points higher than shown here.

The mean score hasn't really gone up much, what has gone up is the score to get into different specialties. This is caused by increased numbers of medical students each year, but the same amount of residency slots.

 

[DocEspana]

Not sure what the real relevance of this statement is... there are several possibilities here. Perhaps the data to which you're referring:
(1) Includes people who took step I but didn't apply for the match.
(2) Includes international students; I'm not sure if the OP's data includes IMG's.
(3) Is from a different year or set of years (which does NOT affect the analysis herein).
(4) Includes people who entered the SF match (uro, plastics, optho, maybe something else) and didn't go through ERAS.

Also, the scores for a given residency may go up over time, but the national average has only increased marginally (a few points) in the last 10-15 years and the SD is basically the same. The percentiles in the chart are thus still valid, and completely unrelated to the average score for any given specialty.

Nah. none of those 4 apply as the OP and the commeter you quoted are both discussing only applicants, the data includes all sorts of international students, they're both citing the 2009 numbers and none of them are including the early match programs. I know what little math anomaly was referring to. Its loss to rounding, I think. The same source that provides the number of people to each specialty also provides a number of people all-together. The numbers within their own document, when added up, dont match up to the total they claim for whatever reason. I've noticed it, but never actually questioned it. Just assumed it was due to rounding. I imagine I'm right, but other people can take time out and confirm or deny this.

EDIT: It might also simply be some small percentage of drop out at each level of measurement. Went back to check the document. The document cites a source early on as listing the total applicants (USMD, USDO, Foreign grad, 5th pathway, Re-applicant, USIMG) as being 29,888 people. Two or three pages later it breaks down the applicants by specialty using a different source and has only 28,717 applicants. I've noticed previously (though I am *not* going to sit around for 10 minutes adding up the values to get a number) that the more detailed breakdowns for each specialty, when all added together, is even less than the 28,717.

 

[Rothbard]

nm

 

[RapplixGmed]

I'm going to bump this for this year. Cheers and good luck guys!

 

[shan564]

Thanks for the bump and the effort. Looks like some interesting data...

 

[docjohn101]

anyone have the data for percent of people who pass and fail step 1?

A link would be awesome too.

 

[Aclamity]

This is nice. Anyone know if NBME or NRMP came out with official percentiles for different scores?