UWorld vs NBME Score Prediction

[testt]

This site gets thrown around a lot, https://usmle-score-correlation.blogspot.com, and I was wondering if anyone here had a large mismatch between their NBME predicted score and the UWorld predicted score on that chart. If you took the Step which one did you find more accurate?

Currently my I am almost done with UWorld, doing it random untimed and my average is around an 83% which according to that chart is like a 265. My NBME predicted score, however ,is a 250 average (took 2 NBME so far).

I know it sounds crazy neurotic but I'd like to get into a pretty competitive specialty/residency so the difference in score is pretty important. Am I doing something wrong with the NBMEs? Or is UWorld an unreliable source for a predicted score?

---

Members don't see this ad.

 

[TheOther]

This site gets thrown around a lot, USMLE SCORE CORRELATION, and I was wondering if anyone here had a large mismatch between their NBME predicted score and the UWorld predicted score on that chart. If you took the Step which one did you find more accurate?

Currently my I am almost done with UWorld, doing it random untimed and my average is around an 83% which according to that chart is like a 265. My NBME predicted score, however ,is a 250 average (took 2 NBME so far).

I know it sounds crazy neurotic but I'd like to get into a pretty competitive specialty/residency so the difference in score is pretty important. Am I doing something wrong with the NBMEs? Or is UWorld an unreliable source for a predicted score?

I'm not sure how well the data was put together for that website, but a good number of people seem to have found it pretty spot-on. One thing to keep in mind though: the data from the website is retrospective while the NBMEs measure your current progress. So your UWorld % likely puts you at a much higher potential range than someone who, let's say, averages 50% on their first random pass. Hypothetically, if you had 2-3 weeks left to study for your exam, it would be reasonable for you to bring your NBME scores from a 250 up to a 265.

I'm in the middle of Step 1 prep too so these are just some thoughts I had. Best to you.

 

[Kalydeco]

This site gets thrown around a lot, USMLE SCORE CORRELATION, and I was wondering if anyone here had a large mismatch between their NBME predicted score and the UWorld predicted score on that chart. If you took the Step which one did you find more accurate?

Currently my I am almost done with UWorld, doing it random untimed and my average is around an 83% which according to that chart is like a 265. My NBME predicted score, however ,is a 250 average (took 2 NBME so far).

I know it sounds crazy neurotic but I'd like to get into a pretty competitive specialty/residency so the difference in score is pretty important. Am I doing something wrong with the NBMEs? Or is UWorld an unreliable source for a predicted score?

In my experience, NBME scores are the better predictor (though for me, the UWorld self assessments were the best). People approach UWorld in different ways, which is in my opinion the flaw in the assumptions powering that chart. To use my personal experience as an example, I was much more concerned with learning from the question explanations rather than answering questions correctly, and I finished UWorld (tutor mode, random) with something like a 64%, correlating to a mid-230s score on the chart. I outperformed that "prediction" by more than 20 points. Ultimately, I think you should set your expectations to align with your scores on NBME exams and UWSAs. I also think it's ridiculous to "expect" a 265, for two reasons: pretty much no one is guaranteed a score that high because there is an element of luck in which 280 questions you happen to get, and if you're expecting a 265, why keep studying at all? Just spend the rest of your dedicated study time on a beach in Tahiti.

But, beyond this whole discussion of predictive value that seems to dominate the SDN step 1 conversation, it's a much better use of the time to study areas where you're relatively weak than to obsess over what predictive algorithm best approximates your score.