O
ohio FMG
With the new CK format and revised question pool are the NBME Step 2 CK self assessments still a valid indicator of how you'll score? If so, which test is more representative of how you'll perform on the new exam?
in light of the usmle people discovering their q bank had been breached (via that test prep company...you can see it on the usmle website) i think this new 44 format ck test is from a brand new question pool. i think any old nbme tests, uw, kaplan, and any other q banks are no longer good predictors of scores. just my theory (and based on my experience with the exam)
in light of the usmle people discovering their q bank had been breached (via that test prep company...you can see it on the usmle website) i think this new 44 format ck test is from a brand new question pool. i think any old nbme tests, uw, kaplan, and any other q banks are no longer good predictors of scores. just my theory (and based on my experience with the exam)
Your reasoning doesn't make any sense if you think about it. The nbme tests were good predictors of performance for CK prior to the change from 46-->44. I think it's very safe to say that performance on CK a month ago would correlate well to performance on the new CK (for example, someone who scored 250 a month ago would most likely score very close to 250 this month and vice versa). Because of those facts, I have to assume that the nbme exams will still correlate closely to performance on the new CK tests. The same thing happened last summer when we took step 1, they were changing from 50-->48 questions and I don't think there was any change in scores.
Even if they did rewrite the entire USMLE Step 2 CK question pool (I'm sure they didn't), they would rewrite them in a way that the difficulty and subject matter was basically the exact same and by the time they had a large statistical sample to work with, the scores would end up being the same. So, yeah, the NBME exams are still fine.
I suppose the main point is "am I going to pass, or how am I going to score?". I agree with all of the opinions here, they all sound reasonable, and they wouldn't need to write 14,000 new questions in the first month, 2-3000 would do for now.
Either way, the length of statistical input of test takers into the score estimator bank would still give you fairly accurate results, and I'd think more accurate if you average somewhere between your Uworld predictor and NBME predictor.
I just took it, and my test had very few "most common", or easily recognizable scenarios. 80% were longer, and more confounding distractors than 90% of Uworld questions. I had a very hard time staying on time.
Get a lot of sleep.