Regarding the recent posting by 'abs':
well, when you annihilate the exam down to the differences being 'only' 100 questions separating the Mean from the 90th percentile, sure, that sounds much closer than the percentile scores might suggest. Why not continue (and be correct) and state that somewhere near 90-95% of all test takers scored between 400 and 600? That makes things sound even closer that all test takers were essentially within 200 points of each other? My point is that, of course, the majority are withing the middle of the Gaussian curve; that's inherent in the design of the exam, from the test writers to the reviewers from the various Colleges who sit for the COMLEX as Faculty Reviewers. Every student taking the COMLEX Level 1 in June, regardless of what test 'form' one received (e.g., there were 2 different examinations at my institution) had his/her score normalized against everyone else taking that particular examination. The Mean score was still scored a 500, per NBOME design, regardless of the examination taken.
I myself did well on this exam, but I can confidently say that those in my circle (myself included) who likewise did well commanded a stronger understanding of basic science concept and data than those who did closer to the Mean. The COMLEX is not perfect, but inherent in its design (the length, you'll recall, was 750 questions this year: 187-189-187-187) is an attempt by the NBOME to offer questions of varying difficulty such that good guessing and chance are eliminated as mitigating factors...*for the majority of test takers*. The purpose of the exam's design is to develop as good a picture as possible of the distribution of scores over the ranges, with minimal number of examinees falling below 400 or above 650.
I will not deny that, yes, there are some who guessed well for nearly 16 hours this past June, nor will I deny there were those whose luck was down for those same 16 hours and guessed poorly. What I will refute, on the basis of statistics, is that all receiving good scores were necessarily good guessers, and those who did less well were any worse guessers. Given to 50-100 people, this notion may hold more water; given to over 2000 students, I highly doubt it, though I will say that there probably are a large number of students who have greater or lesser abilities for endurance. I am the first to say that there *were* many questions that were "poorly" written; however, everyone taking the test had the same ambiguities, misunderstandings, and contradictions with which to contend over those 16 hours of examination.