From another
paper:
Acad Med. 2005 Oct;80(10):910-7.
Validity of the Medical College Admission Test for predicting medical school performance.
Julian ER1.
Author information
Results:
Grades were best predicted by a combination of MCAT scores and uGPAs, with MCAT scores providing a substantial increment over uGPAs. MCAT scores were better predictors of USMLE Step scores than were uGPAs, and the combination did little better than MCAT scores alone. The probability of experiencing academic difficulty or distinction tended to vary with MCAT scores. MCAT scores were strong predictors of scores for all three Step examinations, particularly Step 1.
CONCLUSIONS:
MCAT scores almost double the proportion of variance in medical school grades explained by uGPAs, and essentially replace the need for uGPAs in their impressive prediction of Step scores. The MCAT performs well as an indicator of academic preparation for medical school, independent of the school-specific handicaps of uGPAs.
-----
TBH I think it would be better if the following were to happen:
1. Schools independently decide on a pass/fail threshold for GPA since it doesn't really predict anything and is a much more predictor than the MCAT in terms of performance *in* medical school. These should be clearly advertised on a schools website. "3.2 and below need not apply" is enough.
2. Stop grading the mcat on a numerical percentile scale. As long as percentiles exist on a continuum, the 99%ile is going to fare better than the 95%ile no matter how much AAMC begs schools to stop caring so much about numbers. The MCAT should instead be graded on a tiered basis. Some more sophisticated maths would be necessary to figure out how to do this but for example: first you figure out a reasonable threshold for the McATs predictability of Step performance; if a 520 does not do significantly better than a 515 but a 515 does significantly better than a 510 then some threshold in between 510 and 515 should be chosen as a threshold. E.g 500-510 = A; 510-514 = AA, 515-520 = AAA; 521-528 = S. While schools will still clamor to fill their classes with the highest possible scorers, the small differences between nearby percentiles are erased (which is, in my opinion, the larger issue at hand). If people are borderline scorers (514, 520 etc) their score report should include a confidence interval like... 95% Confidence: AAA-AAA-S. Like for GPAs, MCAT thresholds for schools should be clearly advertised on the school website and the integrity of the information should be verifiable with the MSAR. "We will only consider applicants with AA mcat scores and above" for example.
3. All interviews should be a combination of MMI and personal interviews. The MMI correlates better with standardized clinical performance than either personal interviews or the MCAT. This is surprising but makes some intuitive sense given that it measures one's ability to reason given little (or conflicting) information and a short time to respond.
4. Like I said earlier, cap on the total number of possible applications. People shouldn't be applying to 30-40 schools. 15 is a reasonable cap. Decreasing the total number of applications does something in the ways of reducing the advantage some gain by simply throwing money at the application cycle, it decreases the app burden on medical schools because schools have to be more carefully chosen. Some superstars hold like 5-10 acceptances by the end of the cycle. This phenomenon is even worse in the MD/PhD cycle where if you read the posts following the cycle most accepted students hold 3-5 acceptances while 50% of all eventual MD/PhD students sit on 0 or on the waitlist. This is not good for students, waitlisted applicants, or the schools. This also discourages people from coming up with terrible school lists like "only the top 20 plus my state schools and I have a
LizzyM <70". Forcing people to think more carefully about their application choices will improve the experience for everyone involved, secondary revenue be damned.