It means I have literally seen people with their intestines partially extruded from their abdomens who have complained less than you guys are about a $12 test that takes less than 2 hours. After finishing Step 1, Step 2 CK, Step 2 CS, Step 3, your specialty boards (+/- oral exam), possibly your subspecialty boards, and then getting fleeced by MOC for the rest of your careers, CASPer will seem like a youthful opium dream.
If you are referring to the paper linked by
@WheezyBaby above, that study doesn't even involve CASPer, but rather a homebrew SJT.
I also would not be so quick to judge a test based solely on the r. I'll give you an example. At my institution the MCAT is a poor predictor of preclinical performance as long as the scores are above a certain threshold. Go below that threshold land the MCAT quickly becomes a much stronger predictor of academic difficulties. We see a similar phenomenon with preclinical performance and USMLE scores, with the former being most predictive at the extremes, and much weaker in the middle (which comprises most of the students). Even if these associations are not super strong they are still useful, as they help flag individuals who are at higher risk of needing intervention.
You also have to look at the incremental validity of the test relative to its cost (both monetary and time). A quick, inexpensive test does not necessarily need to add a huge amount of value to a multivariate predictive model in order to justify its use.
To that end, it would be helpful to know what question(s) the people behind the GME pilot are actually asking, and what they are hoping to gain from trying CASPer. But neither of us are privy to that, so we are missing important information.
Look, there are plenty of problems to be had. MMIs all have systematic biases, unstructured interviews don't correlate with anything, but few question them because they are a familiar and time-honored way of doing things. Step scores have been misused for years. At the root of all this evil, however, is the phenomenon of over-application. Since no one wants to try capping residency applications, I think it will take the development of widespread preference signaling to bring this whole mess somewhat under control, and mitigate the need for more GME pilots.