The actual vs predicted graduation stats support exactly what I am saying. In 2011 (the latest free version of USNWR I could find), Yale had a predicted 6-year graduation rate of 96% based on the stats of the incoming class. In that year, the actual graduation rate was 97%. So Yale’s graduation rate was almost exactly what we’d expect based on the impressive stats of the incoming class. Some of the other ivies didn’t fare as well and had a graduation rate that was below predicted based on the stats of the incoming class. If the grad rate is approximately the same as predicted based on high academic stats of the admitted class or in some cases a little lower grad rate than expected, it doesn’t imply it is “very difficult to fail” at that college. It only suggests the college is admitting stellar students who are unlikely to fail where ever they attend college.
I’ve taken classes at colleges with a wide variety of selectivities including Stanford, RPI, SUNYA, Syracuse, Wyoming, … Fewer failed at Stanford, but I did not get the impression that it was “very difficult to fail” at Stanford compared to the others, only that a smaller portion of students did failing quality work, and the few that did fail exams generally rectified the situation quickly. The one where I found it most difficult to fail was Wyoming, but that may have related to being an online program. Nevertheless, Wyoming had a substantial failure rate, likely the highest among the colleges I attended.