Many of you know I'm now a professor after attending USNA and a lib-arts college (from which I graduated) in the late 80s/early 90s. From a professor's point of view, anecdotally, we (my colleagues and I) are also skeptical about these ratings because admincritters can, and do, game the systems.
For instance, you would think that if 10,000 students apply and 1,000 are admitted, then the admission rate is 10%, right? Not necessarily. Some admincritters can (and do) weight the numbers differently. They fudge. Perhaps that 1,000 admitted this year includes 100 who were admitted last year and deferred admission. Then only 900/10,000 of this year's matriculants applied this year. Voila! A 9% admission rate, highly selective. But wait! Perhaps in addition to those 10,000 applications there were 2,335 incomplete applications. Now your acceptance rate is 900/12,335, less than 7.5%.
This gets really messy in cost of attendance. My SLAC, like many LACs, has a relatively high tuition (high $20Ks). However, we are also a need-blind school. Students are admitted based on merit, not ability to pay, and only students from very very well-off families pay sticker price. My SLAC is also great about not saddling students with student loans (we have a cap). So, while my SLAC and others look staggeringly expensive, financial aid is generous and our kids don't have to work 20 hours a week to pay for books and food.
And speaking of food...at Nearby U, which is trying very hard to move up in the Princeton rankings, there was a minor flap last spring when the food service company (it was revealed) served students a gourmet dinner, asked them to rate the meal, then submitted those ratings as Nearby U's "quality of food" data.
More and more high-quality colleges are opting out of these kinds of ratings for just these reasons - it's too easy to game the system, and what parents and prospective students see as a result is misleading or untrue information. As with anything you're going to invest 4+ years and tens of thousands of dollars: caveat emptor, both in the choice and in selection methods.