One of the best assessments of college is the job and the paycheck. The chart above tells the tale (taken from here). Another table from the Wall Street Journal gives data by major, which is very interesting (hint: liberal arts comes out looking good)--see this post for details and comments. I argue there that the government is in the best position to do detailed studies by institution if they really are interested in learning outcomes. I had an opinion piece published about that idea in University Business a while back. Maybe the new administration would be more receptive to such an idea.
Another measure of success of higher education is the percentage of the population with degrees.
These are all solid, usable statistics with no statistical goo added. Notice that there aren't any averages in sight (unless you consider a percent to be an average over a binary variable).This contrasts with the generally opaque assessments of learning outcomes. Part of the allure of standardized testing is that at least the outcome seems to be clear: it arrives in the form of a crisp number like the ones above. These can be turned into pretty graphs. The problem there is whether or not they actually mean anything. I've argued that they can, if the complexity of what's being tested is appropriately low. Testing calculus (low complexity) may work, but testing critical thinking (high complexity) probably won't.
