This is interesting, but it doesn't prove college is "worth it". We know that at least 40% of the wage premium of attending college (and possibly considerably more) are due to ability bias - smarter people get paid more and smarter people go to college.
(I do suspect there is some individual benefit to college, from some mix of beneficial skill increases and socially harmful signalling. But the data in this article doesn't begin to address this question.)
Absolutely, correlation between college degree and employment does not show any causation between the two. Not only is intelligence a factor in going to school and getting a job, but so is parental income, responsibility, upbringing, etc. If only more journalists had taken a statistics class while they were at college we might not see these same fallacies repeated over and over in the media.
Separating the college group to reflect marks could be a little more interesting, compare under achieving but passing college students to those who didn't attend. Of course there will still be bias but it gives a little better picture on how much smarts plays into it and how much just having the piece of paper plays into it.
http://econlog.econlib.org/archives/2012/01/correcting_for.h...
(I do suspect there is some individual benefit to college, from some mix of beneficial skill increases and socially harmful signalling. But the data in this article doesn't begin to address this question.)