One flaw is the failure to control for intelligence. Or socioeconomic background. Or any number of other factors that influence lifetime earnings that also positively correlate with college attendance.
If you were to only look at people who grow up in a very-middle class household (a very ill-defined term I admit) and who scored in the X to Y percentile (not sure what good numbers would be) on some measure of intelligence then I am nor sure that the numbers would be clearly in favor of college.
Additionally, to observe the value of a college education, you have to compare it to education received without going to college, while excluding the people who choose to not go to college because they were not interested in furthering their education. The high school graduate that spends hours every evening scouring the internet, or the library, for educational purposes is not comparable the person of equal formal education who goes home to watch sitcom TV each night.
If you were to only look at people who grow up in a very-middle class household (a very ill-defined term I admit) and who scored in the X to Y percentile (not sure what good numbers would be) on some measure of intelligence then I am nor sure that the numbers would be clearly in favor of college.