We collected personality (Big Five) and demographic characteristics, and ran incentivized experiments measuring cognitive skills (non-verbal IQ, numeracy, backward induction/ planning), and economic (time, risk) preferences, with 100 students at a small public undergraduate liberal arts college in the Midwestern US as part of a larger study that collected the same measures from 1,065 trainee truckers. Using standardized (z-score) versions of our variables we analyze their relative power to predict (1) timely graduation (four years or less), (2) graduation in six years or less, and (3) final GPA. The proactive aspect of Conscientious (but not the inhibitive one) has a large and robust positive effect on all three outcomes, and Agreeableness has a robust negative effect on both graduation outcomes, but not on GPA. Economic time preferences predict graduation in four years, and GPA. Cognitive skill measures predict as expected if entered individually in a multivariate model, but when all variables compete it is only our backward induction measure ("Hit15") that weakly predicts graduation in four years, and strongly predicts graduation in six years. Trainee truckers work in a different vocational setting and their results are appropriately different, but there is a common element: Hit15 also predicts their job success (completing a one year employment contract that makes training free). We interpret Hit15 as capturing a specific part of the cognitive skills required for self-management in non-routine settings – thinking backward from future goals to make the best current choice – that is not well measured by existing instruments, and suggest this deserves further scientific and institutional scrutiny.