I use economic theory and estimates of a semiparametrically identified structural model to analyze the role played by credit constraints, uncertainty and preferences in explaining college attendance. A methodology for inferring information available to the agent from individual choices is proposed and implemented. The test distinguishes which of the unobserved (to the analyst) components of future outcomes are known to the agent and which are unknown to both at a given stage of the life cycle. I use microdata on earnings, schooling and consumption to infer the agent's information set and estimate a model of college choice and consumption under uncertainty with equilibrium borrowing constraints. I estimate that 80% and 44% of the variances of college and high school earnings respectively are predictable by the agent. Moving to a no tuition economy increases college attendance from 48% to 50%. When people are allowed to smooth consumption, college increases to nearly 58%. General equilibrium effects not withstanding, credit constraints have a larger effect than previously suggested.