Forecasting credit hours

This paper employs Tobit to estimate retention probabilities and credit hou rs at two universities. The innovation is that this technique examines cred it-hour choice with the decision to depart the university treated as the ch oice of zero credit hours. Tobit is appropriate for this problem because it recognizes the tower bound of zero on credit hours and incorporates this b ound into the parameter estimates and forecasts. Models are estimated for c redit hours in a single year and cumulative hours over a six-year horizon.