A reliable and valid technique for measuring college participation rates is proposed, based on national data. The objective is to determine if factors other than state demographic factors systematically influence college attendance by state. Multiple regression equations are used to predict enrollments by using demographic variables such as population by gender, race, and income. The difference between the expected enrollment and actual (residual) is mapped for the 50 states. Using Higher Education General Information Survey data, enrollments are evaluated for females, males, first-time freshmen, full- and part-time students, and for private and public institutions. Major findings are as follows: states with more students in public institutions than predicted tended to be states with lower than predicted enrollment in private institutions, and vice-versa; higher than predicted states in one category of enrollment tend to be higher in other categories except for public versus private, while lower states tend to be lower on other categories also; part-time enrollment is the most variable in actual enrollments and in residuals; states with lower enrollments than predicted tend to border states with higher than predicted enrollment; and the Plains states tend to be lower than predicted on several categories. (SW)