Why Universities Have Different Graduation Rates Than Predicted: A Residual Analysis Approach

This paper examines differences in graduation rates at colleges and universities using both quantitative and qualitative analysis. Building on previous research this paper first uses regression equations to predict what variables influence graduation rates. After controlling for dozens of variables such as standardized test scores, this analysis then looks directly at those schools with significantly higher or lower predicted values. These schools are examined to see what variables might be important in determining graduation rates that are not easily quantified or are just not in data sets. In the quantitative section, results confirm that socioeconomic status, selectivity and other well-established variables highly influence graduation rates. The qualitative research suggests that information such as how well known schools are may influence outcomes, such as those with visible sports teams over-perform smaller, less well-known schools that do not have a visibility beyond academics.