Teasing apart the relationships between enrollment, graduation, and time on aid is difficult.
The literature on welfare use duration typically focuses on spell analyses using survival analysis. This type of modeling is not appropriate for examining duration across multiple spells, the measure
of time on aid most relevant in the current policy discussion. It is possible to use linear regression
models to examine the correlation between college enrollment and number of months on aid.
Because there appear to be few disparities in attrition across various subgroups, results should reflect
difference in time on aid rather than difference in time in the NLSY.