Student Demographics and Fall-to-Fall Retention

Objective and Scope

Examine the correlation between various student demographics and fall-to-fall retention to help identify characteristics of students that, on average, have higher or lower likelihood of being retained.

Analysis included fall semester first-time post-secondary full-time bachelors degree-seeking students over the last five years and whether they returned to ISU the subsequent fall or not. The focus of the study revolved around student demographics including Age, Sex, Race/Ethnicity, First Generation and Pell Eligibility (proxy for income).

As many factors may contribute to retention that may also be correlated with these demographics it is essential to control for the externalities that would otherwise misrepresent the ISU equity (fairness) story. As such, a multivariate logistic regression was performed to better isolate if, after controlling for a closer to complete picture of the student, differences in demographics still exists. Control variables included various characteristics about the student prior to their arrival at ISU (e.g. high school activity), financial variables (e.g. aid received), and initial ISU activity (e.g. chosen course taking behavior). See the “Variables Considered” section for more information.

Key Findings

High School GPA was a better predictor of fall-to-fall retention than any student demographic considered in this study. Overall differences in retention between these different demographics appear to be more of a result of differing starting points, on average. For example, overall, males have lower retention rates than females. However, males start at ISU with, on average, a lower high school GPA. After controlling for high school GPA and other factors, the effect of sex was statistically insignificant. Statistically significant differences between certain demographics did exist, but with relatively small effect sizes, varied confidence intervals and sensitivity to changes in statistical modeling. A highlight of these findings can be found below:

Sex: Statistically insignificant.

Race/ethnicity: Students who are Hispanic were slightly more likely to be retained while students who were white were slightly less likely to be retained. In both cases, statistically significant but small effect sizes and sensitive to model changes.

Age: Older students were slightly more likely to be retained after controlling for other factors. Significant but effect size was relatively small. In addition, low sample size on older entering students is a limitation to this finding.

First generation: Students who were first generation college students were slightly less likely to be retained.

Pell eligibility: Students who were pell eligible were slightly less likely to be retained.

Variables Considered

The interactive tables below displays the various variables considered, how each was defined and how retention varies within the variable.