We can model this relationship as follows:
\[ \operatorname{Next Term GPA} = \alpha + \beta_{1}(\operatorname{Prior Cum. GPA}) + \beta_{2}(\operatorname{N terms abroad}_{\operatorname{1\ Term}}) + \beta_{3}(\operatorname{N terms abroad}_{\operatorname{2\ Term}}) \]
The model explains about 30% of the variance in the next term’s GPA.
| Characteristic | Beta | 95% CI1 | p-value |
|---|---|---|---|
| Prior Cum. GPA | 0.89 | 0.80, 1.0 | <0.001 |
| Terms Abroad | |||
| Zero | — | — | |
| 1 Term | 0.11 | 0.02, 0.20 | 0.022 |
| 2 Term | 0.10 | 0.02, 0.18 | 0.017 |
|
1
CI = Confidence Interval
|
|||
Before 2012, there was a relationship between Study Abroad participation and Pell eligibility. Now roughly the same proportion of Pell-eligible and non-eligible students participate in Study Abroad, either for one or two semesters.
\[ \log\left[ \frac { P( \operatorname{grad} = \operatorname{1} ) }{ 1 - P( \operatorname{grad} = \operatorname{1} ) } \right] = \alpha + \beta_{1}(\operatorname{Prior Cum. GPA}) + \beta_{2}(\operatorname{N Terms Abroad}_{\operatorname{1\ Term}}) + \beta_{3}(\operatorname{N Terms Abroad}_{\operatorname{2\ Term}}) \]
| Characteristic | OR1 | 95% CI1 | p-value |
|---|---|---|---|
| Prior Cum. GPA | 4.90 | 3.24, 7.58 | <0.001 |
| Terms Abroad | |||
| Zero | — | — | |
| 1 Term | 3.09 | 1.68, 6.08 | <0.001 |
| 2 Term | 1.91 | 1.19, 3.11 | 0.008 |
|
1
OR = Odds Ratio, CI = Confidence Interval
|
|||
Odds ratios in logistic regression are not as intuitive as probability. This resource may assist with interpretation.↩