Reflection Questions:

1. Which predictor had the strongest relationship with GPA?

According to the multiple linear regression test performed both stress and coping are statistically significant predictors of GPA. Adjusted R square of 0.621suggests that 60.8% of the variability in GPA can be explained by the predictor variables stress and coping.

Stress is a strong predictor of GPA, higher levels of stress were associated with lower GPA, where for every unit increase in stress GPA decreases by 0.04 (β = -0.671, p < 0.001). Additionally, coping moderately strong predictor of GPA, higher levels of coping were associated with higher GPA, where for every unit increase in coping GPA increases by 0.04 (β = 0.501, p < 0.001). These results suggest that coping may buffer the effects of stress on GPA.

The effect of stress was greater than the effect of coping.

2. Did the hierarchical model significantly improve prediction?

Adding coping and social support significantly improved GPA prediction indicated by a R squared change of 0.274 (p <0.001).

3. How well did stress and coping predict passing/failing?

Nagelkerke R square indicates that this model explains about 38.9% of the variance in pass/fail status suggesting a moderate level of prediction power. The overall accuracy of 80%, means the model performs reasonably well but struggles more with correctly predicting those who pass (46.7%) compared to those who failed (91.1%).

The binary logistic regression output indicates that as stress decreases the likelihood of passing decreases (B = -0.117), and the OR of 0.89 (95% CI: 0.782, 1.012; p =0.076) indicates that this is not statistically significant. However, as coping increases the odds of passing significantly increases (B = 0.363) with an OR of 1.4 (95% CI: 1.152, 1.792).

4. Was coping a statistically significant mediator?

Coping does not significantly mediate the relationship between stress and GPA because stress does not strongly predict coping, the indirect effect size is 0.0036 (95% CI: -0.0037, 0.0123)
However, coping independently improves GPA.

5. Did social support moderate the effect of stress on GPA?

The interaction effect (B = -0.0002, p = .8924) is not statistically significant, this means social support does not moderate the impact of stress on GPA. Furthermore the 95% CI: 0.0027 – 0.0024 includes zero, reinforcing that social support does not change the relationship between stress and GPA. Additionally, the suggests the moderation effect adds virtually no explanatory power to the model.

6. Were linear regression assumptions met?

The Scatterplot between standardized residuals and standardized predicted values produced a random cloud of points spread evenly around the horizontal line, demonstrating homoscedasticity.

The history of the standardized residuals had an approximately bell shaped distribution and Q-Q plot showed that points lie closely along the diagonal line with very minor deviations.

The VIF values for variables in the regression model, were all beneath 5 which suggest there’s no problematic multicollinearity in the model.