Please help!
I am investigating the relationship between perfectionism, coping and stress as one part of my master's thesis. I conducted a hierarchical regression with maladaptive perfectionism in step 1 and avoidant coping in step 2. (based on significant correlation results)
Model 1 was statistically significant (with only maladaptive perfectionism in the model).
Model 2 was no longer statistically significant (with maladaptive perfectionism and avoidant coping in the model. Model Summary Table: Sig. F Change = .258.
Do I report the statistics (Beta etc) for Model 1 to explain that only maladaptive perfectionism is a statistically significant predictor of stress?
Do I need to report any findings from model 2, even though it is not significant? sould I report p = .26 or p > .05?
Below is what I have written so far, but I don't know what to say next.
The final two-step hierarchical multiple regression presented in Table 5 was conducted to test the ability of maladaptive perfectionism as measured by Discrepancy to predict Stress as the outcome variable, after controlling for Avoidant Coping. In Step 1, Discrepancy significantly contributed to the regression model, F (1, 101) = 21.97, p < .001, and explained 18% of the variance in Stress. When Avoidant Coping was introduced in Step 2, the model was no longer statistically significant, F (1, 100) = 1.30, p = .26/ >.05.
What you have written sounds good. What about a multiple standard regression analysis and then you give the tables with beta and B and t-test and p as well as the tests for multicollinearity and part correlation.
Béatrice Marianne Ewalds-Kvist thank you for the feedback. I conducted the standard multiple regression and found the same result - that maladaptive perfectionism is a significant predictor of stress but avoidant coping is not. I will report the findings as you suggested. I appreciate your time and consideration.
You write-up sounds fine. But I would provide a sentence as well addressing the correlation between your two predictors. If they are highly correlated with one another, it makes sense that including them together produces a nonsignficant model, and i would put them in two separate models. But if they are only moderately correlated, then what you have sounds fine.
One last note, I would also report R-squared change from model 1 to model 1 as opposed to F-change.
Cory
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression