What is the appropriate way to report the relationships among three interrelated variables in an observational study using GLM or mediation models?
Variables: Depression, Neuroticism, Mindfulness
Regression models - All of the relationships are statistically significant and passed assumption checks.
Mindfulness predicts Depression: R2 = .247, unstandardized coefficient = -5.32
Mindfulness predicts Neuroticism: R2 = .503, unstandardized coefficient = -3.88
Neuroticism predicts Depression: R2 = .732, unstandardized coefficient = 1.02
Should I simply report all findings, starting with the "strongest" relationship, Neuroticism predicts Depression?
Multiple Regression models - All of the relationships are statistically significant and passed assumption checks. Is there a benefit to reporting these model results beyond what is gained from the above regression models?
Mediation models - All of the relationships are statistically significant. All Sobel test results are statistically significant.
See attached file with results of mediation models and Sobel tests. Would I report all models or only that Neuroticism mediates the role of Mindfulness on reducing Depression, and Depression mediates the role of Mindfulness on reducing Neuroticism?
What is your hypothesis? It might be, for example, that mindfulness moderates the relationship between neuroticism and depression.
I'd strongly recommend drawing a causal diagram (wonderful online tool here http://www.dagitty.net which is really helpful in defining the relationships implied by your model).
- Badges
- Science topic
- Similar topics
- Psychology