I want to test the research hypothesis that whether married and single individuals differ in well-being depends on their age. I chose to use the PROCESS procedure. In the model summary, the age * marital status interaction term is not statistically significant. However, when I look at the Conditional effects of X on Y at different values of the M, there are some statistically significant coefficients. Am I entitled to also report these differences although the interaction term in the model summary is non-significant?
Have you violated the assumption of homogeneity of regression slopes or why do you want to use J-N? The J-N is the strongest alternative to ANCOVA when the assumption of homogeneity of regression slopes has been violated. Why not use MANOVA, with several age groups; then you can read the main and interaction effects or check the link out: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3203541/
Thank you for your answer! I have learned about the JN approach as one alternative of investigating where are located the differences, is this not its use? I liked the idea of not distinguishing age groups beforehand, but let the analysis show me where the differences lie.
Hi Georgios,
Generally, if the interaction is not significant, I would not look at the J-N regions.
But specifically, the answer can get complicated. Significant regions can arise due to various data peculiarities or due to some theoretically meaningful trend that did not emerge because of lack of power. One might explore these regions and try to understand if they yield meaningful information that might be utilized in a next study.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS