I am currently working on the results of a 2x2 design. I used a MANOVA and there was only one significant main effect of one condition (the main effect from the other condition is unsignificant and no interaction effect has been found). However, the follow up ANOVAs for the non-significant main effect present some significant results.
Is it possible to comment follow up ANOVAs despite the related main effect from the MANOVA turned out to be not significant? If so, which are the precautions to be taken when commenting such results?
The MANOVA is telling you that main effect A becomes negligible when main effect B is held constant (is controlled statistically), even though the ANOVA shows a significant relationship between A and the response variable (Y) when there is no control for B. Is there a theoretically plausible causal order linking A, B, and Y? If A causes B and B causes Y, then the results say the effect of A on Y is indirect, mediated through B. But if B causes both A and Y, the the same results say the relationship between A and Y is spurious, an artifact of their common cause B.
Suggest you read the attached before you do anything else. Best wishes David Booth
Yes, you can comment on the follow-up ANOVAs despite the non-significant main effect in the MANOVA, but interpret the results with caution. The significant results in follow-up ANOVAs may indicate that the individual dependent variables show significant differences that the overall MANOVA did not capture. When discussing these results, mention the discrepancy between the MANOVA and ANOVA findings and consider potential reasons for this difference. Be cautious when drawing conclusions and consider additional analyses to validate the results.
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication