I am facing a situation wherein both the direct effect and indirect effect are insignificant but moderation effect is significant between the same set of independent (X) and dependent variables (Y).
Attached are the results.
How to interpret this scenario?
P.S. I am using covariance-based SEM. I am not testing mediated-moderation, but plain mediation and plain moderation between X and Y.
I'm not sure I understand the question. Moderation and mediation are different concepts. Its fine to have moderation without mediation and vice versa (and indeed mediated moderation). It just means that the direct X variable and moderator effects on Y are not additive - the overall effects aren't simply X + M but depend on the combination of the two.
Its typical to think of this as the moderator influence the strength of the effect of X on Y but statistically there is no distinction between X and M in the model. So its just as valid to think of X influencing the strength of the effect of M on Y.
Dear Thom Baguley ,
It is ok to have a statistically significant moderation effect rather than mediation or indirect impact. Your findings are justified automatically.
@Harmanjit Seeing at your results, it seems that the direct relationship between X and Y (primarily insignificant) is moderated in such way that it becomes significant at high level of moderator. It would be interesting to know your IV, DV and Moderator and also the type of data for thrice of them (Likert/Categorical/Binary). Moreover, you also need to mention what method of interaction you used (+1/-1 SD or Percentile).
Thanks Imran Anwar
I am not comfortable in sharing the variables as of now.
However, the scenario which you are describing, i.e. "the direct relationship between X and Y (primarily insignificant) is moderated in such way that it becomes significant at high level of moderator"; are there any similar cases in literature which you have come across? If yes, please share.
Regarding the second part of your query, all variables are continuous (Likert). Also, I used +1/-1 SD method to check moderation.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling