I have attached the screenshot for my results in Process Macro -- So the value of FRGN is "0" for all rows whereas my coding had options of "0", "1" and "2" -- does this mean when FRGN variable is "1" and "2" the results are not significant and only significant when FRGN is "0" ?
Does this mean that my moderated mediation is only significant when FRGN is "0" ? and not significant when FRGN is "1" or "2" ?
It seems that there went something wrong.
1. If you have a multicategorical variable (>2 groups) you need to specify this explicitly in PROCESS, otherwise it will treat it as metric (two groups will automatically treated as such).
2. When a moderator is treated as metic, by default 16th, 50th and 84th percentile of the moderator are used to calcualte the effect. Here you have no variation in the moderator and exactly the same values, suggesting that it may be singular? Without further knowledge about your variables and the rest of the output it is hard to make an educated guess. Can you provide more information?
P.S.: it just came into my mind, if you did not explicitly specified your moderator as multicategotrical AND your "0" category has much more cases (much larger n) than the other two categories, 16th, 50th, and 84th percentile would all have the value "0" and PROCESS would only show the indirect effect for this (which doesn't make any sense). Therefore, the moderator has not to be singular, but maybe with very different n for each group, with group "0" the by far largest?
Rainer Duesing thank you for the answer, yes that is the case. "0" has much more cases. But when I run Johnson neyman test, then it shows all categories i.e. "0", "1" and "2" and the effect is increasing when it is "1" and "2" . So I can conclude that as the results move towards "1" and "2" , the effect increases ? even though in summary of indirect effect it only gives "0" only. ?
The Johnson Neyman test does also make no sense, since you only have three categories. Just define the moderator als multicategorical and everything is fine. You just have to decide if you want to use dummy coding or something else.
- Badges
- Science topic