I am doing a study on the relationship between religiosity and visitation to religious healers. Before adding interaction term in the regression, the multilevel model showed religiosity as significant, but after I added the religiosity variable (i.e., frequency of prayers) as interaction terms with other control variables education, gender, urbanity, and such, the original variable frequency of prayers turned insignificant.
- what am i doing wrong here?
- is it because I added too many interactions with it?
- should I use another religiosity indicator (e.g., frequency of reading the Bible) along with the frequency of prayers as an interaction term with education, gender, urbanity?
It would be really helpful for me if someone tell me what to do in this situation.