I used a hierarchal model, where my model testing only my main relationship is not significant.
and my models where I test my interaction effects separately do show the interaction effects as being significant, however in both those models my main relationship is still not significant ( and quite contradictory indicating the opposite effect as what is regarded as the norm) .
If the significance of my moderators (as interactions) do have value, how would I approach it in my analysis?
Do I state due to the moderators (as interactions) being significant that the main relationship does show significance but that it is conditional on the moderator being present? Since the moderator is the independent variable x moderator.
Are there any papers or books I could reference in this scenario?
An everyday scenario may help.
The annual average temperature of Death Valley and Honolulu are both around 77 degree F, but one place is known for it's harsh weather and another for mild weather because if you look at the temperature by month and/or by day/night, you'll see the difference--a life and death difference if you are a human trying to travel across the regions. You may consider this a case where the main effect is not significantly different, but the interaction of temperature and month or temperature and hours are significantly different.
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation