I was wondering if someone could help me interpret an interaction.
I'm using Hayes' PROCESS (SPSS) and have one categorical IV (gender) which is dummy coded 0 = female (blue line), 1 = male (red line), and one continuous moderator. The interaction is significant but negative.
When I've probed it, simple slopes shows it's only significant at +1 SD, b = -0.87, SE = 0.41, p < .05.
I guess my question is, is the beta of "-0.87" an indication that it's the female group that's significant?
Before posting I swapped the IV for the moderator, and vice versa in PROCESS. The result showed only a sig. result for the female group (0), this time with a positive beta. I'm convinced it's to do with the coding. Is anyone able to shed some light on the matter?
Thanks guys.
(Not sure this is entirely what you're asking so apologies in advance if my answer below does not provide the information you're looking for.)
I'm assuming that the continuous moderator is denoted as X.
The beta for the interaction (product) term X*Sex tells you whether there is a difference in the slopes between males and females. A beta of zero for the interaction term X*Sex would indicate that the slope coefficients are the same for both males and females. In that case, the regression lines are parallel for males and females (i.e., they do not differ across the two groups; there is no interaction).
Here, the slope for males is negative, whereas the slope for females is positive, so there is a non-zero difference in the slopes between groups. The beta for the interaction term X*Sex should thus be non-zero. If this beta is statistically significant, this means the slopes differ significantly between males and females. The sign of the beta coefficient is related to the coding of the groups (i.e., which group is coded as 0).
To find out whether the relationship between the continuous moderator (X) and your DV (Y) is significant in Group Female, just look at the beta for X (assuming female is coded as 0). The beta for X gives you the relationship in Group Female (or, more generally speaking, the group that's coded as 0). The p value for that beta indicates whether the relationship between X and Y is significant in Group Female. (You could do the same for males by recoding gender as 0 = male, 1 = female and then looking at the beta for X again to find out whether it's significantly different from zero in Group Male.)
I hope this helps.
The term to examine is the interaction effect, X*Sex -- if it is significant, than the slopes are different.
FYI this is a very simple regression to specify and interpret. Using something as complex as the Hayes macro may be getting in your way rather than helping.
Thank you Christian Geiser and David L Morgan for your responses. I really appreciate the depth of detail that you provided Christian, it certainly helped answer my question regarding coding. And David I think you're right, using PROCESS has probably just complicated matters. Thank you both!
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS