Hi everyone. I have to assess a moderation model through regression on spss based on Baron and Kenny's model (1986). However, i found in the initial evaluation of the model that the interaction term (non-meancetered values) is not statistically significant. Should I keep on with the evaluation of the mean-centered interaction term or should i make any alteration in the proposed model?
Hello Katerina,
The short answer is, yes, do try the centered interaction term. The reason for centering is to reduce (sharply) the redundancy/overlap of the interaction term's information with that of the variables from which it is generated. Significance of individual model variates, estimated regression coefficients, and overall model R-squared can change as a result of centering. That said, you still may find that the interaction term (centered or not) might not be noteworthy in explaining DV differences.
Good luck with your work.
The above suggestions make sense but,
The coefficient for the interactive term is the same for Mean centering vs. non mean-centering analysis. It changes the coefficients of independent and moderator variable.
In your case, you cannot get the significant interactive result with the mean centering one, but in general, you need to use mean-centering appraoch due to the issues Prof. Morse desribed above.
Katrerina it is assumed that you have developed the model on the basis of some theory and if there is still insignificant interaction then no need to worry at all because you only have to give argument and reasons that why that particular moderator did not moderate. Research does't force you to bring significant result.
Regards
M.imran
Please refer to:
1. Landau, S. and Everitt, B. S.(2004). A Handbook of Statistical analyses using SPSS.
2.Howitt, D. and Cramer, D.(2008). Introduction to SPSS.
Regards,
Zuhair
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression