I'm still learning and confusing to choose which method to use for my study. What is the simple method should i use if i have 3 independent variables, 1 moderate and 1 dependent variable
It depends (primarily) on the scale level of your dependent variable (DV). If your DV is continuous (metrical, interval level), you could use multiple linear regression analysis, specifically, moderated linear regression. If your DV is categorical, some version of (moderated) logistic regression would be suitable.
In either case, examining a moderator (interaction) effect in regression involves including one or more product (interaction) terms to reflect the potential interactive (moderator) effect of one or more variables. For more information, see
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park: Sage.
Cohen, J., Cohen, P., West, S. G. & Aiken, L. S. (2003). Applied multiple regression/correlation analysis for the behavioral sciences. Mahwah, NJ: Erlbaum. (Chapters 7 and 9)
Hayes, A. F. (2018). Introduction to mediation, moderation, and conditional process analysis (2nded.). New York: Guilford Press.
Create three interaction terms between each of your IVs and moderator. That is IV1*M, IV2*M and IV3*M.
Now you can run a linear regression including all the IVs, interactions and moderator.
If an interaction term is significant it suggests M moderates the effect of the corresponding IV on Y.
This assumes Y is a variable with continuous measurement properties.
Hope that helps
Good luck..
I would also add to consider your sample size (total number of rows or patients) you have. This will also impact your final choice in terms of model complexity.
my DV using interval scale (5 point Likert scale), how about PLS (SEM) Christian Geiser , is it more comprehensive compare with moderate analysis in SPSS? but I'm not familiar with PLS (SEM), and from my reading most of the article that have moderating in their model, they are using PLS (SEM) to analyze it. based on my observation also that PLS (SEM) quite complicated and I'm afraid that i cant justify in the report. which method is appropriate for my study?
Paul F Burke did you suggest to use moderated analysis? if more than 2 IV, i should use multiple regression right? as in my framework have 3 IVs, 1 moderated and 1 DV
From what I can understand this would be suitable. You would use SEM if you had a set of measures describing your latent variable(s) , say using 5 likert scales reflective of your single underlying DV. Hope that helps. Paul
- Badges
- Science topic
- Similar topics
- Statistics
- Survey