I need to establish the impact of one variable (x[predictor-Independent]) on another variable (y[Outcome-dependent]), having a third variable playing a moderating role. This is premised on the assumption that there is an interaction between the predictor and the moderator variables with respect to their effect on the outcome variable.
let z be the third variable then form the regression y=a + bx +c(x*z) + d*z +eps run this regression. Test H0 c=0. If you can reject H0 then there is an interaction between x and z.and so z moderates x. See this link for details:
https://www.bing.com/search?q=interaction+and+moderation&qs=n&form=QBRE&sp=-1&pq=interaction+and+moderation&sc=6-26&sk=&cvid=C69D762C4C284B4C9F6287CD22979C6F
nice plots and so on can be done to check these things. See interaction in Montgomery, Design and Analysis of Experiments. Best, David Booth
another useful link is:
http://core.ecu.edu/psyc/wuenschk/MV/multReg/moderator.pdf
BTW please notice this is trivial in R. Just use the lm function.
In addition, CBSEM and PLSSEM with a moderating variable may be used depending on how variables are measured.
- Badges
- Science method
- Similar topics
- Methodology