forum.smartpls.com

Hi all, In my interaction analysis in smart pls, the beta value of the interaction term is negative (should be positive as per the model assumption), though the beta values of the moderator and the independent variable are positive. The t-value for all the 3 variables are significant and the r-squared value of the dependent variable also increases due to the moderating effect. Should we consider the negative beta while interpreting moderating effect or can we stick with change in r-squared value and ignore beta value? Please clarify.

Hi,

You have collinearity problem, check it.

Regards,

Thanks. Will do and get back.

Hi
I did check for collinearity problems using SPSS and the VIF values for all my variables are above 1 and less than 5.

Hi,

If no have collinearity,this is supression effect. So, the sign change after you include moderation effect. For clear this issue may be you can asses Sympson Paradox Ratio.

Best regards,

will check out and get back. Thanks

i also think something is wrong with collinearity