Hi Mehwish,

multicolinearity is an issue of two or more predictors in ONE structural or regression equation (hence, one dependent variable).

If you mean (perhaps you can clarify) that two dependent variables are very highly correlated, the coefficients of the predictors will be similar but the standard errors will (as far as I imagine) not be affected (this is the core issue in multicolinarity).

HTH

Holger

Dear Holger,

Thank You for your reply.

You mentioned that "multicolinearity is an issue of two or more predictors". In my case if i don't consider the mediating variable and take the case e.g. Quality (IV) --> Performance (2 dimensional formative construct). Then how i should check the multicollinearity of IV with a formative DV.

Mehwish

Hi Mehwish,

as I said AFAIK there cannot be a to high relationship between an IV and a DV. Of course, parameter estimates have to make sense (and a too high effect would be suspicious) but the standard errors won't be affected (too the contrary, a higher Rsquare reduces the standard errors).

But I can imagine that I don't get your point. Perhaps, drawing a path diagram would help....

Best,

Holger

Dear Mehwish,

As Holger said, multicolllinearity only occurs if you have more than one independent variable in your equation. Multicollinearity is caused by collinearity among your independent variables not among the dependent variables. So if there is only one independent variable in your equation, multicollinearity is no issue. Thus from your description, multicollinearity might only be an issue for the coefficient estimates of your formative indicators in the formative measurement equation.

Best regards,

Florian

مرحبا عزيزي