Page 1 of 1

Evaluating endogeneity in a model

Posted: Mon Apr 24, 2017 5:41 am
by Brusset_X
The study of a PLS model these days is not complete if endogeneity is not evaluated.

José Benitez, Jörg Henseler and José Roldan in this aisel.aisnet.org/cgi/viewcontent.cgi?ar ... =amcis2016paper present a methodology to do this with ADANCO. Is there a way to evaluate such a phenomenon and eventually to correct it in a model using SmartPLS?

Thank you.

Re: Evaluating endogeneity in a model

Posted: Mon Apr 24, 2017 9:00 pm
by jmbecker
We are definitely aware of this important topic, but as far as I know there is currently no PLS software (also not ADANCO) that offers a build-in feature to treat endogeneity. This has a simple reason: Endogeneity in PLS is an under-researched topic and there is no established procedure to tackle endogeneity problems.
Generally, the topic of endogeneity is certainly more complex than implementing a 2SLS approach into PLS models.

However, as PLS uses simple OLS regression on the latent variable scores in the (inner) structural model to obtain the final path coefficients, any “conventional” method to treat endogeneity in OLS regression would probably be an option for PLS as well.

I particularly like the following article on endogeneity which you may find useful as well:
Papies, D., Ebbes, P., & van Heerde, H. J. (2016). Addressing Endogeneity in Marketing Models. In P. S. H. Leeflang, J. E. Wieringa, T. H. A. Bijmolt, & K. H. Pauwels (Eds.), Advanced Methods in Modeling Markets (pp. forthcoming).  You may contact the authors.
But there are many more useful resource on tackling endogeneity, for example, using the control variable approach or the instrumental variable approach.

Re: Evaluating endogeneity in a model

Posted: Tue Apr 25, 2017 12:20 pm
by Brusset_X
I thank you ery much for this reply.
I did not find the book or the chapter that you mention by Papies, Ebbes and Heerder. I will write to them.
I tried using a control variable as instrumental in a 2SLS calculation in SPSS and got correct results.