I am doing CFA to test for common method bias using an unmeasured latent construct method. The average common variance I am getting from the common factor is about 31% which is high (as should be less than 25%).

I checked for differences in p-values of the trait loadings (between CFA model with ULC and without ULC) and p-values become insignificant for only one scale.

What can I do to control the effect of common method variance statistically in this case?

I am doing a moderation analysis and I have 6 variables

  • 1
    $\begingroup$ Try looking into the work of Terence (T. J. G.) Tracey for info on a method of partialing out the first unrotated factor as a "general factor" attributable to social desirability or common method bias. $\endgroup$ – rolando2 Oct 14 '12 at 22:36

You might also wanted to try using a marker variable (Williams, Hartman, & Cavazotte (2010). Method Variance and Marker Variables: A Review and Comprehensive Marker Technique. the Organizational Research Methods) which is simply explained in this youtube. http://www.youtube.com/watch?v=w7zZCBlRXog&lc=5D_7EzRSjHAjwrJJdM0epb8dEiSNe_zaTs0AopwzZ1s&lch=email_reply&feature=em-comment_reply_received


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.