Say the model I am working with is simply:

M <- IV
DV <- IV + M

All three constructs are obtained using the same survey, hence the common method bias issue. I thought about allowing the error terms of all three correlate, but then SEM can't compute standard errors.

Is there another way to counter this common method bias using lavaan?

  • $\begingroup$ Questions about lavaan syntax are off topic here and will be closed. I have edited the question, because it is more broad than just lavaan. $\endgroup$ – Jeremy Miles Jan 11 at 18:17
  • $\begingroup$ And I think the answer is no. $\endgroup$ – Jeremy Miles Jan 11 at 18:17
  • $\begingroup$ Well, I know one way conceptually for SEM to counter this is to add a common latent factor (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003; Williams, Edwards, & Vandenberg 2003). But I am not sure if lavaan can do this... as it is unobserved variable/I have no data for it. $\endgroup$ – Carrie Jan 14 at 6:47
  • $\begingroup$ You can add it, but the model won't be identified. $\endgroup$ – Jeremy Miles Jan 14 at 16:59
  • $\begingroup$ ... Unless M, IV and DV are latent and have at least three (?) indicators each. This is true of every SEM program. $\endgroup$ – Jeremy Miles Jan 14 at 17:00

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.