I am carrying out a mediation analysis using PROCESS in spss. According to Field a mediation effect occurs if the confidence intervals for the indirect effect do not include 0 - this is the case but the predictor to mediator pathway is non significant, although there is a sign effect for predictor to outcome, and sign effect mediator to outcome.

Does this still mean there is a mediator effect?

  • $\begingroup$ Who is "Field" and where is this said? $\endgroup$
    – Glen_b
    Commented Jun 17, 2013 at 9:21
  • 1
    $\begingroup$ The old style of looking at mediation (and moderation) viewed them as either present or not. The seminal article is by Baron & Kenny. More recent work views mediation as a continuum, from non-existent to weak to strong (see work by e.g. MacKinnon) $\endgroup$
    – Peter Flom
    Commented Jun 17, 2013 at 10:53
  • $\begingroup$ Field is Andy Field, lecturer and writer of statistics books, and he says this on a you tube video $\endgroup$ Commented Jun 17, 2013 at 16:46
  • $\begingroup$ There is no command or menu item called PROCESS in SPSS, so I am wondering what you are actually doing. $\endgroup$
    – JKP
    Commented Jun 17, 2013 at 17:12
  • $\begingroup$ I am using a macro by Hayes for moderation and mediation analysis which he calls PROCESS $\endgroup$ Commented Jun 21, 2013 at 12:52

1 Answer 1


If your a path (from predictor to mediator) is not significant then you DO NOT have a mediation effect. The proposed mediator has an effect on the outcome variable and your proposed predictor also has an effect on the outcome variable, just NOT by the mediator you proposed. And yes, PROCESS is a plug-in that you can install in SPSS has the most current methodology for examining mediating effects, so that is the best tool you can use at the moment.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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