0
$\begingroup$

I ran 2 mediation models each with two mediators in R (lavaan) and am looking for help with interpretation.

Model 1:

  • Paths a1 and a2 = significant
  • Paths b1 and b2 = not significant
  • Path c' = significant (direct effect)
  • Path c = significant (total effect)

I want to know what I can conclude about the relation between X1 and Y. Can I argue that there is both an effect of X1 on Y (direct) as well as an additive effect (through total effect = direct + indirect)?

Model 2:

  • Paths a1 and a2 = significant
  • Paths b1 and b2 = not significant
  • Path c' = not significant (direct effect)
  • Path c = significant (total effect)

I want to know what I can conclude about the relation between X2 and Y. Can I argue that there is an effect of X on Y when you consider the additive effects of the direct and indirect paths (i.e., the total effect)?

Theoretical mediation model as figure

Thank you!

Improve this question
$\endgroup$
5
  • 1
    $\begingroup$ Most people won't be familiar with what a1, b2, etc mean. Can you rephrase for a more general audience? $\endgroup$
    – Jeremy Miles
    Jul 14, 2020 at 15:25
  • 1
    $\begingroup$ I also removed the R and lavaan tags, as these are not relevant. $\endgroup$
    – Jeremy Miles
    Jul 14, 2020 at 15:26
  • 1
    $\begingroup$ I'm not clear why the second model was run. $\endgroup$
    – Michelle
    Jul 15, 2020 at 1:32
  • $\begingroup$ @Michelle Model 2 used a different predictor (X2), all other variables were the same as in Model 1 $\endgroup$
    – Sybil
    Jul 15, 2020 at 20:34
  • $\begingroup$ X is significantly related to M1 and M2, but M1 and M2 are not significantly related to Y. X is also significantly related to Y. For the total effect model, c doesn't include M1 or M2 and thus doesn't measure them. If you just want to do X against Y directly, you could/should use a regression. SEM isn't useful when you only have the one path. $\endgroup$
    – Michelle
    Jul 15, 2020 at 21:03

1 Answer 1

Reset to default
1
$\begingroup$

Model 1. I want to know what I can conclude about the relation between X1 and Y. Can I argue that there is both an effect of X1 on Y (direct) as well as an additive effect (through total effect = direct + indirect)?

  • You can state that X1 is significantly associated with Y. However, you cannot make claims about the indirect effect until you statistically test the indirect effects (not the individual a1, b1, a2, b2 paths from which the indirect effects are comprised). You can do this by adding these codes to your estimation model: ab1 := a1b1 ab2 := a2b2

Model 2. I want to know what I can conclude about the relation between X2 and Y. Can I argue that there is an effect of X on Y when you consider the additive effects of the direct and indirect paths (i.e., the total effect)?

  • The same deal applies here. X is significantly associated with Y. You cannot make claims about the indirect effects until you actually test them.
Improve this answer
$\endgroup$

Your Answer

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

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