I have tested a simple mediation model in which motivation at 4 week follow up mediates the effect of treatment assignment on treatment outcome at 12 weeks using the model indirect effect command. While I understand the results, I am unclear as to how to explain the procedures in the methods section and report the findings in the results section. With regard to the methods section, is there a reference that can be used for this procedures (aside from the mplus manual)? I know I should report the estimates, standard error and estimate/standard error. The manual refers to the delta method but what statistic is being reported. I assume the estimate/standard error is similar to a z score and therefore, +/- 1.96 is significant. Is this assumption correct?
Check for references by MacKinnon. There is one for confidence intervals in our user's guide reference list that may address this. Mplus implements indirect effects that are the product of the two regression coefficients, that is, a times b. The Delta method is the method for calculating standard errors of the indirect effects. The ratio is a z score.