Anonymous posted on Wednesday, July 20, 2005 - 11:45 pm
A reviewer asked whether the models I ran in M-Plus (regular SEM mediation models) were full information maximum likelihood (FIML) models or not. The reviewer said FIML models "yield consistent estimates of structural parameters by correcting for simultaneity between the endogenous variables and the disturbance terms of the statistical model."
Does M-Plus generally do this as an estimation method or is there a way I can write the code to make it do it?
Thank you for your help.
bmuthen posted on Thursday, July 21, 2005 - 12:24 am
Mplus uses FIML as the default in most cases (check the output for the estimator used). The reviewer quote you give is strange if it is a general statement about SEM mediational modeling estimated under FIML. It is not relevant to all SEM mediation models estimated by FIML - such as those models that do not have/need simultaneity (reciprocity). But if your theory calls for simultaneity, Mplus can do it.
Anonymous posted on Thursday, July 21, 2005 - 6:54 pm