I am running a path analysis and have noticed that there are two different ways of running the analysis, one where means and variances are specified in the 'Model' command (e.g., var1; var2;[var1];[var2]; var2 ON var1;) and one in which they are not (e.g., simply var2 ON var1). As far as I can tell, the outputs seem to be identical but is there any difference between these two specifications?
The means, variances, and covariances of observed exogenous variables should not be included in the MODEL command. A regression model is estimated conditioned on these variables. With no missing data, continuous outcomes, and maximum likelihood estimation, the results are the same when the means, variances, and covariances of observed exogenous variables are included in MODEL command. In other cases, this is not the case. When they are included, unnecessary distributional assumptions are made about them.