I recently ran some analyses using Mplus that I previously ran using LISREL. I notice that the AIC statistic reported by Mplus is quite a bit larger than what I got with LISREL. Any idea why this may be the case? I'm using Mplus version 3.01 Thanks.
There are different ways of computing AIC. The formula we use is shown in the Technical Appendices. This topic is discussed by Adrian Raftery in Sociological Methodology.
QianLi Xue posted on Tuesday, November 13, 2007 - 12:42 pm
Can MPLUS generate information criteria (e.g. AIC, BIC) for a saturated model. I assume that we can calcuate based on the loglikelihood for H1, but students would like to see them in the output. Thanks for any help you may provide.