Message/Author 

Yao Wen posted on Monday, May 02, 2011  5:43 pm



Hi, I am doing a two level SEM model study. The input data is a covariance matrix. I didn't find the example for this kind of input. So I am not quite sure how to set up the sample size. Are there two sample sizes that one is for group size, the other one is for individual size? Or, is only one sample size needed which indicates the average sample size? Thank you! 


See Example 13.1 and pages 431 and 432 of the Version 6 user's guide. 

Yao Wen posted on Wednesday, May 04, 2011  9:31 pm



Thank you for the reply! But in my case, I only have one level2 covariance matrix as input to obtain the estimates of parameters for both levels. I am not sure if it is possible to do this in Mplus. 


This will not work. You need raw data. 


Hello, I'm trying to apply a 2level regression analysis with about n=200 students in k=9 classes. I'm only interested in effects on level 1. From reading papers by Hox and Maas, I had got the impression that the small number of classes does not seriously affect the estimated parameters on level 1. Is that true? In other words, would a random intercept model that considers only level1 predictors be reasonable for my dataset? Thank you! 


This sounds plausible. You could compare this to a Bayesian analysis where we have found fewer cluster units are needed. 


Thank you for your quick response. Analyses with ML and Bayes estimation yielded very similar results. Could you please give me a hint to the reference where the Bayesian estimation is reported? 


See the following which is available on the website: Asparouhov, T. & Muthén, B. (2010). Bayesian analysis of latent variable models using Mplus. Technical Report. Version 4. See also, Browne, W.J. & Draper, D. (2006). A comparison of Bayesian and likelihoodbased methods for fitting multilevel models. Bayesian Analysis, 3, 473514. 

Back to top 