Lars Hauge posted on Tuesday, June 30, 2009 - 4:59 am
Dear Dr. Muthen,
Is it possible in Mplus to save pooled-within asymptotic variance-covariance matrices in the same way as for variance-covariance matrices? [cf. Dyer, N. G., Hanges, P. J., & Hall, R. J. (2005). Applying multilevel confirmatory factor analysis techniques to the study of leadership. Leadership Quarterly, 16(1), 149-167.] Or will raw-data always have to be used to estimate models applying MLR?
You can save this matrix using the SAMPLE option of the SAVEDATA command. See the user's guide for further information. You can analyze it using ML. Raw data are required with MLR.
Lars Hauge posted on Wednesday, July 01, 2009 - 12:57 am
Dear Dr. Muthen,
Thank you for replying. I am able to obtain pooled-within and between variance-covariance matrices applying the following syntax command:
SAVEDATA: SIGB IS D:\Between_cov.dat; SAMPLE IS D:\Within_cov.dat;
These matrices works fine applying ML estimation. However, I would like to be able also to obtain the corresponding asymptotic variance-covariance matrices to be exported and analyzed for illustrative modelling comparison reasons. I know it is possible to obtain ordinary asymptotic variance-covariance in PRELIS to be analyzed with MLR in LISREL. Is it possible to save this corresponding matrix also in Mplus?
I am also trying to obtain the pooled-within and between variance-covariance matrices for a subset of my variables, with the intention to use them for CFAs separately on within and between level (like in the article from Dyer et al 2005 mentioned above). But somehow I donīt get them but only the messages "THE MODEL ESTIMATION TERMINATED NORMALLY - THE H1 MODEL ESTIMATION DID NOT CONVERGE" and "no data saved". I already increased the number of H1 iterations but it didnīt help.