|
|
MAR and Type=RANDOM with AUXILIARY |
|
Message/Author |
|
|
Hi all, I have tried to track down other discussions of this topic and perhaps overlooked something, but couldn't find anything that completely answers my questions. I am using TYPE=RANDOM in order to use TSCORES for analysis of a cohort-sequential design. I would like to also use the AUXILIARY command with the '(m)' option for missing data correlates, but it appears that I cannot: "The 'm' specifier in the AUXILIARY option is not available for TYPE=RANDOM." Is there any way to overcome this limitation, e.g., with a newer version of Mplus? Or is there some fundamental analytic limitation here? Thank you for any advice. Very best, John |
|
|
Try to capture the different cohorts using multiple-group analysis instead. See UGT ex 6.18. |
|
|
I was looking at an example of a linear growth model with missing data on a continuous outcome using a missing data correlate to improve the plausibility of MAR in this link: http://www.statmodel.com/usersguide/chap11/ex11.1.html I was wondering what are the hypothesis tested in relation to the auxiliary variables. In other words, how do I interpret the output regarding the auxiliary part of the model? MODEL FIT INFORMATION Number of Free Parameters 9 Loglikelihood Including the Auxiliary Part H0 Value -1009.197 H1 Value -1007.232 Information Criteria Including the Auxiliary Part Number of Free Parameters 15 Akaike (AIC) 2048.394 Bayesian (BIC) 2097.869 Sample-Size Adjusted BIC 2050.347 (n* = (n + 2) / 24) Chi-Square Test of Model Fit Value 3.930 Degrees of Freedom 5 P-Value 0.5595 Thanks! |
|
|
The auxiliary part is saturated so does not impact the test of model fit. See the tech report Auxiliary Variables Predicting Missing Data on our website at http://www.statmodel.com/techappen.shtml |
|
Back to top |
|
|