|
|
BSEM MI analysis for categorical data |
|
Message/Author |
|
|
Hi Mplus veterans, I want to use Bayesian method to test full measurement invaraince (scalar MI) for categorical data. There are 8 items in 30 different groups. I fixed the factor variance at 1 for all groups. Also, I fixed the factor mean at 0 for the reference group (%c#1%). Is my method correct? Any comments are greatly appreciated. Thank you in advance! SW VARIABLE: NAMES = id group y1-y8; USEVARIABLES = group y1-y8; CATEGORICAL = y1-y8; CLASSES = c(30); KNOWNCLASS = c(group = 1-30); ANALYSIS: ESTIMATOR = BAYES; PROCESSORS = 2; BSEED = 48932; BCONVERGENCE = .05; TYPE = MIXTURE; Biterations = 50000000(1000); MODEL: %OVERALL% f by y1*y2-y8; !factor variance is fixed at 1 for all groups. f@1; %c#1% !factor mean is fixed at 0 for the reference group. [f@0]; OUTPUT: TECH1 TECH8; |
|
|
I would fix the factor variance to 1 in only one group. |
|
Back to top |
|
|