LCA with shared indicators
Message/Author
 Troy Thomas posted on Monday, February 17, 2014 - 1:31 am
I am trying to estimate a latent class model with two latent variables LV1 and LV2 which have categorical indicators (5-point rating scales). LV1 has four indicators V1 -V4 and LV2 has 14 indicators V1-V14.

classes = ca(2) C(3);
Analysis:
Type=mixture;
parameterization=LOGLINEAR;
MODEL:
%overall%
ca with c@0;

MODEL ca:
%ca#1%
[V1\$1-V4\$1];
[V1\$2-V4\$2];
[V1\$3-V4\$3];
[V1\$4-V4\$4];

%ca#2%
[V1\$1-V4\$1];
[V1\$2-V4\$2];
[V1\$3-V4\$3];
[V1\$4-V4\$4];

MODEL C:
%C#1%
[V5\$1-V14\$1](1);
[V5\$2-V14\$2](2);
[V5\$3-V14\$3](3);
[V5\$4-V14\$4](4);

[V1\$1-V4\$1];
[V1\$2-V4\$2];
[V1\$3-V4\$3];
[V1\$4-V4\$4];

%C#2%
[V5\$1-V14\$1](5);
[V5\$2-V14\$2](6);
[V5\$3-V14\$3](7);
[V5\$4-V14\$4](8);

[V1\$1-V4\$1];
[V1\$2-V4\$2];
[V1\$3-V4\$3];
[V1\$4-V4\$4];

This gives the following error:

ONE OR MORE MULTINOMIAL LOGIT PARAMETERS WERE FIXED TO AVOID SINGULARITY OF THE INFORMATION MATRIX. ... THE FOLLOWING PARAMETERS WERE FIXED:41

Parameter 41 is the ALPHA(c) value for CA#1 (LV1)

I may be approaching the model inappropriately. Thanks for any help that you can provide for this problem.
 Linda K. Muthen posted on Monday, February 17, 2014 - 2:04 pm