James Swartz
Would very much appreciate help debugging a vexing problem trying to test for measurement invariance using ESEM for a 3-factor model with 20 dichotomous indicators. I am trying to walk through the 13 sets of comparisons as described by Marsh et al. but am stuck on step 1, which tests for configural invariance. I have shamelessly adapted the code from a paper by Toth et al. where they have continuous indicators. Here is abbreviated code with reduced example item set:
CATEGORICAL ARE item_1-item_6;
GROUPING = PARTICIP(0 = FR, 1 = BYS);
Analysis:
estimator = wlsmv;
rotation = target;
parameterization = theta;
Model:
f1 by item_1 item_2 item_3~0 item_4~0 item_5~0 item_6~0 (*1);
f2 by item_1~0 item_2~0 item_3 item_4 item_5~0 item_6~0 (*1);
f3 by item_1~0 item_2~0 item_3~0 item_4~ item_5 item_6 (*1);
[item_1-item_6];
item_1-item_6;
[f1-f3@0];
item_1 with item_2 (c1);
item_3 with item_4 (c2);
item_5 with item_6 (c3);
Model BYS:
f1 by item_1 item_2 item_3~0 item_4~0 item_5~0 item_6~0 (*1);
f2 by item_1~0 item_2~0 item_3 item_4 item_5~0 item_6~0 (*1);
f3 by item_1~0 item_2~0 item_3~0 item_4~ item_5 item_6 (*1);
[item_1-item_6];
item_1-item_6;
[f1-f3@0];
The error returned is this:
THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES COULD NOT BE COMPUTED. THE MODEL MAY NOT BE IDENTIFIED. CHECK YOUR MODEL. PROBLEM INVOLVING PARAMETER 174.
Inspection of the parameters in TECH 1 indicates parameter 174 is the variance of item_1. I suspect that this code, while it works for continuous indicators, does not work for dichotomous/categorical because the means and variances of categorical variables are not independent. But I am not sure what line(s) of code to change to have this run correctly and still test for configural invariance. Any help would be much appreciated. Been stuck for days!
James A Swartz
This link might be useful: http://www.statmodel.com/discussion/messages/11/5581.html?1589927864
http://www.statmodel.com/discussion/messages/11/5581.html?1589927864
Thanks Everyone - I have looked at that discussion thread and it does not quite answer the issues or provide an example. I have categorical indicators and am trying to do ESEM within CFA... but maybe if I do a pure ESEM approach, these examples will work.
Thanks again.