Message/Author |
|
|
Dear Dr. Muthen, I have a scale with 15 items with a binary response format. I ran an invariance analyses for categorical items, examining configural, metric and scalar invariance. I received feedback that the way I identified my scalar model was overly restrictive because I fixed my item residuals to = 1. And item residual variances were fixed to 1 in my configural and metric invariance models. Below is my code for my scalar invariance model. Can you please advise. !Scalar MODEL FOR Women Reference GROUP MODEL: !!! Factor loadings all estimated f1 BY v1*(L1) v2*(L2) v3*(L3) v4*(L4) v5*(L5) v6*(L6) v7*(L7) v8*(L8) v9*(L9) v10*(L10) v11*(L11) v12*(L12) v13*(L13) v14*(L14) v15*(L15); !!! Item thresholds all free [v1$1-v15$1*]; ! Item residual variances all fixed=1 v1-v15@1; ! Factor mean=0 and variance=1 for identification [f1@0]; f1@1; |
|
|
Continued from above !!! SCALAR MODEL FOR Men FOCAL GROUP MODEL men: !!! Factor loadings all estimated f1 BY v1*(L1) v2*(L2) v3*(L3) v4*(L4) v5*(L5) v6*(L6) v7*(L7) v8*(L8) v9*(L9) v10*(L10) v11*(L11) v12*(L12) v13*(L13) v14*(L14) v15*(L15); !!! Item thresholds all free ! Item residual variances all fixed=1 v1-v15@1; ! Factor mean=now FREE and variance=now FREE [f1*]; f1*; |
|
|
The models we suggest for measurement invariance are described in the Version 7.1 Language Addendum on the website with the user's guide. You can check your input with this. Please limit posts to one Window. |
|
Stace Swayne posted on Tuesday, September 08, 2015 - 8:58 pm
|
|
|
Dear Dr. Muthen, Would you say that the measurement invariance code provided in the Version 7.1 Language Addendum on page 7, which fixes the residuals in all models for scalar invariance is a more restrictive assumption than needed for scalar invariance? |
|
|
I don't see that we state that residual variances are fixed in all groups. |
|
Back to top |