Hi,

I am trying to calculate measurement invariance for my questionaire. I have two groups (1=1 and 2=2). I do have good model fits for configural invariance and metric invariance. But I also get a negative chi square difference test. I used this script for the Mplus Input https://www.pedocs.de/volltexte/2015/11380/pdf/ESP_2015_3_Schwab_Helm_Ueberpruefung_von_Messvarianz.pdf

I have 12 scales, most of them with 3 items, 3 scales have only 2 items. Because I have so many scales it might be better to decribe only the input for the first scales (but I have also specified the other scales in the input in the same way).

This is my input for configural invariance:

MODEL:

!loadings

WE BY

S11_we11@1 (1)

S11_we1* (2)

S11_we10* (3) ;

!intercepts

[S11_we11*] (I1); [S11_we1*] (I2); [S11_we10*] (I3);

!residuals

S11_we11* (E1); S11_we1* (E2); S11_we10* (E3);

!factor

WE*;

!factor

[WE@0];

Model 2:

!loadings

WE BY

S11_we11@1

S11_we1*

S11_we10* ;

!intercepts

[S11_we11*];

[S11_we1*] ;

[S11_we10*] ;

!residuals

S11_we11* ; S11_we1* ; S11_we10* ;

!factor

WE*;

!factor

[WE@0];

OUTPUT:

tech1;

tech4;

stdyx;

sampstat;

MODINDICES STANDARDIZED;

AND for metric invariance:

Model:

!loadings

WE BY

S11_we11* (1)

S11_we1* (2)

S11_we10* (3) ;

!intercepts

[S11_we11*] (I1); [S11_we1*] (I2); [S11_we10*] (I3);

!residuals

S11_we11* (E1); S11_we1* (E2); S11_we10* (E3);

!factor

WE@1;

!factor

[WE@0];

Model 2:

!loadings

WE BY

S11_we11* (1)

S11_we1* (2)

S11_we10* (3) ;

!intercepts

[S11_we11*];

[S11_we1*] ;

[S11_we10*] ;

!residuals

S11_we11* ; S11_we1* ; S11_we10* ;

!factor

WE*;

!factor

[WE@0];

I would appreciate some help very much! How do I have to adjust the input or what do I have to calculate additionally?

Using the Model Test in Mplus only led to the model not being identified.

I am very grateful for any help/ideas!