|
|
Comparing bifactor models embedded wi... |
|
Message/Author |
|
LSH posted on Friday, August 25, 2017 - 1:48 pm
|
|
|
I'm trying to test two competing SEMs. Both contain a bifactor model predicting a latent variable. I would like to assess whether the model fit is improved by allowing the specific factors to predict in addition to the general factor. Model: EmoReg BY D1, D3-D5, D7, D9, D11-D16, D18-D33, D35-D36; NON BY D25 D21 D12 D11 D29 D23; GOAL BY D26 D18 D13 D33 D20; IMPULSE BY D32 D27 D14 D19 D3 D24; STRAT BY D16 D15 D31 D35 D28 D22 D36 D30; CLAR BY D5 D4 D9 D7 D1; EmoReg WITH STRAT@0 NON@0 GOAL@0 IMPULSE@0 CLAR@0; STRAT WITH NON@0 GOAL@0 IMPULSE@0 CLAR@0; NON WITH GOAL@0 IMPULSE@0 CLAR@0; GOAL WITH IMPULSE@0 CLAR@0; IMPULSE WITH CLAR@0; DASS BY str anx dep; DASS ON EmoReg; DASS WITH STRAT@0 NON@0 GOAL@0 IMPULSE@0 CLAR@0; The specific factor syntax is identical except the last line is: DASS ON STRAT NON GOAL IMPULSE CLAR. Is this correct? Thank you! |
|
|
I would change the DASS WITH statement to DASS ON (and fix all the slopes to 0). It may not make a difference but it is more logical. |
|
LSH posted on Sunday, August 27, 2017 - 6:40 am
|
|
|
Thank you very much, I appreciate the help! |
|
LSH posted on Sunday, August 27, 2017 - 8:32 am
|
|
|
One follow-up question: Am I right in understanding that these models are nested? Thank you again! |
|
LSH posted on Sunday, August 27, 2017 - 10:49 am
|
|
|
Update: please disregard my previous post; I believe I have figured it out. Thanks again for your help. |
|
Back to top |
|
|