Help with AMOS/SEM

[roshayu mohamad]

 

Hi, I am doing SEM and analysing my data using AMOS. Having some problem here. Grateful if any of you have any idea how I can fix this.

 

I have 7 factors, and evaluating individual structural model for each factors shows a good fit for all (with some modification). But when it comes to the overall structural model, the fit only close to good fit (GFI, TLI, CFI  0.8XX; RMSEA is good).  How can i fix this, and how can I say that my overall structural model is fit with this ‘bad’ figure?

 

My data is close to 350 with 37 observed variables.

 

Thanks.

[Jeremy Miles]

First, that’s a lot of variables. And these models rarely fit when you have a lot of variables.

 

You have two choices.  You can say that your model doesn’t fit –  and 0.8 is really not fitting (what’s RMSEA?) and stop there.  Or you can try to tweak your model, you can add correlated errors, or you can remove variables.  Look at the modification indices and residuals for clues.  However, when you do this you’re moving into exploratory mode, and it might not be long before you need to consider exploratory factor analysis instead.

 

jeremy

[roshayu mohamad]

Hi Jeremy,

 

I have tried doing 2 things with my model:

 

1) add correlated errors – the overall structural model does fit. even though the error correlation does make sense theoretically i believe, but somehow i feel like the model look ridiculous.

 

2) i removed the variables which cause the problem, but this time, my overall structural model become unidentified.

 

when you say EFA, what are those signs that make it ‘a must ‘ for EFA?

 

 

 

[Author]

Posts