Saturated Model

[DUYAN]

Dear all;
I would be very grateful if you can explain me why the results obtained from saturated model are different from estimation model?
for example, in the saturated model, i have a link which is significant at 5% (path coefficient=1,77) while in the estimation model the same link is significant at 10% (path coefficient = 0,098).
I have 194 observations
I followed tow steps approach for analysis ( in the second step i used the latent variable scores).
Thanks in advance

[plshelpme]

Hi Duyan,

I am new to this, so if I am wrong somebody please correct me.

If I remember correctly, from Hair's book 'A primer on PLS...' the saturated model is a version of the structural model where all relationships are allowed to connect freely. It is used to obtain an estimation of the model fit – as you compare the saturated model with your estimated model. That key difference is probably why you have different values.

Have you assessed your model fit?

[DUYAN]

Hi Duyan,

I am new to this, so if I am wrong somebody please correct me.

If I remember correctly, from Hair's book 'A primer on PLS...' the saturated model is a version of the structural model where all relationships are allowed to connect freely. It is used to obtain an estimation of the model fit – as you compare the saturated model with your estimated model. That key difference is probably why you have different values.

Have you assessed your model fit?

thank you for your response
I have a SRMR value of 0,064 and GoF of 0,541
I saturated the model to know if there is a significant path which is not taken into account in the original model.....

[jmbecker]

The differentiation between saturated model and estimated model is only made for the model fit indices that operate on the complete covariance matrix. I don't understand how you can get different results for the path coefficients?
Did you saturate your model yourself? If yes, it is not suprising that you get different results as you have a different model. With a fully saturated model you might run into multicolllinearity problems, etc.