Increase model fit
Message/Author
 Monika Hupler posted on Tuesday, July 05, 2011 - 8:51 am
Dear Mplus-Experts

How can I increase the model fit of a latent growth curve?
Now my is RMSEA=0.183

I red something that equal error covariance and error covariance can be used to increase the fit. How do I add this aspect to my Mplus syntax?

And are there other means to increase the fit?

Thank you
Monika
 Linda K. Muthen posted on Tuesday, July 05, 2011 - 4:05 pm
Look at modification indices to see where the model does not fit.
 Monika Hupler posted on Tuesday, September 13, 2011 - 7:32 am
Dear Linda,

I am sorry to ask such basic questions, I am in Southamerica at the moment and my acess to Mplus literature and help in person in limited.

Which modelfit indices are you talking about?

The Output says the folowing:
Chi-Square Test of Model Fit (Value = 488.527, df=128, p = .000)
Chi-Square Test of Model Fit for the Baseline Model (Value = 1456.747, df=120, p =.000)
CFI = 0.730, TLI = 0.747
Loglikelihood: H0 Value = -1790.674, H1 Value = -1546.411
Number of Free Parameters = 22
Akaike (AIC) = 3625.349, Bayesian (BIC)= 3674.178, Adjusted BIC = 3604.899(n* = (n + 2) / 24)
RMSEA (Estimate = 0.204, 90 Percent C.I. [0.185,0.223], p(RMSEA <= .05) = .000
SRMR (Value = 0.149)

And which model fit indices indicate, why my model fit is so low, and how do I change that?

Does the model fit only serve to get a higher power for significant results?

Thank you a lot!
 Linda K. Muthen posted on Tuesday, September 13, 2011 - 9:15 am
Ask for the MODINDICES option of the OUTPUT command. The Mplus User's Guide is available on the website. The Topic 1 course handout and video discuss modification indices. You can also Google modification indices.