|
|
| Message/Author |
|
|
|
|
I have a question regarding not calculated Chi^2, CFI and TLI-parameters in Mplus5.
I have got a path model with 8 continous variables including three of them are denoted with plausible values. Therefore I use “Type = Imputation” so that all the parameters are calculated (at once) and I don’t have to calculate the model five times: for each set of plausible values separately.
My problem is that Mplus does not calculate Chi^2 of the Baseline Model (and CFI,TLI-parameters) for the model including all 8 variables using the ML-estimator.
If I omit any variable, Chi^2, CFI and TLI-parameters are calculated. If I calculate the whole model separately for each set of plausible values Chi^2, CFI and TLI-parameters will be calculated too.
Thanks! |
|
|
|
|
| CFI and TLI cannot be computed unless the baseline model converges. |
|
|
|
|
Dear Ms. Muthen,
Thank you for your quick response. But now my problem is, that I do not know how to report my results.
Can I report the standardized model results, the chi^2 test of model fit, the RMSEA and SRMR from the output of the “Type = Imputation”-output and report the CFI and TLI from the five for each set of plausible value separately calculated outputs?
Or do I have to report all results from the five for each set of plausible values separately calculated outputs and use the Rubin-formula to calculate standardized model results?
Thanks a lot for help! |
|
|
|
|
| There is no theory for CFI/TLI with multiple imputations, nor summarizing CFI/TLI over the imputations so you might just as well report all five sets for them. |
|
|
| Back to top |
|
|
