Hi,
I have performed a CFA in R using the lavaan package. I used a robust estimator (MLR) because there was a lack of normality in the data. With a sample of 282 and 5 factors to conduct I have a good Chi-Square/df ratio, a good RMSEA and SRMR BUT I have a relatively low CFI=0,799 and TLI=0,788. Is it too low to report a reasonable fit? Are there any actions that I can do to bring up the CFI and TLI measure? Is there any literature that can help me in Reporting this? Thanks!! Kind regards - Manou
Dear Manou,
Please see similar question posed through ResearchGate 11 Dec. 2015 here is folowing link :
-If RMSEA, SRMR, RMR, PGFI, PNFI, and χ^2⁄df have an acceptable ...
https://www.researchgate.net/.../If_RMSEA_SRMR_RMR_PGFI_P...
Also here useful attached files in topics to help you in your question.
Best regards
Dear Manou,
Please see similar question posed through ResearchGate 11 Dec. 2015 here is folowing link :
-If RMSEA, SRMR, RMR, PGFI, PNFI, and χ^2⁄df have an acceptable ...
https://www.researchgate.net/.../If_RMSEA_SRMR_RMR_PGFI_P...
Also here useful attached files in topics to help you in your question.
Best regards
You need to see the modification indices and draw covariances between the suggested variables. Once all the modification indices are accounted for, only then you would be able to get good indices. In alternate, you need to remove low loading items from the measurement model.
Dear Manou
Some questions
1- Low correlations between your variables
2 - Factorial structure needs to be revised (theoretically)
3 - To execute the modification indexes and be careful not to discard the theoretical model.
An alternative would also be a Bayesian estimator
https://www.statmodel.com/download/IntroBayesVersion%203.pdf
Best regards,
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R Statistical Package