In structural equation modeling analyses on psychological constructs, sometimes we encounter perfect fits with .00 RMSEA etc. I know that this is considered as a misleading finding. I would appreciate opinions and references on this issue.
Dear Çinar Kaya,
As David A. Kenny indicates, the fit refers to the ability of a model to reproduce the data (i.e., usually the variance-covariance matrix). Of course, one expects reasonable fit indexes, but, in practice, it is often impossible obtain perfect coefficients. When you observe a RMSEA value of 0.000, perhaps the most recommended is to check the model specification. In the case of this index, it is recommended to present the 90% confidence interval (90%CI). Moreover, it is suggested in the literature taking into account more than one fit index. For a brief introdution to the fit indexes, perhaps it is interesting to visit the Professor Kenny's homepage (http://davidakenny.net/cm/fit.htm), where you can finding interesting references about this issue. I hope this information can be useful for you.
Regards, Valdiney
Dear Çinar Kaya,
As David A. Kenny indicates, the fit refers to the ability of a model to reproduce the data (i.e., usually the variance-covariance matrix). Of course, one expects reasonable fit indexes, but, in practice, it is often impossible obtain perfect coefficients. When you observe a RMSEA value of 0.000, perhaps the most recommended is to check the model specification. In the case of this index, it is recommended to present the 90% confidence interval (90%CI). Moreover, it is suggested in the literature taking into account more than one fit index. For a brief introdution to the fit indexes, perhaps it is interesting to visit the Professor Kenny's homepage (http://davidakenny.net/cm/fit.htm), where you can finding interesting references about this issue. I hope this information can be useful for you.
Regards, Valdiney
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics