My model fit indices as mentioned above for the sample size of 600 is more than 0.8 but not meeting the rule of thumb of more than 0.9 value can this value be considered and what are the ways to improve the model fit values
Values below .90 for CFI/TLI tend to indicate poor fit. There are general problems with all those indices. My view is that it is best to rely on the chi-square test of model fit and to look at covariance residuals and modification indices to examine sources of model misfit.
Model fit indices provide valuable information about how well your statistical model fits the observed data. While the rule of thumb suggests values above 0.90 for certain fit indices, the interpretation of fit indices should not be solely based on these cutoffs. The choice of fit indices and their acceptable values can depend on the complexity of the model, the nature of the data, and the research context.
Here are some key points to consider:
It sounds like you are referring to fit indices used to assess the goodness-of-fit of a statistical model, often in the context of structural equation modeling (SEM) or similar techniques. Fit indices provide information about how well the proposed model fits the observed data. While there are recommended guidelines for fit index values, the decision of whether a value slightly below the threshold can be considered acceptable depends on the specific context and the overall research goals.
Typically, a fit index value greater than 0.90 is considered good, while values greater than 0.80 might be considered acceptable, especially if you have theoretical or practical justifications for why the model might not achieve a higher fit. The decision to accept or reject a model based on fit indices should be made in combination with other factors, such as theoretical reasoning, model complexity, and practical implications.
If your fit indices are not meeting the desired thresholds, here are some strategies to consider for improving model fit:
Remember that improving model fit should be guided by theoretical reasoning and prior research. While achieving high fit indices is desirable, the ultimate goal is to have a model that accurately represents the theoretical framework and provides meaningful insights into the relationships among variables.
Thank you for this question. I often had this problem. I am learning enormously from the comments already made on this question. Following these, I can modestly say that the Model fit indices examine the fit between the model and the data. CFI and TLI indices below 0.95 indicate a mismatch between the model and the data. The analysis of the modification indices indicates the source of the problem and makes it possible to correct the model (as stated by those who commented before me). Similarly, adjustment problems can be related to theoretical relationships between model elements. I could be wrong, but I think that a model or a modeling must be based on theoretically valid relations before being tested statistically. Some problems with model fitting would arise from the fact that the model variables are not logically related. this is a basic theoretical problem that can adversely affect model fit by showing poor fit indices. See Hu and Bentler (1999) for model fit criteria.
- Badges
- Science topic
- Similar topics
- Instrumentation Engineering
- Instruments