I am facing problem in getting the good fit indices. I have only 400 respondents with a total of 4 variables and 54 items (A: 5 components, 3-4 items each; B: 5 components, 4-5 items each; C: 3 components, 3 items each, DV=D: 3 items). The initial values for fitness indices were:
I repeated the same steps by running CFA for individual constructs, deleting items according to MI, running CFA again for whole model, and deleting items according to MI. I've deleted 13 items from the model.
ChiSq P-value = .000; RMSEA = .068; GFI = .783; CFI = .875; TLI = .864; NFI = .812; ChiSq/df = 2.853
Read this chapter of the book. You will get it clearer. Good luck.
Follow the following proposed fit indices threshold. I will personally suggest to address the issue through the specification of an error co-variance by adding a co-variance between or among items error terms.
GFI>0.90 indicates good fit (Kline 2005).
TLI values exceeding 0.95 indicate good model fit (Hu & Bentler 1998).
CFI values greater than roughly 0.90 may indicate a reasonably good fit of the model. A value of 0.95 is preferable (Kline 2005).
When you have many items, GFI, CFI, TLI, and NFI are not good although RMSEA is acceptable (lower than .05 or .08).
I suggest you to use an item-parceling approach (e.g., use 3 items for each component or use component as an indicator). For details, refer to the following articles.
Beauducel, A., & Wittmann, W. Simulation study on fit indices in confirmatory factor analysis based on data with slightly distorted simple structure. Structural Equation Modeling, 2005, 12, 41-75.
Hair, J. F. J., Anderson, R. E., Tatham, R. L., & Black,W. C. (1995). Multivariate data Little, T. D., Cunningham, W. A., Shahar, G., & Widaman, K. F. (2002). To parcel or not to parcel: Exploring the question, weighing the merits. Structural Equation Modeling, 9, 151–173.analysis (4th ed.). Saddle River, NJ: Prentice Hall.
Thank you for your suggestion.... I try adjust my model again.
ChiSq P-value = .000; RMSEA = .063; GFI = .838; CFI = .914; TLI = .904; NFI = .868; ChiSq/df = 2.589
But..I deleted 20 items (from 54 items) in my model.
Would it be acceptable for model fit????
The CFA can be considered valid if the deletion items are not higher than 20% of the total items. You can refer the book written by Kline (2005).
Hi,
after you modified your model and got the following indeces, I think the model fit.
ChiSq P-value = .000; RMSEA = .063; GFI = .838; CFI = .914; TLI = .904; NFI = .868; ChiSq/df = 2.589
Good luck
Please, I want to interpret warpPLS result. What are the important result to report and their thresholds. Thank you
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling