I am looking for a macro or software that can assist me with calculating sample size for my study. Specifically, I am interested in computing power for the comparative fit index. Are you aware of any such macros/software?
Dear Gargi, I do not know any software that can calculate what you want. However, I am not sure why you would like to calculate that. You have several fit index and\or goodness of fit that you should take into account. Why would you based just in the CFI? Your sample size should be based on the type of analysis that you want to performance, and not on the type of the index fit. Just to be clear, CFI it is one of the most sample size dependable index fit, and you have others fit index, that are not sample dependable as the CFI or TLI. If you could give more information about the type of research\analysis you want to do (sample, instrument, aim, theoretical models, subjects….) it would be easier to discuss the sample size. There are a few guidelines for sample sizes and index parameters related to confirmatory factor analysis or structural equation modeling. Maybe it worthwhile to check them.
If you prefer to compute power for the RMSEA index, you could it with R program. Preacher and Coffman (2006) developed an applet that you can use to calculate power and sample for RMSEA:
http://www.quantpsy.org/rmsea/rmsea.htm
I hope it could help you. I think RMSEA is a better fit index than CFI, not so biased.
You can also use a monte carlo simulation (see Muthén & Muthén, 2002).
Muthén, L. K., & Muthén, B. O. (2002). How to use a Monte Carlo study to decide on sample size and determine power. Structural Equation Modeling, 9(4), 599-620. Available at the Mplus homepage.
Dario, Kim (2005) provides a few equations that provide guidelines for calculating sample sizes for different fit indices. However, the equations can be a little intimidating. I have seen syntax for computing power for RMSEA, Steigers γ, AGFI, GFI, and McDonalds. However, the syntax I had for determining power for CFI was not performing properly. I was recently able to figure that out.
Albert, thank you. In my experience, it's easier to find syntax for computing power for RMSEA and a few other fit indices. It's more challenging to find syntax for the CFI. I agree with you regarding RMSEA, however, I wanted to select a few for my study. CFI was my second choice.
Christoph, thanks for the citation. I will have to look into that for the future.