gillian writes: > I am using AMOS Basic to fit a One-facor and Two-factor > exploratory factor analysis model to some data. I got the result below: > > 1-facotr analysis - Chi-square: 89.286, > degree-of-freedom: 35 with P value nearly 0. 2-factor analysis - Chi-square: > 30.177, degree-of-freedom: 26 with P value 0.260. > > Would someone lighten me on the question below please: > > "Use appriopriate significance test to assess whether > fitted models provide a good fit for the data." > > How to do significance test for this case. Could I just > mention that based on the result obtained, two-factor model fits > better without doing further significance test?
I'm not an expert on Amos or factor analysis, but a few thoughts come to mind. 1. You should try clarifying the question with the person who originally asked it. 2. A formal comparison of two nested models would use the difference in chi-squared values and your degrees of freedom would be the difference in the degrees of freedom. If the difference in chi-squared values is small then the simpler model is appropriate. If the difference in chi-squared values is large then the more complex model is appropriate. 3. In Structural Equations Modeling, there is less emphasis on the chi-square test per se, and more on goodness of fit measures. The simplest one is the chi-square for the model divided by its degrees of freedom. By that criterion, the two factor model looks very good. There are a lot of other criteria and the Amos manual will describe all of these. There's also a wonderful book by Barbara Byrne (Structural Equation Modeling with AMOS) that would be worth an investment, perhaps. You didn't ask, but several things strike me as unusual. First, I didn't think that Amos could fit an exploratory factor analysis model. It can do what is often called confirmatory factor analysis. Second, the degrees of freedom seem a bit strange. It would seem to me that moving from a two factor model to a one factor model should only involve a change in a single degree of freedom (or maybe two degrees of freedom). You have a change of nine degrees of freedom. These questions might just expose my own naivete. The real experts on this topic can be found at SEMNET: http://www.gsu.edu/~mkteer/semnet.html Best of luck. Steve Simon, [EMAIL PROTECTED], Standard Disclaimer. The STATS web page has moved to http://www.childrens-mercy.org/stats. . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
