Should I perform CFA after the EFA or can I move directly to multiple regression analysis? I have 66 indicators of leadership, and I want to check which ones contribute the most to leadership effectiveness. I have performed EFA, resulting 8 constructs from leadership indicators.
Should I perform CFA after that or can I move directly to multiple regression analysis?
Thanks!
 A: Unless you are very careful to make appropriate statistical adjustments, it is generally a bad idea to use the same data for exploratory analysis and confirmatory analysis.  The problems with doing this is that you tend to bias your confirmatory analysis in favour of confirming patterns that were identified in exploratory analysis.  In your exploratory analysis you often select variables as a subset of the available variables, based on observing patterns between those variables and the response of interest.  Performing confirmatory analysis via hypothesis testing with the same data, and with a model that has already incorporated that variable selection (and without statistical adjustment for the selection of variables) will be heavily biased in favour of showing a relationship between the selected variables and the response.
For this reason, when undertaking exploratory and confirmatory analysis, the usual injunction is to use different data for these two parts of the analysis.  (This doesn't necessarily require two lots of data collection, since you can just split your data into two parts via simple-random-sampling.)  The use of separate data for these two parts allows you to undertake confirmatory analysis without bias in favour of acceptance of the relationships identified loosely in exploratory analysis.
In regard to this issue, it is worth considering the warning of the philosopher Ronald Giere:

“If the known facts were used in constructing the model and were thus built into the resulting hypothesis … then the fit between these facts and the hypothesis provides no evidence that the hypothesis is true [since] these facts had no chance of refuting the hypothesis.”
— Giere, R.N. (1984) Understanding Scientific Reasoning, p. 408.

A: First, do these 8 constructs make sense? Are there theoretical reasons for the individual tests to hang together in these 8 constructs?
I would say you need to do some CFA to confirm what you found with your explorations with EFA. 
And what do you mean by move straight to multiple regression analysis? Is this to use the 8 constructs with other variables?
