Confirmatory or explorative factor analysis?

#1
Hi

I did a statistics exam and recieved a pretty bad grade corresponding to an E (I'm a political science student, and I've done many quantitative papers so I wouldn't think I'm completely lost usually). The exam was to conduct a factor analysis to find out whether Schwartz's Human values inventory, consisting of two theoretical factors, can be replicated in Sweden. I was handed a dataset with a 10-variable battery of Schwartz's values. I was also handed an article which found three factors rather than two as supplement (perhaps only to provide the needed theoretical information, I'm not sure). Because the included article found three factors rather than the two, and because I wasn't handed any information about which variables in the battery to specify for Schwartz's original two factors, I did an explorative factor analysis. I would argue that this is acceptable for the stated reasons, though the most obvious choice perhaps would be to conduct a confirmatory factor analysis.

Am I totally wrong?

I haven't spoken to the professor yet, maybe I will. But if the choice to conduct an explorative factor analysis sounds completely off to you, I might just accept that as the reason since I really just wanna move on.

Thanks!
 
Last edited:
#3
I'm afraid I don't have the actual question since they delete it from our portal once the submission date expires.
It was something like (or very close to) this:

Apply a factor analysis to find out whether Schwartz’s value structure can be replicated for Swedish individuals.

It had some explanatory text before stating that Schwartz's theory presents basic human values in two dimensions, too.
The exam was a takehome which had to be written in four days.

I think basically what I'm asking is: does it seem crazy to conduct a explorative factor analysis in this case from a "normal" research kind of view? Perhaps I posted it the wrong place since I'm pretty familiar with basic applied statistics.

Schwartz's theory is well-defined and has been confirmed in different contexts, so it's not theory building. I was asked to check whether it's two-dimensionalness holds in Sweden, I guess, which is why I now think I probably should have conducted a confirmatory analysis instead.
 
#4
I don't know, maybe I answered my own question... Because the theoretical concept consists of two dimensions, I guess the most proper solution would be to test whether these two actually apply in Sweden.