Talk:Factor analysis/Archives/2013

Wordy and imprecise
There is a lot of wordy discussion in this article about the controversy and the differences between exploratory factor analysis and principal component analysis, but I see no precise mathematical description of the difference between the two ANYWHERE in the article. The only math that there is, at the beginning, seems completely consistent with PCA (except it allows for the errors to have non-identical variances). I'm not an expert in this field so I am not qualified to rectify this, but it seems pretty embarrassing. Is this just bad exposition or is it the actual state of the field?

Another problem: a lot of the so-called "contrasts" drawn between the two methods don't refer to common properties or concepts, which (I shouldn't have to say) is essential to drawing a CONTRAST. There are many sentences like this from the introduction: "Latent variable models, including factor analysis, use regression modelling techniques to test hypotheses producing error terms, while PCA is a descriptive statistical technique." That doesn't establish a difference between the two techniques at all, it's just describing properties of each. It is a wordy, jargony way of saying "this ball is different from that ball because this ball is red and that ball is round," a nonsense that is perpetrated over and over and over again throughout the article.

Very disappointing article.108.65.201.197 (talk) 22:27, 11 December 2012 (UTC)

I agree. If the reader didn't already know, I'm not sure the article would make clear to them that Factor Analysis is a stochastic model involving an error term while PCA is a purely mathematical procedure on a given matrix. While the two may have some similarities of intent and apprarance - those are very different starting places. — Preceding unsigned comment added by 62.25.109.195 (talk) 12:05, 11 March 2013 (UTC)

Reliability and Validity
I wanted to suggest adding a section for "Reliability and Validity" pertaining to the use of factor analysis. The article does not link factor analysis with Cronbach's alpha for testing the internal consistency of factors, which would be useful to readers. Also, discussion of the relationship of factors to theoretical constructs as as a determination of the validity would also seem important. Thelema418 (talk) 19:40, 16 March 2013 (UTC)