on the ability of factor analysis to discover factors

I’m linking this post on g by Cosma Shalizi because I’ve only read it once, I haven’t understood it properly, and I don’t want to lose it.

(H/T Freddie DeBoer, via TNC.)

Advertisements

3 thoughts on “on the ability of factor analysis to discover factors

  1. I think Cosma’s point is simpler than his prolixity might suggest: factor analysis always discovers factors, but factors and causes are different things. I find reading the simulation code much clearer than the blog post: http://vserver1.cscs.lsa.umich.edu/~crshalizi/sloth/thomson-model.R

    I’m not sure if Cosma would agree, but I’d be inclined to summarize the Thomson model as an elaboration of the claim that “IQ tests are poorly designed measurements that confound many causes because each test is a sum of independent variables.”

  2. John: Thanks for that — I may have a look. I think I understood the conceptual point at the level you’re suggesting; I’m just not that well acquainted with either the intelligence literature (or, for that matter, the personality literature, which he swipes in the same post) or the quantitative details that make the whole argument go. Not that I don’t trust Cosma on the quantitative part, I’d just like to be able to understand it more deeply.

    I’ve also queued up some of Andy Conway’s papers on this; hopefully that’ll help me understand where the good quantitative people who take g seriously are coming from.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s