Exploratory and Confirmatory
If you have read my other post, I’m all about Farmer’s Talk. Perhaps it’s the innate tendency of me as a DC residence, but I despise acronyms if I can help it. Moreover, I am technically part of the Gen Y crowd, yet I can’t see myself talking in letters which makes no sense never mind bear any cultural distinction of one’s language.
But I digress…
In short, Factor Analysis is a way to find some form of relationships (or lack thereof) between variables (wiki’s version). There are two types:
- Exploratory: Find the factors.
- Confirmatory: Confirm the factors you find in Exploratory (aka EFA from the cool kidz) is legit.
By way of Factor Analysis Introduction
- Part 1: Intro
- Part 2: Rotation
- Part 3: Diff btw EFA & CFA
- Part 4: How many factors should I find?
- Part 5: Dropping unimportant variables in analysis
- Part 6: Common Problems
Another good introduction from Univ. of Hawaii
I’m sure there are tons more reason to do factor analysis than what I bullet below, but here are my experience thus far with Factor Analysis:
- Determine groupings (err.. Factors, Categories, etc.) of variables for segmentation purposes.
- Identify redundant variables for ease modeling by using subset of variables, thereby minimize run-time and/or ‘understanding-of-what-the-heck-you-are-looking-at’ time
In (our beloved, ‘high barrier to entry’) SAS:
- Proc FACTOR
- Proc CALIS
In (people’s republic of) R:
Exploratory: <to be written>
Confirmatory: <to be written>