1

As a conceptual question: in exploratory factor analysis, how is a factor created? I would like to know your simple answer to this simple question.

Imagine, my academic field does not dependent on statistics. (Oh, impossible? -- so just imagine like J. Lennon!)

amoeba
  • 93,463
  • 28
  • 275
  • 317
Bernard
  • 363
  • 1
  • 3
  • 10
  • My shallow understanding of EFA is from Singular Value Decomposition(SVD) of a matrix. It used to reduce your data dimension, i.e you can use a smaller matrix to represent a bigger matrix. Then for each colum (or a row) of your samller matrix, you will create a name for them. The name you created for the column(or row) is called a "factor", which can be anything you think is reasonable by combining varialbes in you small matrix. – Deep North Jul 05 '15 at 13:14
  • 1
    You _have_ to get an idea of PCA before you sit down to try to imagine FA. The very basic imagination is that correlated variables are a bucket of plain flowers, the bunch. And you have one more, gorgeous flower to stick it in to make an ikebana. At what position to other flowers is to stick the flower to make it maximally crown (summarize, explain, accomplish) the idea of the whole specific bucket? First listen to Lennon, then proceed smoothly to [this thread](http://stats.stackexchange.com/q/1576/3277) and subsequently possibly for [these pics](http://stats.stackexchange.com/a/95106/3277). – ttnphns Jul 05 '15 at 18:11
  • P.S. The gorgeous flower I spoke about - You don't really have it (too expensive to buy) - you have to imagine it. – ttnphns Jul 05 '15 at 18:35

0 Answers0