View Full Version : Factor Analysis


Alexis Newton
11-28-2005, 01:43 PM
I am working on a factor analysis assignment and having difficulty getting started in SPSS. Can anyone help? I would appreciate any assistance!

quark
11-28-2005, 02:05 PM
Alex,

Below are some articles I found on google.

Factor Analysis Using SPSS (www.sussex.ac.uk/Users/andyf/teaching/rm2/factor.pdf)

http://www2.chass.ncsu.edu/garson/pa765/factor.htm

Susanbwh
02-08-2006, 02:04 PM
Could anyone explain which is better for my data? I have about 30 items that are related with each other (psychometric ...). Thanks,

JohnM
02-08-2006, 02:12 PM
This link briefly explains the different rotation methods. Varimax is the most common method, as it allows for easier interpretation of factors.


http://www2.chass.ncsu.edu/garson/pa765/factor.htm

Susanbwh
02-08-2006, 02:17 PM
Varimax is the most common method, as it allows for easier interpretation of factors.

How does it matter in statistics or SPSS whether I want to do exploratory or confirmatory analysis? Actually I had about 30 items (40 patients) measuring a chance to be predisposed to some psychatric or neurologic disorders. And we know which items we want to include to extract several factors. We are not using all the items. We want to make sure that we are making the right decision about which items go together to make a factor or factors.

JohnM
02-08-2006, 02:22 PM
In exploratory factor analysis, you're not sure what the factors are, so you measure a bunch of variables related to the phenomena at hand and see how they cluster together.

In confirmatory factor analysis, you know what the factors are, and you want to confirm it with a follow-on study.

Just a comment - a sample size of 40 is not nearly enough for factor analysis. The basic rule of thumb is a sample size of 10 per original variable - so if you have 30 variables, you need at least a sample of 300.

Susanbwh
02-08-2006, 02:55 PM
That is what I thought. So, I can only use several items, then. What about correlation coefficient??? I can do it with 40 patients???

JohnM
02-08-2006, 03:09 PM
40 should be enough to do a decent correlation.

Susanbwh
02-08-2006, 03:13 PM
Thank you! The last question for today--how do you interpret "degree of freedom"?

JohnM
02-08-2006, 03:27 PM
It's the number of independent measurements that go into the estimation of a parameter (mean, std dev, etc.)

jayachandran
11-19-2006, 12:15 PM
I am working on a factor analysis assignment and having difficulty getting started in SPSS. Can anyone help? I would appreciate any assistance!
also i need how can i do with minitab