|Date: ||Thu, 10 Feb 2005 10:11:28 -0500|
|Sender: ||"SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>|
|From: ||Art Kendall <Art@DRKENDALL.ORG>|
|Organization: ||Social Research Consultants|
|Subject: ||Re: factor analysis and your experience|
|Content-Type: ||text/plain; charset=ISO-8859-1; format=flowed|
What do you want to find out via factor analysis?
What is the nature of the set of variables you have? Are they designed
to measure one or more constructs? Are they a mix of items from that
you want so use to see is there are a few underlying constructs that
might be useful?
When Likert response format items are sufficiently discrepant from
normal, it is because almost all respondents gave the same answer.
Because they have very little variance themselves, they contribute
little if anything to the variance of the summative composite. You
might as well throw them out.
Usually Likert items are consider "not severely" discrepant form
If you wish to explore the difference in methods, start with a
conventional PFA( implicitly using communalities in the diagonal) with
Varimax rotation. Then try using CATPCA and see if the derived factors
differ much as interpreted constructs.
Social Research Consultants
University Park, MD USA
Mpundu Mukanga wrote:
>I dont if my questions make sense to you. Questions: Based on your experience and knowledge of statistics would you perform factor analysis on non-normal data such as likert type dat (ordinal data) if yes what extraction and rotation techniques would you recommend?. if not what alternatives are available?. My goal as mentioned before is enable me look at the data critically and identify factors which i can use to create subscales from a pool of items i have (likert instrument).
>thanks for taking your time to read and help.
>Do You Yahoo!?
>Tired of spam? Yahoo! Mail has the best spam protection around