Re: Regression / Factor Analysis

Posted by Art Kendall on Mar 25, 2012; 7:32pm
URL: http://spssx-discussion.165.s1.nabble.com/Regression-Factor-Analysis-tp5591552p5593727.html

A lot depends on the actual meanings of the variables you are asking about.

What is the purpose of the over all study?  Do you have hypotheses?  Exploring a phenomenon to generate possible meaning?  Developing a prediction but don't care about the meaning of the equation? etc.

What is the dependent variable? How was it measured?
How did you select what the independent variables would be? Are the continuous, nominal, dichotomous or what?

How did you gather your data?  Survey?  Existing records? Experiment? etc.
Are the cases people, projects, pieces of text, etc?
How many cases do you have?

A common reason to use factor analysis is to create summative scales.  A scale is made up of items that appear to be measuring something in common., an attitude, a skill, knowledge, etc.


Roughly, "variables do fit well" in the factor solution" when they do not load strongly, cleanly, and meaningfully on a factor.  In further analysis they might or might not be kept as additional independent (predictor) variables depending on their meaning.






On 3/24/2012 7:32 AM, Sandy wrote:

I have 30-31 independent variables and a dependent variable . I wish to know
which of these variables are significant for the analysis. In other words ,
to identify the variables which can be excluded from the analysis .
I run multiple regression analysis but in this some of the variables are
automatically excluded because of multicollinearity between the variables so
i move my thoughts to Factor Analysis which can be used in data reduction .
As i am new to Factor Analysis i find problem in interpreting the results .I
know Initial and Extraction Communalities which can be used to identify the
variables that do not fit well within the factor solution .

I need help in interpreting Component / Rotated Component Matrix.

Please suggest me the other analysis as well which can solve my purpose .

Note : I know this is not a SPSS query but any help would be highly
appreciated !

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Regression-Factor-Analysis-tp5591552p5591552.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD