Re: Factor Analysis for non-metric variables

Posted by Art Kendall on Jul 15, 2012; 12:30pm
URL: http://spssx-discussion.165.s1.nabble.com/Factor-Analysis-for-non-metric-variables-tp5714188p5714228.html

CATPCA (categorical principal components analysis) can handle all levels of measurement.

However, those seem unusual variables to put in a factor analysis?  Why are you putting all those in one factor analysis?

That message means that some variable(s) is/are perfectly predictable from some other variable(s).
Some common reason for that are:
a typo that put the same variable in the list twice,

more variables than cases,

including a total score and the items that went into it,

when dummy variables are used to represent categories, the user forgot to have one fewer dummies that there are categories of a variable.
A simplistic example would be to represent gender with a dummy for male and a dummy for female.

Check the archives of this list, Google "nabble spssx". for many discussions of this.

After you check these things please post what you are trying to do with the factor analysis.
On 7/13/2012 7:51 AM, christina0822 wrote:

Dear All:

I am now struggling with my MSc dissertation and there are two problems with
Factor Analysis now. The first one is how can I include non-metric variables
into FA, like Gender, Marital status, occupation, residential area,etc? The
other one is every time I tried KMO and Barlett's test , the output never
show up. there is an notification under the correlation matrix: "the matrix
is not positive definite" I don't know what's the meaning for that. And How
can I solve these two problems? Or anyone can kindly see my output result, I
will send it to you.

Many Thanks

Christina

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Factor-Analysis-for-non-metric-variables-tp5714188.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