EFA using HETCOR

[yael918]

I downloaded HETCOR so that I could run an EFA on categorical data.
Could anyone walk me through how to use it? I know how to run an EFA on SPSS, I just am not sure about adding this additional step.
Thanks

[Jon K Peck]

HETCOR simply provides Pearson, polyserial, or polychoric correlations for a set of variables according to their measurement levels.  Once installed (Utilities > Extension Bundles > Install Extension Bundle), you will have a dialog box interface for it  (Analyze > Correlate > Heterogeneous Correlations) as well as the SPSSINC HETCOR syntax.

Note that this one of the few extensions that requires both the Python and R plugins.


Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
new phone: [phone]




From:        yael918 <[hidden email]>
To:        [hidden email],
Date:        12/17/2012 07:23 AM
Subject:        [SPSSX-L] EFA using HETCOR
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

I downloaded HETCOR so that I could run an EFA on categorical data.
Could anyone walk me through how to use it? I know how to run an EFA on
SPSS, I just am not sure about adding this additional step.
Thanks



--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/EFA-using-HETCOR-tp5716960.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

[Art Kendall]

I have not used HETCOR.  But it is a way of generating the correlation matrix.
Then you would use the option in FACTOR that reads in a matrix.

However, have you looked at CATPCA?  It has the ability to see what level of measurement assumptions are necessary. 

One reason I mention this is that ordered categories (e.g., extent response scales, Likert scale) can often be treated as "not severely discrepant form interval level".

In general, treating response scales as nominal level requires many more cases.

If you do not have CATPCA and you items are at least ordered, it would be worthwhile to compare the substantive results of doing FACTOR using the original data and using HETCOR to generate the correlation matrix.  If they come out very close, e.g., if you are developing scales and if the items load cleanly on the same factors so they would have the same scoring key in the scales, then simply report the conventional Factoring and footnote that HETCOR gave substantively the same results.
On 12/14/2012 5:33 PM, yael918 wrote:

I downloaded HETCOR so that I could run an EFA on categorical data.
Could anyone walk me through how to use it? I know how to run an EFA on
SPSS, I just am not sure about adding this additional step.
Thanks



--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/EFA-using-HETCOR-tp5716960.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