Backward Discriminant Analysis

[Eric Black]

Hello - is there a way in SPSS to run a discriminant analysis stepwise
backwards?  All of the options I see seem to be forward but I would need a
backward approach

Thanks

Matt

=====================
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

[SR Millis-3]

I'd suggest staying away from all stepwise methods:

http://www.stata.com/support/faqs/stat/stepwise.html

 

~~~~~~~~~~~
Scott R Millis, PhD, ABPP, CStat, PStat®
Professor
Wayne State University School of Medicine
Email: [hidden email]
Email: [hidden email]
Tel: [phone]

---

From: Matt Kretschmer <[hidden email]>
To: [hidden email]
Sent: Wednesday, September 14, 2011 4:06 PM
Subject: Backward Discriminant Analysis

Hello - is there a way in SPSS to run a discriminant analysis stepwise
backwards?  All of the options I see seem to be forward but I would need a
backward approach

Thanks

Matt

=====================
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]

As Dr. Millis mentioned, stepwise procedures are highly questionable.
However if you have only 2 groups and if for some reason you persist in doing a stepwise approach,
REGRESSION will give the same results.  (use /save to create the the variables and CROSSTABS to get the cross-classification table).

Art Kendall
Social Research Consultants

On 9/14/2011 4:06 PM, Matt Kretschmer wrote:

Hello - is there a way in SPSS to run a discriminant analysis stepwise
backwards?  All of the options I see seem to be forward but I would need a
backward approach

Thanks

Matt

=====================
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