Discriminant Analysis - Testing Equal Covariances

[Peter Erhardt]

I am doing discriminant analysis for 2 groups and I don't see an option
to test whether the covariances between groups are equal.  In other
words, I don't know if I should be using the linear or the quadratic
discriminant functions.  In SPSS this is the "Use Covariance Matrix:
"Within Groups" or "Separate Groups" option.  However, there is not a
third option to test this equality like there is in SAS.

 

Anyone know of a way to do this in SPSS? Either with an option under
discriminant analysis, or a separate test would be fine.

 

I get digest mode usually, so I'd appreciate if you respond to me
directly as well as the list.

 

Thanks,

Pete

 

Peter Erhardt

Data & Research Specialist

Development & Alumni Relations

Rochester Institute of Technology

116 Lomb Memorial Drive

Rochester, NY 14623

585-475-5039 || 585-475-7004 (F)

[hidden email]

 

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

Peter:

I would encourage you to do logistic regression than than discriminant function analysis.  In, logistic regression you don't need to worry about homogeneity of var/cov matrices or multivariate normality.  In addition, the interpretation of the relative contributions of the covariates/predictor variables is much easier than in DFA. In logistic regression, the exponeniated coefficient is an odds ratio.

Scott R Millis, PhD, MEd, ABPP (CN,CL,RP), CStat
Professor & Director of Research
Dept of Physical Medicine & Rehabilitation
Wayne State University School of Medicine
261 Mack Blvd
Detroit, MI 48201
Email:  [hidden email]
Tel: [phone]
Fax: [phone]


--- On Wed, 1/28/09, Peter Erhardt <[hidden email]> wrote:

> From: Peter Erhardt <[hidden email]>
> Subject: Discriminant Analysis - Testing Equal Covariances
> To: [hidden email]
> Date: Wednesday, January 28, 2009, 8:53 AM
> I am doing discriminant analysis for 2 groups and I
> don't see an option
> to test whether the covariances between groups are equal.
> In other
> words, I don't know if I should be using the linear or
> the quadratic
> discriminant functions.  In SPSS this is the "Use
> Covariance Matrix:
> "Within Groups" or "Separate Groups"
> option.  However, there is not a
> third option to test this equality like there is in SAS.
>
>
>
> Anyone know of a way to do this in SPSS? Either with an
> option under
> discriminant analysis, or a separate test would be fine.
>
>
>
> I get digest mode usually, so I'd appreciate if you
> respond to me
> directly as well as the list.
>
>
>
> Thanks,
>
> Pete
>
>
>
> Peter Erhardt
>
> Data & Research Specialist
>
> Development & Alumni Relations
>
> Rochester Institute of Technology
>
> 116 Lomb Memorial Drive
>
> Rochester, NY 14623
>
> 585-475-5039 || 585-475-7004 (F)
>
> [hidden email]
>
>
>
> ====================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

[Swank, Paul R]

I agree with Scott that logistic regression is probably best here but if
you still wish to continue with the discriminant analysis, recognize
that it is basically a MANOVA. So you can use the MANOVA program to
compare the variance -covariance matrices, or at least I assume so.

Paul R. Swank, Ph.D
Professor and Director of Research
Children's Learning Institute
University of Texas Health Science Center
Houston, TX 77038


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Peter Erhardt
Sent: Wednesday, January 28, 2009 7:54 AM
To: [hidden email]
Subject: Discriminant Analysis - Testing Equal Covariances

I am doing discriminant analysis for 2 groups and I don't see an option
to test whether the covariances between groups are equal.  In other
words, I don't know if I should be using the linear or the quadratic
discriminant functions.  In SPSS this is the "Use Covariance Matrix:
"Within Groups" or "Separate Groups" option.  However, there is not a
third option to test this equality like there is in SAS.



Anyone know of a way to do this in SPSS? Either with an option under
discriminant analysis, or a separate test would be fine.



I get digest mode usually, so I'd appreciate if you respond to me
directly as well as the list.



Thanks,

Pete



Peter Erhardt

Data & Research Specialist

Development & Alumni Relations

Rochester Institute of Technology

116 Lomb Memorial Drive

Rochester, NY 14623

585-475-5039 || 585-475-7004 (F)

[hidden email]



==========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

[SR Millis-3]

To build on Paul's comment, DFA seems to be relatively robust to violations of the homogeneity assumption if the ratio of the larger group n divided by the smaller group n is less than 1.5 (Stevens, 1996). In SPSS you can test the assumption using Box's M.



Scott R Millis, PhD, MEd, ABPP (CN,CL,RP), CStat
Professor & Director of Research
Dept of Physical Medicine & Rehabilitation
Wayne State University School of Medicine
261 Mack Blvd
Detroit, MI 48201
Email:  [hidden email]
Tel: [phone]
Fax: [phone]


--- On Wed, 1/28/09, Swank, Paul R <[hidden email]> wrote:

> From: Swank, Paul R <[hidden email]>
> Subject: Re: Discriminant Analysis - Testing Equal Covariances
> To: [hidden email]
> Date: Wednesday, January 28, 2009, 10:08 AM
> I agree with Scott that logistic regression is probably best
> here but if
> you still wish to continue with the discriminant analysis,
> recognize
> that it is basically a MANOVA. So you can use the MANOVA
> program to
> compare the variance -covariance matrices, or at least I
> assume so.
>
> Paul R. Swank, Ph.D
> Professor and Director of Research
> Children's Learning Institute
> University of Texas Health Science Center
> Houston, TX 77038
>
>
> -----Original Message-----
> From: SPSSX(r) Discussion [mailto:[hidden email]]
> On Behalf Of
> Peter Erhardt
> Sent: Wednesday, January 28, 2009 7:54 AM
> To: [hidden email]
> Subject: Discriminant Analysis - Testing Equal Covariances
>
> I am doing discriminant analysis for 2 groups and I
> don't see an option
> to test whether the covariances between groups are equal.
> In other
> words, I don't know if I should be using the linear or
> the quadratic
> discriminant functions.  In SPSS this is the "Use
> Covariance Matrix:
> "Within Groups" or "Separate Groups"
> option.  However, there is not a
> third option to test this equality like there is in SAS.
>
>
>
> Anyone know of a way to do this in SPSS? Either with an
> option under
> discriminant analysis, or a separate test would be fine.
>
>
>
> I get digest mode usually, so I'd appreciate if you
> respond to me
> directly as well as the list.
>
>
>
> Thanks,
>
> Pete
>
>
>
> Peter Erhardt
>
> Data & Research Specialist
>
> Development & Alumni Relations
>
> Rochester Institute of Technology
>
> 116 Lomb Memorial Drive
>
> Rochester, NY 14623
>
> 585-475-5039 || 585-475-7004 (F)
>
> [hidden email]
>
>
>
> ==========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

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