IIA test (or Hausman test) in SPSS

[Ange]

Is it possible to conduct a IIA test (Independence from Irrelevant Attributes) in either Logistic or Multinomial Logistic Regression ??

The Hausman Test is often used. Is that, or other tests available in SPSS?

Thank you!

[Bruce Weaver]

Judging from the Wikipedia page on the Hausman test (which I'd never heard of), it should not be too hard to program it in the matrix language.  

    http://en.wikipedia.org/wiki/Hausman_test

The formula shown near the top, for example would look like this, I think:

compute H = T(B1-B0)*GINV(VarB0-VarB1)*(B1-B0).

where (if I followed):

B0 = a column vector of coefficients under the null hypothesis
B1 = a column vector of coefficients under the alternative hypothesis
VarB0 = a column vector of squared standard errors for B0
VarB1 = a column vector of squared standard errors for B1

HTH.

Ange wrote

Is it possible to conduct a IIA test (Independence from Irrelevant Attributes) in either Logistic or Multinomial Logistic Regression ??

The Hausman Test is often used. Is that, or other tests available in SPSS?

Thank you!

[David Marso]

It would seem that since B0 and B1 are Vectors, VarB0 and VarB1 would be matrices since there are covariances among the B's.  Looks like the MATRIX math would follow as indicated ;-)
--

Bruce Weaver wrote

Judging from the Wikipedia page on the Hausman test (which I'd never heard of), it should not be too hard to program it in the matrix language.  

    http://en.wikipedia.org/wiki/Hausman_test

The formula shown near the top, for example would look like this, I think:

compute H = T(B1-B0)*GINV(VarB0-VarB1)*(B1-B0).

where (if I followed):

B0 = a column vector of coefficients under the null hypothesis
B1 = a column vector of coefficients under the alternative hypothesis
VarB0 = a column vector of squared standard errors for B0
VarB1 = a column vector of squared standard errors for B1

HTH.

Ange wrote

Is it possible to conduct a IIA test (Independence from Irrelevant Attributes) in either Logistic or Multinomial Logistic Regression ??

The Hausman Test is often used. Is that, or other tests available in SPSS?

Thank you!

[Poes, Matthew Joseph-2]

It's natively supported as a model option in SAS.  If the poster has access to this program, I would imagine that would be easier than trying to program it into the matrix procedure.

Matthew J Poes
Research Data Specialist
Center for Prevention Research and Development
University of Illinois
510 Devonshire Dr.
Champaign, IL 61820
Phone: [phone]
email: [hidden email]


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of David Marso
Sent: Wednesday, March 28, 2012 10:19 AM
To: [hidden email]
Subject: Re: IIA test (or Hausman test) in SPSS

It would seem that since B0 and B1 are Vectors, VarB0 and VarB1 would be matrices since there are covariances among the B's.  Looks like the MATRIX math would follow as indicated ;-)
--


Bruce Weaver wrote

>
> Judging from the Wikipedia page on the Hausman test (which I'd never
> heard of), it should not be too hard to program it in the matrix language.
>
>     http://en.wikipedia.org/wiki/Hausman_test
>
> The formula shown near the top, for example would look like this, I think:
>
> compute H = T(B1-B0)*GINV(VarB0-VarB1)*(B1-B0).
>
> where (if I followed):
>
> B0 = a column vector of coefficients under the null hypothesis
> B1 = a column vector of coefficients under the alternative hypothesis
> VarB0 = a column vector of squared standard errors for B0
> VarB1 = a column vector of squared standard errors for B1
>
> HTH.
>
>
>
> Ange wrote
>>
>> Is it possible to conduct a IIA test (Independence from Irrelevant
>> Attributes) in either Logistic or Multinomial Logistic Regression ??
>>
>> The Hausman Test is often used. Is that, or other tests available in
>> SPSS?
>>
>> Thank you!
>>
>

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/IIA-test-or-Hausman-test-in-SPSS-tp5600767p5600856.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

This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal. Thank you.


This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal. Thank you.

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

[Bruce Weaver]

You could be right, David.  I was assuming the Var0 and Var1 terms were just the variances -- the terms on the main diagonals of the covariance matrices.  It's a little hard to tell on that page.  If I was programming it, I'd find a worked example somewhere and make sure I could duplicate the results!

David Marso wrote

It would seem that since B0 and B1 are Vectors, VarB0 and VarB1 would be matrices since there are covariances among the B's.  Looks like the MATRIX math would follow as indicated ;-)
--

Bruce Weaver wrote

Judging from the Wikipedia page on the Hausman test (which I'd never heard of), it should not be too hard to program it in the matrix language.  

    http://en.wikipedia.org/wiki/Hausman_test

The formula shown near the top, for example would look like this, I think:

compute H = T(B1-B0)*GINV(VarB0-VarB1)*(B1-B0).

where (if I followed):

B0 = a column vector of coefficients under the null hypothesis
B1 = a column vector of coefficients under the alternative hypothesis
VarB0 = a column vector of squared standard errors for B0
VarB1 = a column vector of squared standard errors for B1

HTH.

Ange wrote

Is it possible to conduct a IIA test (Independence from Irrelevant Attributes) in either Logistic or Multinomial Logistic Regression ??

The Hausman Test is often used. Is that, or other tests available in SPSS?

Thank you!