MIXED: Why no z (or Wald) test for covariance parameters?

[Bruce Weaver]

In the table of Covariance Parameters produced by MIXED, the estimates and their standard errors are shown, but the user is left to their own devices if they want to compute z = estimate / SE.  I understand that such z-tests may have some limitations...but still, some authors do look at them to get an indication as whether more explanatory variables may be needed.  E.g., Singer & Willett do this a lot in Applied Longitudinal Data Analysis.  So, to any SPSS folks who are reading this, how about it?  Can we have z and p added to that table in future (perhaps with a footnote commenting on the limitations of the test)?  Thanks for considering.

[SPSS Support]

Hi Bruce,

Wald tests and confidence intervals for the covariance parameters have been available since the initial implementation of the MIXED procedure. Our reaction to the concerns with validity due to testing of a null value on the boundary of the parameter space was simply to not make these default output. To get Z values, significances, and confidence intervals, just go into the Statistics dialog and check the box labeled "Tests for covariance parameters" in the Model Statistics grouping. In command syntax, it's the TESTCOV keyword on the PRINT subcommand.

David Nichols

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver
Sent: Wednesday, September 02, 2009 3:40 PM
To: [hidden email]
Subject: [SPSSX-L] MIXED: Why no z (or Wald) test for covariance parameters?

In the table of Covariance Parameters produced by MIXED, the estimates and their standard errors are shown, but the user is left to their own devices if they want to compute z = estimate / SE.  I understand that such z-tests may have some limitations...but still, some authors do look at them to get an indication as whether more explanatory variables may be needed.  E.g., Singer & Willett do this a lot in Applied Longitudinal Data Analysis.  So, to any SPSS folks who are reading this, how about it?  Can we have z and p added to that table in future (perhaps with a footnote commenting on the limitations of the test)?  Thanks for considering.


-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/
"When all else fails, RTFM."

NOTE:  My Hotmail account is for posting only, and is not monitored regularly.
If you wish to send me an e-mail, please use the address shown in my sig file.
--
View this message in context: http://www.nabble.com/MIXED%3A-Why-no-z-%28or-Wald%29-test-for-covariance-parameters--tp25265343p25265343.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

[Bruce Weaver]

Ah, very good.  I obviously failed to follow the advice in my own sig file (i.e., RTFM)!  Thanks David.

Bruce

SPSS Support wrote

Hi Bruce,

Wald tests and confidence intervals for the covariance parameters have been available since the initial implementation of the MIXED procedure. Our reaction to the concerns with validity due to testing of a null value on the boundary of the parameter space was simply to not make these default output. To get Z values, significances, and confidence intervals, just go into the Statistics dialog and check the box labeled "Tests for covariance parameters" in the Model Statistics grouping. In command syntax, it's the TESTCOV keyword on the PRINT subcommand.

David Nichols

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[e-mail]] On Behalf Of Bruce Weaver
Sent: Wednesday, September 02, 2009 3:40 PM
To: [e-mail]
Subject: [SPSSX-L] MIXED: Why no z (or Wald) test for covariance parameters?

In the table of Covariance Parameters produced by MIXED, the estimates and their standard errors are shown, but the user is left to their own devices if they want to compute z = estimate / SE.  I understand that such z-tests may have some limitations...but still, some authors do look at them to get an indication as whether more explanatory variables may be needed.  E.g., Singer & Willett do this a lot in Applied Longitudinal Data Analysis.  So, to any SPSS folks who are reading this, how about it?  Can we have z and p added to that table in future (perhaps with a footnote commenting on the limitations of the test)?  Thanks for considering.


-----
--
Bruce Weaver
[e-mail]
http://sites.google.com/a/lakeheadu.ca/bweaver/
"When all else fails, RTFM."

NOTE:  My Hotmail account is for posting only, and is not monitored regularly.
If you wish to send me an e-mail, please use the address shown in my sig file.
--
View this message in context: http://www.nabble.com/MIXED%3A-Why-no-z-%28or-Wald%29-test-for-covariance-parameters--tp25265343p25265343.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to [e-mail] (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
[e-mail] (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