Re: identifying cause of message "error involving DFB10_1"

[Stuart Richardson]

On Mon, 14 Aug 2006 09:45:08 +1000, Simon Freidin <[hidden email]>
wrote:

>Resolution number: 48077  Created on: Aug 19 2004
>
>Problem Subject:  Cox regression won't save partial residuals or DFBETAs
>with ten or more parameters in the model
>
>Problem Description:  I'm running a Cox regression in SPSS, and have
>attempted to save partial residuals or DFBETAs, but I get a log message
>
>"ERROR INVOLVING PR10_1"
>
>when attempting to save partial residuals, or
>
>"ERROR INVOLVING DFB10_1"
>
>when attempting to save DFBETAs, followed by a warning reading:
>
>"Using the specified name for a new variable would create two variables in
>the working file with the same name.
>This command is not executed."
>
>My Cox regression is run, but the desired variables are not saved. What's
>going on?
>
>Resolution Subject: Program defect
>
>Resolution Description:
>This is a program defect in SPSS, occurring when DFBETAs or partial
>residuals are specified to be saved without specifying saving of other
>termporary variables (or when specifying just DFBETAs and XBETA). It has
>been reported to SPSS Development. We apologize for the inconvenience.
>Fortunately, a simple workaround exists: simply also specify saving of one
>of the other optional variables, such as the survival function, and your
>variables should be saved.
>
>At 08:15 AM 14/08/2006, you wrote:
>>I'm attempting to save df Betas in Cox regression with time dependent
>>covariates.  I'm using a time dependent covariate with 3 covariates and 3
>>interactions.  Each time I've attempted to save the df betas, I've

recieved

>>the following error:
>>
>>Error involving DFB10_1
>>
>>The following warning also appears.
>>
>>Warnings
>>Using the specified name for a new variable would create two variables in
>>the working file with the same name.
>>This command is not executed.
>>
>>I've searched the file repeatedly for any df beta variables that might
>>remain from earlier runs but don't find any.  My questions: is the error
>>involving DFB10_1 likely related to the warning about duplicate variables?
>>If it is, how can I delete the duplicate variable I can't find?  What else
>>might cause the error message?
>>
>>I'm running SPSS 12.0.
>
>
>Research Database Manager and Analyst
>Melbourne Institute of Applied Economic and Social Research
>The University of Melbourne
>Melbourne VIC 3010 Australia
>New Tel: [phone] New Fax: [phone]
>http://www.melbourneinstitute.com/hilda/

Thank you for the response.  However, after implementing your solution, I
am now recieving the following error message:

"When there are time-dependent covariates, DFBETA is the only SAVE
statistic that can be computed. Other requested statistics will not be
saved.
Using the specified name for a new variable would create two variables in
the working file with the same name.
This command is not executed."

Your solution does work as you describe when used without time dependent
covariates.  Any other thoughts as to what I can do, having, as I do, a
time dependent covariate?