Missing Values in Categorical Covariates in Cox Regression

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Missing Values in Categorical Covariates in Cox Regression

Atai Winkler

Hi

 

I am developing a series of Cox regression models. One observation (site_id 10434) has very high DFBETA statistics for the categorical covariate season. I therefore want to exclude this observation from the analysis and so I type

 

IF (site_id = '10434')  season = ''.

EXECUTE.

 

and then rerun the Cox regression model. Unexpectedly, this second run produced outputs (survival probability and cumulative hazard) for this record.

 

However, if I set a scale variable used in the model to $SYSMIS (see below), the second run does not produce outputs – as expected.

 

IF (site_id = '10434')  no_previous_planned_proactives_grp = $SYSMIS.

EXECUTE.

 

Why doesn’t setting a categorical covariate to ‘’ for a record exclude the record from the Cox regression procedure?

 

Thank you for your help

 

Atai

 

Dr Atai Winkler

Principal Consultant
PAM Analytics

 

[hidden email]

Skype: ataiwinkler

 

www.pamanalytics.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
Reply | Threaded
Open this post in threaded view
|

Re: Missing Values in Categorical Covariates in Cox Regression

Jon Peck
did you declare blank as a missing value?

On Wed, Jan 8, 2020 at 11:07 AM Atai Winkler <[hidden email]> wrote:

Hi

 

I am developing a series of Cox regression models. One observation (site_id 10434) has very high DFBETA statistics for the categorical covariate season. I therefore want to exclude this observation from the analysis and so I type

 

IF (site_id = '10434')  season = ''.

EXECUTE.

 

and then rerun the Cox regression model. Unexpectedly, this second run produced outputs (survival probability and cumulative hazard) for this record.

 

However, if I set a scale variable used in the model to $SYSMIS (see below), the second run does not produce outputs – as expected.

 

IF (site_id = '10434')  no_previous_planned_proactives_grp = $SYSMIS.

EXECUTE.

 

Why doesn’t setting a categorical covariate to ‘’ for a record exclude the record from the Cox regression procedure?

 

Thank you for your help

 

Atai

 

Dr Atai Winkler

Principal Consultant
PAM Analytics

 

[hidden email]

Skype: ataiwinkler

 

www.pamanalytics.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


--
Jon K Peck
[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