log of both x and y axis in Cox regression plots

[bgreen]

Hello,

Below is the syntax I have been using to perform a Cox regression. I
would like to confirm whether this syntax generates a log of both the
x and y axes. If not, how would I amend the code to obtain such plots?

Any assistance is appreciated,

regards

Bob



COXREG
   intdaysu  /STATUS=event_v('TRUE' 'FALSE')
   /PATTERN BY mhc
   /PATTERN BY pre
   /PATTERN BY strgende
   /PATTERN BY group
     /CONTRAST (pre)=Indicator  /CONTRAST (mhc)=Indicator  /CONTRAST (strgende
  )=Indicator  /CONTRAST (group)=Indicator
   /METHOD=ENTER pre mhc intageat strgende group
   /PLOT HAZARD LML
   /CRITERIA=PIN(.05) POUT(.10) ITERATE(20) .

# obtain plots for observed categorical variables

COXREG
   intdaysu  /STATUS=event_v('TRUE' 'FALSE')/ STRATA=mhc
     /PLOT HAZARD LML.

COXREG
   intdaysu  /STATUS=event_v('TRUE' 'FALSE')/ STRATA=pre
     /PLOT HAZARD LML.
COXREG
   intdaysu  /STATUS=event_v('TRUE' 'FALSE')/ STRATA=strgende
     /PLOT HAZARD LML.
COXREG
   intdaysu  /STATUS=event_v('TRUE' 'FALSE')/ STRATA=group
     /PLOT HAZARD LML.

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

[Marta Garcia-Granero]

Bob Green wrote:
> Below is the syntax I have been using to perform a Cox regression. I
> would like to confirm whether this syntax generates a log of both the
> x and y axes. If not, how would I amend the code to obtain such plots?

I think you should add the other covariates, in order to get the oberved
LML plot for one covariate ADJUSTED for the others.

* For the 1st covariate, the observed LML plot syntax should be *.
COXREG   intdaysu  /STATUS=event_v('TRUE' 'FALSE')
  /STRATA=mhc
  /CONTRAST (pre)=Indicator
  /CONTRAST (strgende)=Indicator
  /CONTRAST (group)=Indicator
  /METHOD=ENTER pre intageat strgende group
  /PLOT HAZARD LML
  /CRITERIA=PIN(.05) POUT(.10) ITERATE(20) .

* This was your syntax *.

# obtain plots for observed categorical variables
COXREG
  intdaysu  /STATUS=event_v('TRUE' 'FALSE')/ STRATA=mhc
    /PLOT HAZARD LML.


HTH,
Marta

--
For miscellaneous statistical stuff, visit:
http://gjyp.nl/marta/

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

[bgreen]

Marta,

Many thanks for both your replies. Inboxes all too often are overflowing.

For the time being, I'll just focus on the
log-log plots of the categorical predictors. Can
I check, whether the time variable (x axis) is
also log transformed?  The statistician I have
been consulting wanted both plot axes log transformed.

Your assistance us much appreciated.

regards

Bob



At 06:52 PM 6/03/2009, Marta García-Granero wrote:

>Bob Green wrote:
>>Below is the syntax I have been using to perform a Cox regression. I
>>would like to confirm whether this syntax generates a log of both the
>>x and y axes. If not, how would I amend the code to obtain such plots?
>
>I think you should add the other covariates, in order to get the oberved
>LML plot for one covariate ADJUSTED for the others.
>
>* For the 1st covariate, the observed LML plot syntax should be *.
>COXREG   intdaysu  /STATUS=event_v('TRUE' 'FALSE')
>  /STRATA=mhc
>  /CONTRAST (pre)=Indicator
>  /CONTRAST (strgende)=Indicator
>  /CONTRAST (group)=Indicator
>  /METHOD=ENTER pre intageat strgende group
>  /PLOT HAZARD LML
>  /CRITERIA=PIN(.05) POUT(.10) ITERATE(20) .
>
>* This was your syntax *.
>
># obtain plots for observed categorical variables
>COXREG
>  intdaysu  /STATUS=event_v('TRUE' 'FALSE')/ STRATA=mhc
>    /PLOT HAZARD LML.
>
>
>HTH,
>Marta
>
>--
>For miscellaneous statistical stuff, visit:
>http://gjyp.nl/marta/
>
>=====================
>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

[Marta Garcia-Granero]

Bob Green wrote:
> For the time being, I'll just focus on the log-log plots of the
> categorical predictors. Can I check, whether the time variable (x
> axis) is also log transformed?  The statistician I have been
> consulting wanted both plot axes log transformed.
I don't think the LML plots have X axis log transformed. Edit manually
the charts and change the X scale to log (don't use safe log). save a
chart template and apply it to eery LML plot.

Just curious: why does your statistician want the X axis log scaled?

Marta

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

[Marta Garcia-Granero]

I wrote:
> Bob Green wrote:
>> For the time being, I'll just focus on the log-log plots of the
>> categorical predictors. Can I check, whether the time variable (x
>> axis) is also log transformed?  The statistician I have been
>> consulting wanted both plot axes log transformed.
> I don't think the LML plots have X axis log transformed. Edit manually
> the charts and change the X scale to log (don't use safe log). Save a
> chart template and apply it to every LML plot.
>
Alternatively, log-transform the time variable and run Cox regression.
See this example:

TEMPORARY.
COMPUTE FuTime=LG10(FuTime).
COXREG FuTime
 /STATUS=Event(1)/STRATA=reclevel
 /CONTRAST (stage)=Indicator(1)
 /METHOD=ENTER stage
 /PLOT LML.


Marta

--
For miscellaneous statistical stuff, visit:
http://gjyp.nl/marta/

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