Survival analysis via GENLINMIXED: Any good worked examples?

[Bruce Weaver]

The FM says that GENLINMIXED can be used to perform "interval censored" survival analysis, and it gives this example:

GENLINMIXED
 /FIELDS TARGET=result2
 /TARGET_OPTIONS DISTRIBUTION=BINOMIAL LINK=CLOGLOG
 /FIXED EFFECTS=duration treatment period age USE_INTERCEPT=FALSE
 /BUILD_OPTIONS TARGET_CATEGORY_ORDER=DESCENDING  
   INPUTS_CATEGORY_ORDER=DESCENDING.

- The procedure fits a model for the target result2, using duration, treatment, period, and age as main
effects.

- The BUILD_OPTIONS subcommand specifies that the category order for the target and all factors is
descending values of factor levels; thus, the first category of each categorical field is used as the
reference category.

- The TARGET_OPTIONS specification assumes that result2 has a binomial distribution. A complementary log-log link function relates the probability of result2 to a linear combination of the predictors, excluding an intercept term.

Can anyone direct me to a nice worked/annotated example of this type of analysis?  Someone has asked me for help, but have never (so far) done this type of analysis myself, and my Google search on <spss genlinmixed survival> is not turning up anything very helpful looking.

Thanks,
Bruce

p.s. - Happy Thanksgiving (on Monday) to any fellow Canucks on the list.

[Maguin, Eugene]

Hi Bruce,
I'm going to assume that what is called "interval censored" is the same as discrete time survival. A number of years ago now, Paul Allison wrote a little green Sage book (#46: Event History Analysis) wherein he described how discrete time data could be restructured and analyzed using logistic regression. A very good, more recent book is Singer and Willett's Applied Longitudinal Data Analysis (ALDA). They discuss and compare cloglog and logit (p. 420-426). Without testing, I'd guess that GenLin would work as well as GenLinMix--unless there is a mulitevel structure involved.
Gene Maguin



-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver
Sent: Friday, October 09, 2015 2:37 PM
To: [hidden email]
Subject: Survival analysis via GENLINMIXED: Any good worked examples?

The FM says that GENLINMIXED can be used to perform "interval censored"
survival analysis, and it gives this example:

GENLINMIXED
 /FIELDS TARGET=result2
 /TARGET_OPTIONS DISTRIBUTION=BINOMIAL LINK=CLOGLOG  /FIXED EFFECTS=duration treatment period age USE_INTERCEPT=FALSE  /BUILD_OPTIONS TARGET_CATEGORY_ORDER=DESCENDING  
   INPUTS_CATEGORY_ORDER=DESCENDING.

- The procedure fits a model for the target result2, using duration, treatment, period, and age as main effects.

- The BUILD_OPTIONS subcommand specifies that the category order for the target and all factors is descending values of factor levels; thus, the first category of each categorical field is used as the reference category.

- The TARGET_OPTIONS specification assumes that result2 has a binomial distribution. A complementary log-log link function relates the probability of result2 to a linear combination of the predictors, excluding an intercept term.

Can anyone direct me to a nice worked/annotated example of this type of analysis?  Someone has asked me for help, but have never (so far) done this type of analysis myself, and my Google search on <spss genlinmixed survival> is not turning up anything very helpful looking.

Thanks,
Bruce

p.s. - Happy Thanksgiving (on Monday) to any fellow Canucks on the list.




-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Survival-analysis-via-GENLINMIXED-Any-good-worked-examples-tp5730756.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]

Hi Gene, and thanks for responding.  I was thinking GENLINMIXED will be needed, because the person who asked for my help said they are "trying to figure out whether SPSS can perform a mixed effect model on survival data with data that are clustered (i.e. center as random effect)."  And later, they added this:

"I have used this command [GENLINMIXED, presumably] for mixed effect models for binary and continuous dependent variables but I have not been able to find how SPSS deals with survival or censored outcome data. It does come with an option of censored dependent variable but I have not figured out how to incorporate time to event as in cox regression."

The reason I asked about "interval censored" survival analysis is because that's what the FM says GENLINMIXED can do.  Having now done a little more reading, I think you're right in suggesting this refers to discrete time survival analysis.  I don't know if the person who contacted me has a discrete time situation or continuous time-to-event.  

In any case, given that I've never estimated the type of model they're asking about, it sounds like a rabbit hole down which I cannot presently go, as I have many other fish of my own to fry!  (How's that for a mix of metaphors?)  ;-)

Thanks again.
Bruce

Maguin, Eugene wrote

Hi Bruce,
I'm going to assume that what is called "interval censored" is the same as discrete time survival. A number of years ago now, Paul Allison wrote a little green Sage book (#46: Event History Analysis) wherein he described how discrete time data could be restructured and analyzed using logistic regression. A very good, more recent book is Singer and Willett's Applied Longitudinal Data Analysis (ALDA). They discuss and compare cloglog and logit (p. 420-426). Without testing, I'd guess that GenLin would work as well as GenLinMix--unless there is a mulitevel structure involved.
Gene Maguin



-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver
Sent: Friday, October 09, 2015 2:37 PM
To: [hidden email]
Subject: Survival analysis via GENLINMIXED: Any good worked examples?

The FM says that GENLINMIXED can be used to perform "interval censored"
survival analysis, and it gives this example:

GENLINMIXED
 /FIELDS TARGET=result2
 /TARGET_OPTIONS DISTRIBUTION=BINOMIAL LINK=CLOGLOG  /FIXED EFFECTS=duration treatment period age USE_INTERCEPT=FALSE  /BUILD_OPTIONS TARGET_CATEGORY_ORDER=DESCENDING  
   INPUTS_CATEGORY_ORDER=DESCENDING.

- The procedure fits a model for the target result2, using duration, treatment, period, and age as main effects.

- The BUILD_OPTIONS subcommand specifies that the category order for the target and all factors is descending values of factor levels; thus, the first category of each categorical field is used as the reference category.

- The TARGET_OPTIONS specification assumes that result2 has a binomial distribution. A complementary log-log link function relates the probability of result2 to a linear combination of the predictors, excluding an intercept term.

Can anyone direct me to a nice worked/annotated example of this type of analysis?  Someone has asked me for help, but have never (so far) done this type of analysis myself, and my Google search on <spss genlinmixed survival> is not turning up anything very helpful looking.

Thanks,
Bruce

p.s. - Happy Thanksgiving (on Monday) to any fellow Canucks on the list.




-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Survival-analysis-via-GENLINMIXED-Any-good-worked-examples-tp5730756.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