turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- Analytics
- /
- Stat Procs
- /
- Proc Genmod GEE datatable arangement

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

04-04-2010 01:51 AM

I have repeated measures from the same individuals, and I need to account for temporal correlations...my outcome measure is continuous (daily score on a withdrawal scale over three weeks - including 1 week of baseline "smoking as usual" and 2 weeks of abstinence) - and I have been advised that a GEE analysis is appropriate. I see that Genmod does GEE - but I'm not sure if it will take my continuous outcome data - all the examples I can find are for logistic regression type analysis with categorical outcome data....

I can set my dataset up for a repeated measures analysis in Proc Mixed ok - its simply stacked with a column for day - a column for phase (baseline or withdrawal week 1 or 2) - and my outcome variable (withdrawal score). Then you just say:

proc mixed data = data

class subject_id phase

model symptom_score = day phase day*phase

repeated subject = id

run

But it seems from the examples I can see that proc genmod does not want the data in a stacked format - but instead with each repeated measure (i.e. each day of symptom data in my example) in its own column. However with this format - I don't understand how you specify the outcome variable (i.e. in the example above - there is just 1 column with all the withdrawal scores in it - but now it would be 21 columns if I dont stack my data.....I don't understand how all the day column could have the same name and thus be specified as a single variable for the GEE analysis outcome.

Here is my GEE code:

proc genmod

data= Cws.Score_data;

/*Marker2 = days;Marker3 = Phase*/

class Subject_id Marker2;

model I_had_mood_swings = Marker2 /type3

link=identity covb;

repeated subject=Subject_id

/type=exch maxiter=25000 covb corrb;

run;

Any help greatfully received.

Thanks,

Davo

I can set my dataset up for a repeated measures analysis in Proc Mixed ok - its simply stacked with a column for day - a column for phase (baseline or withdrawal week 1 or 2) - and my outcome variable (withdrawal score). Then you just say:

proc mixed data = data

class subject_id phase

model symptom_score = day phase day*phase

repeated subject = id

run

But it seems from the examples I can see that proc genmod does not want the data in a stacked format - but instead with each repeated measure (i.e. each day of symptom data in my example) in its own column. However with this format - I don't understand how you specify the outcome variable (i.e. in the example above - there is just 1 column with all the withdrawal scores in it - but now it would be 21 columns if I dont stack my data.....I don't understand how all the day column could have the same name and thus be specified as a single variable for the GEE analysis outcome.

Here is my GEE code:

proc genmod

data= Cws.Score_data;

/*Marker2 = days;Marker3 = Phase*/

class Subject_id Marker2;

model I_had_mood_swings = Marker2 /type3

link=identity covb;

repeated subject=Subject_id

/type=exch maxiter=25000 covb corrb;

run;

Any help greatfully received.

Thanks,

Davo

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to deleted_user

04-08-2010 10:23 AM

PROC GENMOD uses the same data arrangement as PROC MIXED -- one response variable with multiple observations containing the repeated measurements of each subject. This is illustrated in the examples in the GENMOD documentation:

9.1: http://support.sas.com/onlinedoc/913/getDoc/en/statug.hlp/genmod_index.htm

9.2: http://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/genmod_toc.htm

You can use any of the response distributions offered by the DIST= option including the continuous distributions such as normal, gamma, etc.

9.1: http://support.sas.com/onlinedoc/913/getDoc/en/statug.hlp/genmod_index.htm

9.2: http://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/genmod_toc.htm

You can use any of the response distributions offered by the DIST= option including the continuous distributions such as normal, gamma, etc.