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
- /
- How to run proc GLIMMIX for repeated measures

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
- Highlight
- Email to a Friend
- Report Inappropriate Content

03-18-2017 02:10 PM

Hello,

I use SAS 9.4 and I have a data set like this:

ID pcs year disease

1 45 1 1

1 47 2 1

1 50 3 1

2 42 1 0

2 43 2 0

2 46 3 0

3 40 1 0

3 40 2 0

3 49 3 0

...

I want to see how the presence or absence of disease affect the PCS scores rate of change from year 1 to year 3 for each subject. I run the following codes:

proc glimmix data=have;

class disease (ref='0');

model pcs = disease year disease*year / cl;

run;

Then I got the results but I do not think it is correct, since the estimate for intercept is around 12, but the average pcs at year 1 is over 40 instead of 12. Does anyone can help me figure out the problem?

Thank you!

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

03-18-2017 10:30 PM

In GLIMMIX , use RANDOM to get repeated measures ( R- side random effect). proc glimmix data=have; class disease (ref='0') id year ; model pcs = disease year disease*year / cl; random year/subject=id type=ar(1) residual; run;

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

03-19-2017 01:41 PM

Thank you for the reply! When I run your codes I met another problem.

I can get parameter estimates for intercept, disease, year, and disease*year. But when I add some other covariates into the model, i found that the estimates for intercept and disease changed but the estimate for disease*year remained the same. I am not quite sure if this is correct? (I think the estimate for disease*year should change and the estimates for intercept and disease should remain the same).