Depending on your data and your objectives, SAS has all sorts of procedures to deal with repeated measures. You seem to have selected Proc GLM. Do you have specific questions? I don't currently have access to SAS but I've used it and the manuals are available online. Even if I can't answer your question, being more specific may encourage someone else in the community to speak up.

Mr. Charles White, Thanks for your attention. The collected data was for fodder shrubs survival which was collected monthly. I don't want to have the month a treatment.

Thank's.

Yahya

I assume that your data are, similar to the data in the link, in the form of percent survival over a period of several months. I further assume that your data are from multiple scrub species, planting schemes, or some other treatment.

I'm not sure what to assume about your objectives. Are you:

1. Looking for differences in some kind of overall tendency to survive (mean survival?) between groups?

2. Looking to see if the change in survival (slope) is different between groups?

3. Developing a model for how survival changes for each group?

4. Looking for something completely different?

You're probably doing 1 but I'd like to be sure before I go too far.

http://ilri.org/infoserv/Webpub/fulldocs/MultiPur/Jordan.htm

Thank's. I am agree with you about the link. My data is similar to those in the link which was done by some of my colleges whom leave Rangeland Directorate few years ago. In your comment No. 1 and 3 are from my objectives in the present study.

Yahya Al-Satari,

I'll give you two answers, one for each of your research questions.

With regard to modeling survival, I recommend that you do include the month information. There are SAS procedures directly intended to do this. Where the capitalized words are values you supply and the lower case words are SAS syntax, a starting point for such an analysis could be as follows:

proc lifereg data=SHRUBS plots=all;

strata SPECIES;

time MONTH;

test SPECIES;

run;

With regard to testing for differences in survival, you may want a model that explicitly subtracts out the variability due to month. One of many ways to do this is using your GLM procedure of choice as follows:

proc glm data=SHRUBS plots=all;

model SHRUBS = SPECIES MONTH;

run;

The primary test to report is the test for SPECIES as opposed to MONTH or the overall F-test. MONTH is a nuisance variable in the model and you're just subtracting out the effect.

Hopefully, you can take it from here.

Chuck

http://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_lifetest_sect003.htm

Mr. Charles White,

Thanks, I was done your review as seen in the attach SAS file but the run process Not achieved. May be I have some mistakes in the commands, Please, review the SAS file. All the best.

As I said before, I'm experienced with SAS programming but I don't have a legal license to test my SAS code at this time. With that said, I can see some of the potential issues associated with how you interpreted what I wrote. Based on looking at your actual data and the variable names you used in your dataset, minimum required modifications are as follows:

proc lifereg data=Atriplex plots=all;

strata PDate;

time MONTH;

test PDate;

run;

proc glm data=Atriplex plots=all;

model Survival = PDate MONTH;

run;

Note that I used your data to do basic testing of the analytical approach using R (not SAS) and the approach seems sound. If my SAS code still doesn't work for you, the first thing I would recommend that you look into is if you need to declare the "type" of variables used in the procedures.

I apologize for taking so long to get back to you.

Here is a Proc example of my analyses:

a two way repeated measure (2*5)complete randomized design.

proc glm data=a;

class metal;

model y1-y10=time/ss3;

repeated Time 2, Phase 5;

run;quit;

Cheers

Response variable SURVIVAL seems to be binary, and then PROC GLM or PROC MIXED were not recomended. Try using PROC GLIMMIX to fit a generalizad linear model or some kind of Cochran-Armitage test with PROC FREQ:

proc sort data=atriplex;

by PDate;

* what is pdate?;

proc freq data=atriplex;

by PDate;

tables survival*MONTH / trend;

run;

Dear colleges; Thanks you very much for your response.