SL

For your first question you would use proc freq with the trend option in the table statement to invoke the Cochran-Armitage trend test. A starting point would be as follows:

proc freq data = data set; by region; tables disactrn*trtsort / trend; run;

There are further options and statements that can be added to do a more complete analysis including plot statements. I don't believe that you would need to transform or recode the data.

For the ANOVA you can use either proc ANOVA or proc glm. Proc glm is the more preferred method and a starting point for that, assuming disactrn is the response variable, is as follows:

proc glm data = data set; class region; model disactrn=trtsort; /*or reverse it or add more variables depending on what you want to test*/ run;

Play with these with your data they should get you started. Check the documentation for other options and statements that cn be added to enhance your analyses.

Rich

On Mon, 10 Jan 2011 15:14:16 -0600, SAS_learner <proccontents@GMAIL.COM> wrote:

>Hello all, > >My data looks somewhat like this > >region trtsort disactrn >Asian 1 0 >Asian 2 1 >Asian 3 1 >Asian 4 0 >Asian 5 1 > > >I need to calculate > >[1] 1-sided dose-response relationship from Cocharan -Armitage test ( I >have one Placebo and 5 treatment arms) and this table is by regions. > >If I can use Proc Freq is there way I need to have the data to get p-value . >what are the options that I need use. A small example is greatly >appreciated. > >For the second table > > >region trtsort disactrn >Asian 1 10 >Asian 2 23 >Asian 3 11 >Asian 4 10 >Asian 5 31 > >For above data I need to calculate > > >1) 1-sided ANOVA with trt as fixed factor comparing all the LXXX vs.Placebo >2) 1-sided ANCOVA with trt as fixed factor and the baseline value as >covariate comparing all LXXX Vs. Placebo > > >A small example is greatly appreciated. > >thanks >SL