Hello, Producing the table shown is easy of course, but I have no idea on how to detect 'statistical differences between them'. The fact that this question was/is still un-answered makes me think your question is not entirely clear and/or your data are not suitable for this test. Here's the program to produce the table: data new; input day $ live death; cards; Day_0 37 0 Day_6 36 1 Day_12 21 16 Day_18 8 29 Day_24 2 35 ; run; proc means data=new /*nway*/ MAX noprint; var live; output out=MaxLive(drop=_TYPE_ _FREQ_) max=MaxLive; run; data want(drop=MaxLive); if _N_=1 then set MaxLive; set new; Total=live+death; Survival=live/MaxLive; format Survival percent7.2; run; proc print; run; /* end of program */ Cheers, Koen ... View more

Hello, You should specify a window with weights in parentheses for the MOVAVE transformation operation. And you give a weight of 0 to the current observation. Like this: data have; input ID Price; cards; 1 10 2 0 3 20 4 50 ; run; proc expand data=have out=want method=none; convert price=price_avg / TRANSFORMOUT=(movave (0.33333 0.33333 0.33333 0)); run; (If you don't like this approach, you will have to use PROC TIMEDATA or the datastep.) Cheers, Koen ... View more