Adaptive Tests of Significance Using Permutations of Residuals with R and SAS

APPENDIX C

SAS MACRO FOR MULTIPLE COMPARISONS PROCEDURES

Note: The SAS macro adaptall.sas must also be included. It is in Appendix B and on the file adaptall.sas.

/*

Notes:

• 1) This macro must be included in a SAS program that also has the “adaptall” macro.
• 2) No missing values are allowed in the dependent or independent variables.
• 3) If variables are formatted, then the formats will need to be placed before the macro.
• 4) The macro is to be used with a one-way layout. Do not specify anything for xr or classr; they are not used as inputs in this macro.
• 5) The indicator variable for the groups should be specified as the xa and classa variables.
• 6) The number of independent variables should not be specified; it will be calculated by this macro before the call to the awls macro.
• 7) The number of observations must not exceed 10,000.

*/

%macro adaptmc(dataset=, y=, xa=, classa=,
                  seed1=, seed2=, seed3=, nperm=2000);
proc sort data=&dataset out=dsorted;
  by &classa;
data dk;
  retain k 0 ;
  keep k ;
  set dsorted end=last;
    by &classa;
  if first.&classa then do;
    k=k+1;
    call symput (‘km’,k);
    end;
  if last then output;
  run;
%do imc=2 %to &km;
%do jmc=1 %to &imc−1;
 data expand;
   length lower upper $ 8;
   retain lower upper;
   retain c 0;
   file print noprint;
   keep &y &xa newclass ia;
   set dsorted end=last;
     by &classa;
   if first.&classa then do;
       c = c + 1;
       if c = &jmc then lower=&classa;
       end;
   if c = &imc then do;
                              newclass=lower;
                              ia=1 ;
                              upper=&classa;
                              end;
                         else do;
                              newclass=&classa;
                              ia=0 ;
                              end;

if last then ...