O'Reilly logo

Adaptive Tests of Significance Using Permutations of Residuals with R and SAS by Thomas W. O'Gorman

Stay ahead with the world's most comprehensive technology and business learning platform.

With Safari, you learn the way you learn best. Get unlimited access to videos, live online training, learning paths, books, tutorials, and more.

Start Free Trial

No credit card required

APPENDIX I

SAS MACRO FOR CONFIDENCE INTERVALS

 /*
    Note: If variables are formatted then the formats will
      need to be placed before the macro.
  */
options nocenter linesize=80 nonotes ;
%macro adaptci(dataset=, y=, xr=, xa=, seedl=, seed2=,seed3=,
                  maxsteps=8000);
data dseed;
   nextl=&seedl;
   next2=&seed2;
   next3=&seed3;
data dupdate;
  c=0;
data dgarth(keep=&y &xr &xa);
  retain n 0;
  set &dataset;
n = n + 1 ; call symput(‘nobs’,trim(left(put(n,8.) ) ) ); output; proc reg data=dgarth; model &y=&xr &xa/clb; titlel ‘ Regression Analysis with 95% Confidence Intervals ’; /* Now begin finding the adaptive estimate for parameter */ proc glm data=dgarth noprint; model &y=&xr; output out=dresid predicted=yhat1 residual=res; title2 ‘Adaptive weighted least squares analysis’; %adaptwtsl(datares=dresid, resvars=res, nobs=&nobs, nvars=l); data dall; merge dgarth dresid dweights; proc print data=dall ; var &y &xr &xa res weight saswt; title2‘ Data, after adjustment SASWT=weights used by SAS’; proc reg data=dall outest=dunperm outseb noprint; model &y = &xr &xa; weight saswt; data dtunperm(keep=tunperm beta); set dunperm; retain beta; if _type_ eq ‘PARMS’ then beta=&xa; if _type_ eq ‘SEB’ then do; tunperm=beta/&xa; output; end; file print noprint; titlel ‘Adaptive Weighted Least Squares Confidence Intervals’ title2 ‘Search method based on method proposed by ’; title3 ‘Garthwaite, P. H., Biometrics, vol. 52, 1996.’; put ; proc reg data=dgarth outest=dtheta outseb noprint; ; model &y=&xr &xa; data dupdate; ...

With Safari, you learn the way you learn best. Get unlimited access to videos, live online training, learning paths, books, interactive tutorials, and more.

Start Free Trial

No credit card required