Tonidandel - Sampling with Replacement

Scott Tonidandel

SAMPLING WITH REPLACEMENT

The following program is a SAS macro that samples with replacement individual cases from an existing data set and creates an output file that consists of multiple "resampled" data sets. Below you will find a description of the macro, the SAS code, and an example Call Statement. For more information see: Tonidandel, S, Overall, J. E., & Smith, F. (2004). Use of resampling to select among alternative error structure specifications for GLMM analyses of repeated measurements. International Journal of Methods in Psychiatric Research, 13, 24-33.

/*****************************************************************************************/

/* */

/* Macro name: %RESAMP */

/* */

/* Purpose: Sampling With Replacement: Using a 'population' sample, generate sets of */

/* resampled data from the population. */

/* */

/* Call Statement: */

/* */

/* %RESAMP (INP=,OUTP=,SS=,DS=,SEED=); */

/* */

/* REQUIRED parameters: none */

/* */

/* OPTIONAL parameters: */

/* */

/* INP= input SAS dataset containing the 'population' sample data. If a SAS */

/* dataset is not specified, the most recently created SAS dataset (_LAST_) */

/* is used as input to the macro. */

/* */

/* OUTP= SAS dataset name for the output SAS dataset. If a SAS dataset is not */

/* specified, the output SAS dataset default is _RESAMP1_. */

/* */

/* SS= Sample size to generate for each resampled data set. */

/* Default is 200 per grp. */

/* */

/* DS= Number of resampled data sets to generate. Default is 100 sets. */

/* */

/* SEED= Seed used to initiated stream of random numbers for selection of */

/* observations. Must be an integer < 2**31-1. If seed is <=0 then the */

/* clock time is used to initiate. Default is 248531. */

/* */

/* Printed output: none */

/* */

/* Output SAS dataset: */

/* */

/* name specified by: */

/* OUTP= (unspecified, then _RESAMP1_) */

/* variables: */

/* all variables from input SAS dataset */

/* _SET_ data set identifier, ranges from 1 to DS */

/* _ID_ number of the record within the set. */

/* */

/*****************************************************************************************/

%MACRO RESAMP(INP=_LAST_,OUTP=_RESAMP1_,SS=200,DS=100,SEED=248531);

DATA &OUTP;

_PNT_=CEIL(_NOBS_*RANUNI(&SEED));

SET &INP POINT=_PNT_ NOBS=_NOBS_;

RETAIN _SET_ 1;

_ID_+1;

IF _ID_>&SS THEN DO;

IF _SET_=&DS THEN STOP;

ELSE DO;

_ID_=1; _SET_=_SET_+1;

END;

%MEND RESAMP;

/****************************************************************************************/

/*EXAMPLE CALL STATEMENT*/

%RESAMP(INP=work.origdata,OUTP=work.rsmpdata,SS=300,DS=100,SEED=80732);

run;