Good morning,    I am trying to create a matched case/control data set from a case (n=1,646,667) and Control (n=313,552). I have gone through the awesome article called "Simplified Matched Case-Control sampling using proc surveyselect"  http://www2.sas.com/proceedings/sugi29/209-29.pdf   But I have been unable to make it work.    I thought the sample size could be an issue, but the Case part (larger data set) runs fine only with a warning that my log is exceeds the size and needs to close, while the control part (smaller data set) crashes completely. The problems appear during the proc surveyselect part   I have listed my code below which is 99% similar to the one of the article.     Does anybody have a suspicion of what is happening?   data control; length index $25; set HDP_SCRH.c88090_mcc_control ; Index=compress(cust_gendr_cd || age_band_cd || zip3 || rx_mms || bio2016 || tmc_band || risk_band || ccm_eng_2016 || ccm_id_2016 || er_band || inp_band || fip_inn_Band || fip_oon_band || fop_inn_Band || fop_oon_band || prof_inn_Band || prof_oon_band); length_=length(index); run;   PROC FREQ DATA= Control NOPRINT; TABLES INDEX/LIST MISSING OUT=CTRLCNT (/*KEEP=INDEX COUNT*/ RENAME=(COUNT=CTRLCNT));   data Case; length index $25; set HDP_SCRH.c88090_mcc_CASE; Index=compress(cust_gendr_cd || age_band_cd || zip3 || rx_mms || bio2016 || tmc_band || risk_band || ccm_eng_2016 || ccm_id_2016 || er_band || inp_band || fip_inn_Band || fip_oon_band || fop_inn_Band || fop_oon_band || prof_inn_Band || prof_oon_band); length_=length(index); run;   PROC FREQ DATA= case NOPRINT; TABLES INDEX/LIST MISSING OUT=CASECNT (/*KEEP=INDEX COUNT*/ RENAME=(COUNT=CASECNT)); DATA ALLCOUNT; MERGE CASECNT (IN=A) CTRLCNT (IN=B); BY INDEX; IF CASECNT > 0; IF A AND NOT B THEN CTRLCNT = 0; _NSIZE_ = MIN(CASECNT,CTRLCNT); IF _NSIZE_ GT 0; run; PROC SQL; CREATE TABLE ELIGIBLE_CONTROLS AS SELECT * FROM CASE WHERE INDEX IN (SELECT INDEX FROM ALLCOUNT) ORDER BY INDEX; QUIT; PROC SQL; CREATE TABLE ELIGIBLE_CASES AS SELECT * FROM CONTROL WHERE INDEX IN (SELECT INDEX FROM ALLCOUNT) ORDER BY INDEX; QUIT; PROC SURVEYSELECT DATA =ELIGIBLE_CASES SAMPSIZE = ALLCOUNT METHOD = SRS SEED=890 OUT=SELECTED_CASES; STRATA INDEX; RUN; PROC SURVEYSELECT DATA =ELIGIBLE_CONTROLS SAMPSIZE = ALLCOUNT METHOD = SRS SEED=134 OUT=SELECTED_CONTROLS; STRATA INDEX; RUN; ... View more

Hello,   I am using the lag function which works for parts of my data set, while it doesn't in other parts (?) I am using this very simple data step below: Does anybody know what's wrong with this code?   Thank you!   data adjustment3; set adjustment3; if count_cluster=lag1(count_cluster) then Max_Salary_previous_grade=lag1(max_salary); else Max_Salary_previous_grade=0; run;       TABLE GET   count_cluster Max_salary Max_Salary_previous_grade   1            56,768                                                      -     1            70,741                                            56,768 ok! 2            49,957                                                      -     3            50,000                                                      -     4            34,360                                                      -     4            49,790                                            70,741 Error! 4            53,220                                            49,790 ok!                     TABLE WANT   count_cluster Max_salary Max_Salary_previous_grade   1            56,768                                                      -     1            70,741                                            56,768   2            49,957                                                      -     3            50,000                                                      -     4            34,360                                                      -     4            49,790                                            34,360   4            53,220                                            49,790         ... View more