TY - JOUR
AU - AUSTIN, HARLAND
AB - AMERICAN JOURNAL OF EPIDEMIOLOGY VoL 124, No. 1 Printed m U SA Copyright © 1986 by The Johns Hopkins University School of Hygiene and Public Health All rights reserved Brief Reports POSSIBILITY OF SELECTION BIAS IN MATCHED CASE-CONTROL STUDIES USING FRIEND CONTROLS1 W. DANA FLANDERS AND HARLAND AUSTIN In case-control studies, a control is fre- neous levels of the matching factors. Thus, quently selected for each case matched on for example, if matching on friendship at- one or more potentially confounding fac- tains comparability between cases and con- tors such as age, sex, or race. The purpose trols on age, race, and socioeconomic sta- of the matching is the attainment of effi- tus, the bias could still remain among cient control of confounding by these fac- young, white men of similar socioeconomic tors in the analysis (1). Obtaining matched status. Therefore, the bias would not be controls, however, is often costly, and their corrected by the appropriate matched anal- use becomes infeasible as the number of ysis. This selection bias could occur, for matching factors increases (2). Matched example, in a case-control study of occu- controls can be sought from rosters of hos- pation using friend controls, since people 
TI - POSSIBILITY OF SELECTION BIAS IN MATCHED CASE-CONTROL STUDIES USING FRIEND CONTROLS1
JO - American Journal of Epidemiology
DO - 10.1093/oxfordjournals.aje.a114359
DA - 1986-07-01
UR - https://www.deepdyve.com/lp/oxford-university-press/possibility-of-selection-bias-in-matched-case-control-studies-using-I1zyxvY0Er
SP - 150
EP - 153
VL - 124
IS - 1
DP - DeepDyve
ER -