Understanding and minimizing epidemiologic bias in public health research

Abstract

Awareness of potential biases is important for both researchers and policy-makers in public health; for researchers when designing and conducting studies, and for policy-makers when reading study reports and making decisions. This paper explains the meaning and importance of epidemiologic bias in public health and discusses how it arises and what can be done to minimize it. Examples of counting participants in a meeting, to which many policy-makers can relate, are used throughout the paper to illustrate bias in general, random error and systematic error, the effect of sample size, the three main categories of bias (selection, information and confounding), stratification and mathematical modeling.