Systematic distortion of results or findings from the true state of affairs, or any of several varieties of processes leading to systematic distortion. In everyday usage, “bias” often implies the presence of emotional and/or political prejudices that influence conclusions and decisions. These prejudices may occur in public health sciences, for instance in epidemiology as a consequence of a conflict of interest, but any of several other factors often account for the presence of bias. Epidemiologists have identified more than 20 varieties of bias, but several of these are variations of fundamental flaws in design, methods, procedures, or logical reasoning. The most common types include selection bias, observer bias, bias due to confounding, and bias due to errors of logical reasoning. Several of the biases encountered in epidemiological and social science investigations may include more than one of these unless investigators exercise great care with methods, design, and procedures. Other causes are systematic variation of measurements and/or statistical summary measures (means, rates, measures of association, etc.) and flaws in study design, data collection and analysis, or interpretation of the evidence.
Subjects: Medicine and Health.