Measurement Errors in Epidemiology

The basis for all epidemiological research is an accurate and precise measurement of exposure. Imprecision in the assessment of an exposure or of confounders can lead to potentially strong biases, which can be either towards or away from the null. This chapter illustrates the importance of good measures of exposure, demonstrates the potential impact of imprecise or inaccurate measurement, and describes statistical methods that can be used to minimise and correct for measurement error bias. The methods are illustrated using a real example from the field of diet and cancer.