Bookmarks
Assessing and taking account of measurement error using repeat measurements of exposure and outcome variables
Safe People
University of Bristol
Academic Institute
Rosie CornishDr Willy Steven KamgangProfessor Kate TillingProfessor Deborah Lawlor
Safe Projects
B3764
Most measures in epidemiology are subject to error; despite this, the impacts of measurement error are often overlooked. Measurement error in an outcome variable will generally result in a loss of precision whereas measurement error in an exposure variable (or other covariate) will generally lead to bias in regression coefficients. In this project we will quantify measurement error in systolic blood pressure, weight and height using repeated measurements made on a subsample of individuals attending ALSPAC clinics. We will then assess the impact of measurement error in these variables on regression coefficients and their standard errors.
Will help other users of ALSPAC data understand the extent of measurement error in these key variables and its likely impact on analyses.
04/05/2021