Random measurement error may affect an exposure variable (risk factor). A number of steps can be taken to reduce the bias that this introduces. These are best undertaken during the study design stage.
Random measurement error in an exposure variable tends to bias the estimate of the regression slope coefficient downwards the null. This is termed regression dilution or attenuation bias.
Ways to reduce the error:
(1) reduce the random measurement error (with a more accurate and precise measurement device or with more careful technique)
(2) obtain repeated measurements of the exposure variable from each subject
Increasing the sample size will not correct the problem. It will make the estimates "more precisely wrong".
To read more or access our algorithms and calculators, please log in or register.
