English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

This article is a response to an off-the-record discussion that I had at aninternational meeting of epidemiologists. It centered on a concern, perhapswidely spread, that measurement error adjustment methods can induce positivebias in results of epidemiological studies when there is no true association. Itrace the possible history of this supposition and test it in a simulationstudy of both continuous and binary health outcomes under a classicalmultiplicative measurement error model. A Bayesian measurement adjustmentmethod is used. The main conclusion is that adjustment for the presumedmeasurement error does not 'induce' positive associations, especially if thefocus of the interpretation of the result is taken away from the pointestimate. This is in line with properties of earlier measurement erroradjustment methods introduced to epidemiologists in the 1990s. An heuristicargument is provided to support the generalizability of this observation in theBayesian framework. I find that when there is no true association, positivebias can only be induced by indefensible manipulation of the priors, such thatthey dominate the data. The misconception about bias induced by measurementerror adjustment should be more clearly explained during the training ofepidemiologists to ensure the appropriate (and wider) use of measurement errorcorrection procedures. The simple message that can be derived from this paperis: 'Do not focus on point estimates, but mind the gap between boundaries thatreflect variability in the estimate'. And of course: 'Treat measurement erroras a tractable problem that deserves much more attention than just aqualitative (throw-away) discussion'.

Does adjustment for measurement error induce positive bias if there is no true association?