Table 1 Bias in trials due to measurement: a preliminary framework

Our preliminary framework consists of three elements:

(A) What sorts of bias can arise from measurement reactivity (and what are the relationships to the existing well-known forms of bias)?

 1. Main effects of measurement on trial outcome

 2. Where there is an interaction between measurement and trial-arm status on outcome

 3. Where measurement results in study dropout

(B) How does measurement produce changes in people (i.e. what are the mechanisms)?

 1. Measurement changes the performance of that behaviour or reports of performance of that behaviour

 2. Measurement changes emotional states or reports of emotional state

 3. Measurement changes questionnaire or study completion rate

(C) What are the characteristics of measurement, people and context that can lead to or moderate the risk of such biases (i.e. when might measurement reactivity be anticipated)?

 1. Features of measurement that produce reactivity

 2. Features of outcome measurement (may be the same as above)

 3. Features of participants being measured

 4. Other features of context surrounding measurement or trial not captured in other categories