Bias reduction involves identifying possible sources of bias and taking steps to reduce it. This is one way of improving the validity of an evaluation.
Types of bias include,
- Selection bias: For example, project sites chosen for field visits are more easily accessible, even though they are not representative or treated as a purposeful sample or only participants who completed a program are interviewed, ignoring those who dropped out.
- Respondent bias: For example, participants in an intervention might provide ratings that are more positive than warranted because they fear a negative evaluation will lead to an intervention being terminated (deliberate respondent bias) or because of the placebo effect of being involved in a program (unintended respondent bias).
- Evaluator bias: For example, evaluators rating activities or conditions might provide ratings that are more positive than warranted because of confirmation bias which leads them to notice events and conditions that match their theory of change (unintended evaluator bias).
Examples
The QuIP approach (Qualitative Impact Assessment Protocol) includes a number of strategies to reduce pro-project and confirmation bias including where possible the use of a double-blind design where neither the team conducting the interviews nor the respondents are aware of who has commissioned the evaluation or which project is being evaluated.
Last updated:
This page is a Stub (a minimal version of a page). You can help expand it. Contact Us to recommend resources or volunteer to expand the description.
'Bias reduction' is referenced in:
Approach
Framework/Guide
- Rainbow Framework :
Method