Research Bias Vocabulary

Social desirability bias occurs when participants give answers they believe are socially acceptable or expected rather than their true behaviour. For example, participants may say they always read terms and conditions, when in reality they never do. Mitigation techniques include: asking about past behaviour rather than intentions, framing questions as 'what do most people do?', and prioritising behavioural observation over self-report.

Confirmation bias in analysis means the researcher (consciously or unconsciously) seeks out, emphasises, and remembers evidence that supports what they already believed, while minimising evidence that contradicts it. Mitigations: involve multiple researchers in analysis, use structured note-taking, conduct affinity mapping with the whole team, and actively seek disconfirming evidence.

Facilitator bias occurs when the researcher conducting the session inadvertently influences participant responses — through leading questions ('did you find the checkout confusing?'), nodding or reacting to certain answers, or their tone of voice. Neutral facilitation requires open-ended questions, neutral probes ('can you tell me more?'), and trained awareness of non-verbal reactions.

Survivorship bias in user research means you only hear from users who are still using the product — missing the perspective of people who tried and abandoned it. This skews findings towards a more positive view of the product. To counter it, recruit churned users or look at data from users who dropped off during onboarding, not just those who successfully completed the flow.

Recency bias in synthesis means the researcher gives disproportionate weight to findings from the last few sessions because they're most fresh in memory. Sessions from weeks earlier fade. Triangulation (systematically reviewing all session notes and recordings, not just recalling from memory) and structured analysis methods like affinity mapping combat recency bias by treating all sessions equally in the synthesis.