Confirmation bias makes hypothesis-consistent excerpts easier to retain
Aliases: confirmation bias · selective recall · hypothesis-consistent bias
What it is
Confirmation bias makes researchers more likely to seek, code, remember, or retell material consistent with an existing hypothesis while demanding stronger evidence from counterexamples. It need not involve deliberate distortion; it can enter through probes, field notes, transcript selection, theme naming, and reporting.
Why it happens
Prior hypotheses provide attentional and interpretive frames. Consistent excerpts fit a coherent account, whereas contradictions are easily treated as exceptions or participant error. Memory-based debriefs amplify the asymmetry because vivid expected events are more retrievable. Team consensus and decision pressure can reward a single clean narrative.
Studying it
Retain recordings, transcripts, timestamped notes, and coding versions, then compare immediate impressions with the full record. Specify rival accounts and disconfirming evidence in advance. Use negative-case analysis, reflexive memos, and independent coding of a subset; disagreement exposes interpretive choices rather than proving truth through agreement. Trace report quotations back to context.
Where it stops holding
Prior theory is not itself an error; theory-driven analysis requires a frame, and open coding is not automatically neutral. A second researcher who shares the same assumptions and incentives may reproduce the bias. Member checking can correct facts but does not make participants the sole authority over interpretation.
Applying it
- State the main hypothesis, a rival account, and evidence that would change the conclusion.
- Separate observation from interpretation and retain counterexamples with context links.
- Have someone outside moderation audit raw material for evidence against the main story.
- Test whether added negative cases change theme boundaries, not merely how many supporting quotes remain.
Related
- Same group: Q1.14.2 Closed questions constrain response space · Q1.14.3 Stakeholder presence effects · Q1.14.4 Separating moderation and note-taking
- Adjacent: Q1.07 Researcher effects and interpretation · Q2.08 Qualitative coding and thematic analysis
- Search terms:
confirmation bias·negative case analysis·reflexivity