| name | blind-consider-the-opposite |
| description | Use when evidence seems to confirm an existing belief, when research keeps agreeing with the hypothesis, when interpreting ambiguous data, or when a first explanation feels satisfying. Applies Lord, Lepper & Preston (1984) — the one debiasing instruction shown to work: consider the opposite. |
Consider the Opposite
Overview
Lord, Lepper & Preston (1984, JPSP): telling people to "be unbiased and objective" does nothing — biased assimilation of evidence continues unchanged. But one instruction reliably works: "ask yourself at each step whether you would have made the same high or low evaluations had exactly the same study produced results on the other side of the issue." Generating the opposite case isn't a virtue exercise; it mechanically forces retrieval of evidence and interpretations your confirmation-driven search skipped (Wason, 1960: people test hypotheses by seeking confirming instances by default).
Objectivity is not a stance you adopt; it's a specific counterfactual you compute.
When to Use
- Every source you check agrees with you (search-term bias: you searched for the confirmation)
- Interpreting ambiguous evidence — logs, user feedback, benchmark results — that "fits"
- A diagnosis felt right immediately and subsequent checks kept "verifying" it
- Evaluating a study/article/benchmark whose conclusion you like
The Procedure
- The swap test: same evidence, opposite conclusion — would you accept the methodology then? If you'd reject the method when it disagrees with you, reject it now.
- Argue the reverse for real: write 3 concrete reasons the opposite of your belief could be true. Fewer than 3 with substance = you haven't searched, you've confirmed.
- Re-run the search inverted: if you googled "why X is better than Y", now search "X problems / Y advantages". Search queries inherit your hypothesis.
- Explain the anomalies: list the data points your belief explains worst (not best). A belief audited only against its best-fitting evidence is untested.
Quick Reference
| Symptom | Counter-move |
|---|
| All evidence agrees | Invert the search query |
| Explanation felt right instantly | 3 substantive reasons for the opposite |
| Methodology looks fine (and agrees with you) | Swap test |
| "The data clearly shows…" | Which data fits worst? |
Common Mistakes
- Generating strawman opposites that are easy to dismiss — the opposite case must be the one a smart opponent would make (see blind-steelman).
- Considering the opposite after committing publicly — do it while the belief is still cheap to change.
- Confusing this with indecision. You still conclude; you just conclude after the counterfactual, not instead of it.
