You get:
- smoothing over disagreements to tell a clean story
- citing “mixed findings” without explaining why
- missing the most interesting questions (why do studies disagree?)
- reviewers pointing out contradictions you should have addressed
- building on findings that later studies contradict
But contradictions have causes:
- methodological differences (experiment vs. survey, lab vs. field)
- sample differences (students vs. professionals, culture A vs. B)
- measurement differences (how constructs are operationalized)
- temporal differences (findings change over time)
- publication bias (significant findings publish, nulls don’t)
Without contradiction hunting, you miss the real science.
This prompt identifies and explains disagreements across papers.
Assume the role of a meta-research scientist who hunts contradictions. Your task is to identify and explain disagreements between papers. Generate: 1. CONTRADICTION SUMMARY TABLE | Finding | Paper A | Paper B | Direction of disagreement | 2. POSSIBLE EXPLANATIONS (ranked by likelihood) - Methodology (different designs, measures, analyses) - Sample (different populations, settings, time periods) - Definition (different constructs or operationalizations) - Statistical (power, multiple comparisons, p-hacking) - Publication bias (one finding is true, the other is file-drawer) 3. MODERATOR HYPOTHESES - What variable might explain the disagreement? - How would you test it? 4. WHAT'S NEEDED TO RESOLVE - Replication study (with what design?) - Meta-analysis (with what search strategy?) - Theoretical integration (when is each finding true?) 5. IMPLICATION FOR YOUR WORK - Which finding do you trust more? Why? - How will you frame this debate? INPUTS: Paper A (full text or detailed summary): [PASTE] Paper B: [PASTE] Your research question: [PASTE] Field/discipline: [PASTE] RULES: - Assume both papers are honestly conducted (no accusations of fraud without evidence) - Look for "boundary conditions" — contradictions often reveal when a finding holds - Distinguish between statistical and practical significance - Note if the contradiction is actually a replication failure
- Use this when your literature review finds “mixed results” — that’s an opportunity, not a problem.
- Run this on the two most-cited papers in your field that disagree.
- Use the “moderator hypotheses” to generate your own research ideas.
- Frame your research question as “when does X happen?” rather than “does X happen?” — that’s how you publish.
- Cite both sides of the contradiction — reviewers will respect your even-handedness.
Paper A:
“Eysenck & Calvo (1992). Anxiety and performance: The processing efficiency theory. Found anxiety impairs performance on complex tasks but not simple tasks. N=80, lab experiment using math problems.”
Paper B:
“Beilock & Carr (2005). When high-powered people fail. Found high-pressure conditions impair performance on well-learned tasks (choking). N=60, lab experiment with golf putting.”
Your research question:
“When does anxiety help vs. hurt performance?”
Field/discipline:
Sport Psychology / Cognitive Performance
This framework improves outcomes by forcing:
- contradiction table (visual, clear, honest)
- ranked explanations (hypotheses to test)
- moderator identification (when is each finding true?)
- resolution path (what research is needed)
- implication framing (how to cite responsibly)
Great contradiction hunting doesn’t resolve debates — it reveals the boundary conditions that make both findings true.
Build Better AI Systems
Subscribe for advanced prompt engineering, AI coding tools, debugging frameworks, and practical strategies for developers and engineers.