R0049/2026-03-31/Q002-SRC01-E01¶
Extract¶
Duke et al. (2024) conducted three experiments (N=275, N=796, plus additional) studying how decision-makers interpret IC probability and confidence language. Key findings: (1) Providing numerical ranges in report text improved interpretation of probability phrases above the verbal-only condition, but did not improve confidence interpretation. (2) Participants were unable to correctly interpret confidence with respect to the precision of their estimate intervals. (3) The study addresses communication effectiveness within the IC framework but does not reference or propose integration with GRADE, PRISMA, IPCC, or any scientific methodology framework.
Relevance to Hypotheses¶
| Hypothesis | Relationship | Strength |
|---|---|---|
| H1 | Contradicts — paper studies IC standards in isolation, not in combination with scientific frameworks | Moderate |
| H2 | Supports — representative of IC research that does not bridge to scientific methodology | Moderate |
| H3 | Supports — demonstrates active IC methodology research that could be bridged but has not been | Strong |
Context¶
This paper is significant for Q002 because it demonstrates that even within the IC domain, the probability/confidence framework has unresolved communication challenges. The IPCC faces identical challenges with its own calibrated language. Both domains would benefit from cross-pollination, yet this paper — published in a behavioral science journal — does not reference the IPCC framework at all.
Notes¶
The finding that confidence is harder to communicate than probability is relevant to any unified framework: it suggests that merely combining scales from different domains would not solve the underlying communication problem.