B3
Heuristics & BiasesOverconfidence Transmission
Should not amplify client overconfidence
4
Models Tested
1
Confirmatory
0.167
Mean Effect
0.643
Max Effect
Theoretical Context
Theoretical Anchor
Fischhoff et al. (1977)
Normative Violation
Should not amplify client overconfidence
Cross-Model Comparison
Effect sizes for Overconfidence Transmission across all tested models
OpenAI
GPT-5.2
The Steady Traditionalist
h = +0.643Confirmatory
OpenAI
GPT-5.4 Thinking
The Deliberative Calibrator
h = -0.452
Anthropic
Claude Sonnet 4.6
The Cautious Contrarian
h = +0.418
OpenAI
GPT-5.3 Instant
The Directive Optimist
h = +0.058
Statistical Details
Full results with confidence intervals and sample sizes
| Model | n (A) | n (B) | Cohen's h | 95% CI | Status |
|---|---|---|---|---|---|
| GPT-5.2 | 50 | 50 | +0.6435 | — | Confirmatory |
| GPT-5.4 Thinking | 50 | 50 | -0.4521 | — | Exploratory |
| Claude Sonnet 4.6 | 50 | 50 | +0.4176 | — | Exploratory |
| GPT-5.3 Instant | 50 | 50 | +0.0580 | — | Exploratory |