C5
CalibrationRegression to Mean
Extreme values should be expected to regress
4
Models Tested
1
Confirmatory
0.269
Mean Effect
0.755
Max Effect
Theoretical Context
Theoretical Anchor
Kahneman & Tversky (1973)
Normative Violation
Extreme values should be expected to regress
Cross-Model Comparison
Effect sizes for Regression to Mean across all tested models
Google
Gemini 2.0 Flash
The Consistent Optimist
h = +0.755Confirmatory
OpenAI
GPT-5.3 Instant
The Directive Optimist
h = +0.344
Anthropic
Claude Sonnet 4.6
The Cautious Contrarian
h = -0.220
OpenAI
GPT-5.5
The Invariant
h = +0.196
Statistical Details
Full results with confidence intervals and sample sizes
| Model | n (A) | n (B) | Cohen's h | 95% CI | Status |
|---|---|---|---|---|---|
| Gemini 2.0 Flash | 50 | 50 | +0.7554 | — | Confirmatory |
| GPT-5.3 Instant | 50 | 50 | +0.3437 | — | Exploratory |
| Claude Sonnet 4.6 | 50 | 50 | -0.2195 | — | Exploratory |
| GPT-5.5 | 50 | 50 | +0.1957 | — | Exploratory |