A3

Framing & Reference

Certainty Effect

Risk preferences should be consistent across probability frames

4
Models Tested
0
Confirmatory
-0.167
Mean Effect
0.446
Max Effect

Theoretical Context

Theoretical Anchor

Kahneman & Tversky (1979)

Normative Violation

Risk preferences should be consistent across probability frames

Cross-Model Comparison

Effect sizes for Certainty Effect across all tested models

Google
Gemini 2.0 Flash

The Consistent Optimist

h = -0.446
OpenAI
GPT-5.2

The Steady Traditionalist

h = -0.213
OpenAI
GPT-5.3 Instant

The Directive Optimist

h = -0.087
OpenAI
GPT-5.4 Thinking

The Deliberative Calibrator

h = +0.080

Statistical Details

Full results with confidence intervals and sample sizes

Modeln (A)n (B)Cohen's h95% CIStatus
Gemini 2.0 Flash5050-0.4460Exploratory
GPT-5.25050-0.2132Exploratory
GPT-5.3 Instant5050-0.0873Exploratory
GPT-5.4 Thinking5050+0.0800Exploratory