Table 6.

Proportion correct guesses in model simulations of Yu/Smith experiments vs. actual reported values. Fazly et al. and modified Fazly et al. models guessed via sampling with meaning probabilities; variants which guess via choosing the maximally probable candidate perform at ceiling on all conditions. Significant differences determined via pairwise comparison of predictor p-values under a binomial mixed effects regression model with condition (4×4, etc.) as a predictor of correctness with random subject and item effects.

	4×4	3×3	2×2	significant differences	CI overlaps w/ observed
Propose/Verify	0.54 (95% CI = 0.43–0.63)	0.63 (95% CI = 0.54–0.72)	0.76 (95% CI = 0.68–0.84)	4×4 / 2×2 3×3 / 2×2 4×4 / 3×3	yes / yes / yes
Fazly et al. 2010	0.98 (95% CI = 0.94–1.00)	0.98 (95% CI = 0.94–1.00)	0.99 (95% CI = 0.98–1.00)	4×4 / 2×2 3×3 / 2×2	no / no / yes
Modified Fazly	0.96 (95% CI = 0.95–0.97)	0.97 (95% CI = 0.96–0.98)	0.99 (95% CI = 0.99–1.00)	4×4 / 2×2 3×3 / 2×2	no / no / yes
Pursuit	0.71 (95% CI = 0.62–0.80)	0.84 (95% CI = 0.76–0.91)	0.96 (95% CI = 0.92–0.99)	4×4 / 2×2 3×3 / 2×2 4×4 / 3×3	yes / yes / yes
Reported	0.53 (95% CI = 0.37–0.69)	0.76 (95% CI = 0.62–0.90)	0.89 (95% CI = 0.79–0.99)