Paper Replica Figure 1: Logistic Regression CIs & Paper Table Replicas

Methodology & Presentation

This report presents Paper Replica Figure 1 logistic regression plots alongside color-coded replica tables formatted identically to Table 1 (Numbers & Names) and Table 8 (Cities) from the paper:

Figure 1 Error Bar Confidence Level Controls
Currently Displaying: Standard 95% Confidence Intervals (\(z = 1.96\))

1. Main Experiments: Numbers & Names

Paper Replica Figure 1: Numbers & Names Logistic Regression Plots

Paper Replica Figure 1 for Numbers
Paper Replica Figure 1 for Names

Paper Replica Table 1: Numbers & Names Win Fractions (Color-coded by distance from 0.5)

ComparableMetricDisplayGemini 2.5 FlashGemini 3 FlashClaude Haiku 4.5GPT-5 MiniQwen 3.5 35BRandom Pref Order 1Random Pref Order 2Random Pref Order 3
NumbersHigher0.70***0.78***0.79***0.76***0.72***0.480.550.41*
NumbersPrime (28%)0.64**0.600.620.73***0.72***0.500.450.48
NumbersHigher Order Mag (14%)0.580.74***0.75***0.600.590.540.540.32
NumbersAsc Digit Order (31%)0.440.400.400.410.430.390.390.50
NumbersPalindrome (17%)0.590.530.500.560.630.380.32*0.59
NumbersHigher Sum Digits0.570.62***0.64***0.62***0.570.470.550.48
NumbersMult 5 (35%)0.510.450.410.400.460.560.610.56
NumbersMult 20.460.450.440.410.440.39*0.510.50
NumbersMult 10 (22%)0.520.430.370.370.480.480.610.48
NumbersDesc Digit Order (42%)0.540.520.520.550.540.550.560.59
NumbersHas Repeat Digits (41%)0.550.540.530.510.530.430.37*0.52
NumbersFirst Presented0.59***0.500.63***0.48***0.57***0.500.500.50
NamesHigher Rare Letter Frac (6%)0.820.97***0.680.94***0.780.500.330.42
NamesStarting Vowel (31%)0.76***0.74***0.600.74***0.75***0.480.400.45
NamesHigher Frac of 2020 Babies0.72***0.75***0.65***0.67***0.75***0.510.470.50
NamesHigher Frac of 2000 Babies0.69***0.71***0.62***0.63***0.70***0.480.510.55
NamesHigher Frac of 1960 Babies0.40**0.34***0.42***0.34***0.37***0.480.490.49
NamesHigher Bouba Frac0.60*0.590.530.62**0.60*0.58*0.480.52
NamesHigher Vowel Frac0.62***0.62**0.540.66***0.61**0.550.500.47
NamesHigher Syllable Heur0.560.64*0.500.64*0.570.590.490.48
NamesEnding Vowel (41%)0.550.560.460.65*0.500.550.460.55
NamesFirst Alphabetically0.59*0.540.500.540.550.510.470.52
NamesLonger Length0.480.560.510.530.500.560.560.49
NamesHigher Unique Char Frac0.480.480.510.450.510.510.470.48
NamesHigher Census Freq0.520.490.510.450.500.480.59*0.47
NamesHigher Frac of 1980 Babies0.540.500.510.440.520.480.550.48
NamesFirst Presented0.52**0.510.82***0.510.57***0.500.500.50
Regression Coefficient & 95% CI Summaries (Numbers & Names)

Values shown: Probability - 0.5 [95% CI Lower, 95% CI Upper] Significance Tag

Numbers Summary

Model Gemini 2.5 Flash Gemini 3 Flash Claude Haiku 4.5 GPT-5 Mini Qwen 3.5 35B Random Pref Order 1 Random Pref Order 2 Random Pref Order 3
Metric
Asc Digit Order +0.122 [+0.089, +0.154] *** +0.122 [+0.088, +0.154] *** +0.084 [+0.048, +0.119] *** +0.108 [+0.074, +0.141] *** +0.128 [+0.095, +0.160] *** -0.153 [-0.182, -0.124] *** -0.130 [-0.160, -0.100] *** -0.009 [-0.041, +0.023] n.s.
Desc Digit Order -0.107 [-0.138, -0.075] *** -0.105 [-0.137, -0.072] *** -0.006 [-0.041, +0.030] n.s. +0.059 [+0.025, +0.092] ** -0.072 [-0.105, -0.039] *** +0.114 [+0.083, +0.143] *** +0.223 [+0.197, +0.248] *** +0.243 [+0.219, +0.267] ***
First Presented -0.372 [-0.403, -0.333] *** -0.436 [-0.453, -0.413] *** -0.402 [-0.428, -0.368] *** -0.413 [-0.435, -0.384] *** -0.430 [-0.448, -0.406] *** +0.386 [+0.353, +0.413] *** -0.366 [-0.397, -0.327] *** +0.133 [+0.061, +0.199] **
Has Repeat Digits +0.018 [-0.016, +0.051] n.s. -0.093 [-0.126, -0.059] *** -0.136 [-0.169, -0.101] *** -0.223 [-0.251, -0.194] *** -0.112 [-0.144, -0.078] *** -0.198 [-0.224, -0.169] *** -0.298 [-0.319, -0.276] *** -0.079 [-0.109, -0.047] ***
Higher +0.490 [+0.487, +0.492] *** +0.498 [+0.497, +0.498] *** +0.497 [+0.497, +0.498] *** +0.496 [+0.494, +0.497] *** +0.493 [+0.491, +0.495] *** +0.165 [+0.113, +0.214] *** -0.227 [-0.271, -0.179] *** -0.269 [-0.307, -0.225] ***
Higher Order Mag -0.411 [-0.439, -0.373] *** -0.160 [-0.245, -0.062] * +0.059 [-0.049, +0.161] n.s. -0.379 [-0.417, -0.327] *** -0.399 [-0.431, -0.354] *** +0.352 [+0.298, +0.394] *** +0.402 [+0.362, +0.432] *** -0.272 [-0.332, -0.197] ***
Higher Sum Digits -0.308 [-0.348, -0.261] *** -0.344 [-0.379, -0.300] *** -0.237 [-0.293, -0.174] *** -0.265 [-0.314, -0.209] *** -0.323 [-0.361, -0.277] *** -0.419 [-0.436, -0.396] *** +0.440 [+0.423, +0.454] *** +0.037 [-0.029, +0.101] n.s.
Mult 10 +0.041 [-0.013, +0.095] n.s. -0.073 [-0.127, -0.017] n.s. -0.189 [-0.238, -0.136] *** -0.182 [-0.229, -0.131] *** +0.029 [-0.026, +0.083] n.s. -0.283 [-0.316, -0.245] *** +0.132 [+0.082, +0.180] *** -0.356 [-0.380, -0.329] ***
Mult 2 +0.009 [-0.019, +0.036] n.s. -0.048 [-0.076, -0.019] * -0.051 [-0.080, -0.021] ** -0.069 [-0.097, -0.041] *** -0.006 [-0.034, +0.022] n.s. -0.265 [-0.284, -0.245] *** -0.028 [-0.055, -0.002] n.s. +0.055 [+0.030, +0.081] ***
Mult 5 +0.159 [+0.123, +0.194] *** +0.045 [+0.004, +0.085] n.s. -0.037 [-0.079, +0.006] n.s. -0.029 [-0.070, +0.012] n.s. +0.089 [+0.049, +0.127] *** +0.074 [+0.037, +0.111] ** +0.144 [+0.108, +0.178] *** +0.249 [+0.219, +0.276] ***
Palindrome +0.259 [+0.222, +0.293] *** +0.214 [+0.172, +0.253] *** +0.154 [+0.106, +0.199] *** +0.320 [+0.288, +0.348] *** +0.378 [+0.354, +0.398] *** -0.075 [-0.119, -0.029] * -0.007 [-0.054, +0.040] n.s. +0.260 [+0.224, +0.292] ***
Prime +0.340 [+0.320, +0.358] *** +0.270 [+0.243, +0.295] *** +0.312 [+0.287, +0.334] *** +0.400 [+0.385, +0.413] *** +0.416 [+0.404, +0.427] *** -0.156 [-0.186, -0.126] *** -0.079 [-0.111, -0.046] *** -0.051 [-0.084, -0.019] *

Names Summary

Model Gemini 2.5 Flash Gemini 3 Flash Claude Haiku 4.5 GPT-5 Mini Qwen 3.5 35B Random Pref Order 1 Random Pref Order 2 Random Pref Order 3
Metric
Ending Vowel +0.238 [+0.210, +0.264] *** +0.196 [+0.163, +0.227] *** -0.132 [-0.172, -0.090] *** +0.313 [+0.290, +0.334] *** +0.035 [-0.000, +0.070] n.s. +0.001 [-0.028, +0.030] n.s. +0.040 [+0.010, +0.070] n.s. +0.224 [+0.199, +0.248] ***
First Alphabetically +0.142 [+0.124, +0.159] *** +0.036 [+0.015, +0.057] * -0.044 [-0.069, -0.019] ** +0.031 [+0.011, +0.051] * +0.043 [+0.023, +0.062] *** +0.069 [+0.052, +0.085] *** -0.032 [-0.049, -0.015] ** +0.075 [+0.058, +0.092] ***
First Presented -0.499 [-0.500, -0.499] *** -0.500 [-0.500, -0.500] *** -0.312 [-0.379, -0.221] *** -0.500 [-0.500, -0.500] *** -0.498 [-0.499, -0.497] *** -0.284 [-0.337, -0.220] *** +0.398 [+0.361, +0.426] *** +0.402 [+0.365, +0.429] ***
Higher Bouba Frac +0.345 [+0.314, +0.372] *** +0.372 [+0.343, +0.397] *** -0.209 [-0.266, -0.145] *** +0.408 [+0.388, +0.426] *** +0.246 [+0.201, +0.287] *** +0.265 [+0.228, +0.298] *** +0.016 [-0.034, +0.065] n.s. +0.324 [+0.293, +0.352] ***
Higher Census Freq +0.055 [-0.160, +0.251] n.s. -0.465 [-0.486, -0.418] *** +0.317 [+0.105, +0.429] n.s. -0.404 [-0.458, -0.296] *** -0.350 [-0.431, -0.202] ** +0.438 [+0.376, +0.470] *** +0.374 [+0.264, +0.437] *** -0.355 [-0.426, -0.235] ***
Higher Frac of 1960 Babies -0.378 [-0.434, -0.286] *** -0.126 [-0.270, +0.045] n.s. -0.495 [-0.498, -0.489] *** -0.072 [-0.224, +0.095] n.s. -0.346 [-0.415, -0.238] *** -0.432 [-0.461, -0.385] *** -0.468 [-0.482, -0.445] *** -0.434 [-0.462, -0.388] ***
Higher Frac of 1980 Babies +0.342 [+0.283, +0.387] *** +0.154 [+0.055, +0.242] * +0.029 [-0.093, +0.148] n.s. -0.387 [-0.422, -0.340] *** +0.317 [+0.251, +0.368] *** +0.035 [-0.049, +0.117] n.s. +0.318 [+0.260, +0.364] *** +0.437 [+0.414, +0.455] ***
Higher Frac of 2000 Babies +0.277 [+0.219, +0.325] *** +0.370 [+0.325, +0.405] *** +0.333 [+0.270, +0.381] *** +0.279 [+0.217, +0.331] *** +0.294 [+0.239, +0.341] *** -0.356 [-0.386, -0.320] *** +0.367 [+0.331, +0.396] *** +0.354 [+0.316, +0.386] ***
Higher Frac of 2020 Babies +0.499 [+0.498, +0.499] *** +0.500 [+0.500, +0.500] *** +0.500 [+0.500, +0.500] *** +0.500 [+0.500, +0.500] *** +0.500 [+0.500, +0.500] *** +0.321 [+0.272, +0.362] *** -0.396 [-0.422, -0.363] *** -0.231 [-0.287, -0.166] ***
Higher Rare Letter Frac +0.482 [+0.469, +0.490] *** +0.499 [+0.499, +0.500] *** +0.415 [+0.354, +0.452] *** +0.497 [+0.493, +0.498] *** +0.358 [+0.288, +0.408] *** -0.092 [-0.187, +0.009] n.s. -0.051 [-0.150, +0.053] n.s. -0.068 [-0.167, +0.036] n.s.
Higher Syllable Heur -0.105 [-0.182, -0.023] n.s. -0.054 [-0.143, +0.039] n.s. -0.478 [-0.486, -0.466] *** +0.303 [+0.242, +0.352] *** -0.052 [-0.135, +0.034] n.s. +0.388 [+0.355, +0.414] *** -0.402 [-0.426, -0.371] *** +0.025 [-0.050, +0.099] n.s.
Higher Unique Char Frac -0.135 [-0.191, -0.074] *** -0.154 [-0.212, -0.091] *** -0.316 [-0.361, -0.262] *** -0.154 [-0.209, -0.094] *** -0.023 [-0.087, +0.042] n.s. +0.084 [+0.030, +0.136] * -0.266 [-0.304, -0.223] *** -0.236 [-0.278, -0.188] ***
Higher Vowel Frac +0.206 [+0.151, +0.255] *** +0.395 [+0.366, +0.419] *** +0.352 [+0.305, +0.389] *** +0.395 [+0.367, +0.417] *** +0.392 [+0.364, +0.415] *** +0.148 [+0.097, +0.196] *** -0.313 [-0.345, -0.277] *** -0.357 [-0.383, -0.327] ***
Longer Length -0.079 [-0.154, -0.001] n.s. +0.451 [+0.432, +0.465] *** +0.408 [+0.367, +0.438] *** +0.171 [+0.094, +0.240] *** +0.080 [-0.000, +0.156] n.s. -0.067 [-0.133, +0.002] n.s. +0.256 [+0.199, +0.305] *** -0.393 [-0.418, -0.363] ***
Starting Vowel +0.073 [+0.029, +0.116] * -0.013 [-0.061, +0.036] n.s. -0.000 [-0.059, +0.059] n.s. -0.029 [-0.076, +0.018] n.s. +0.027 [-0.019, +0.072] n.s. -0.274 [-0.300, -0.245] *** -0.057 [-0.096, -0.016] n.s. -0.265 [-0.292, -0.234] ***
Detailed Model-by-Model Regression Tables & Bonferroni p-values (Numbers & Names)

Dataset: Numbers | Model: Gemini 2.5 Flash

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
Higher 4.6123 0.1308 35.2557 0.0000 0.0000 *** 0.4902 0.4873 0.4924 0.4858 0.4932 [4.356, 4.869] [4.237, 4.987]
Has Repeat Digits 0.0705 0.0695 1.0141 0.3105 1.0000 n.s. 0.0176 -0.0164 0.0515 -0.0321 0.0670 [-0.066, 0.207] [-0.129, 0.270]
Palindrome 1.1492 0.0996 11.5426 0.0000 0.0000 *** 0.2594 0.2219 0.2932 0.2035 0.3076 [0.954, 1.344] [0.864, 1.434]
Prime 1.6587 0.0735 22.5671 0.0000 0.0000 *** 0.3401 0.3197 0.3585 0.3097 0.3664 [1.515, 1.803] [1.448, 1.869]
Higher Sum Digits -1.4365 0.1434 -10.0174 0.0000 0.0000 *** -0.3079 -0.3478 -0.2605 -0.3638 -0.2361 [-1.718, -1.155] [-1.847, -1.026]
Asc Digit Order 0.4982 0.0699 7.1312 0.0000 0.0000 *** 0.1220 0.0894 0.1537 0.0740 0.1678 [0.361, 0.635] [0.298, 0.698]
Desc Digit Order -0.4336 0.0673 -6.4408 0.0000 0.0000 *** -0.1067 -0.1377 -0.0748 -0.1517 -0.0599 [-0.566, -0.302] [-0.626, -0.241]
Higher Order Mag -2.3320 0.2052 -11.3637 0.0000 0.0000 *** -0.4115 -0.4390 -0.3732 -0.4488 -0.3512 [-2.734, -1.930] [-2.920, -1.744]
Mult 2 0.0342 0.0558 0.6132 0.5397 1.0000 n.s. 0.0085 -0.0188 0.0358 -0.0313 0.0483 [-0.075, 0.143] [-0.126, 0.194]
Mult 5 0.6604 0.0802 8.2395 0.0000 0.0000 *** 0.1594 0.1232 0.1937 0.1061 0.2089 [0.503, 0.817] [0.431, 0.890]
Mult 10 0.1661 0.1109 1.4980 0.1341 1.0000 n.s. 0.0414 -0.0128 0.0947 -0.0378 0.1187 [-0.051, 0.384] [-0.152, 0.484]
First Presented -1.9182 0.1574 -12.1852 0.0000 0.0000 *** -0.3719 -0.4026 -0.3334 -0.4144 -0.3126 [-2.227, -1.610] [-2.369, -1.467]

Dataset: Numbers | Model: Gemini 3 Flash

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
Higher 6.1507 0.1411 43.6004 0.0000 0.0000 *** 0.4979 0.4972 0.4984 0.4968 0.4986 [5.874, 6.427] [5.747, 6.555]
Has Repeat Digits -0.3751 0.0712 -5.2704 0.0000 0.0000 *** -0.0927 -0.1259 -0.0586 -0.1408 -0.0427 [-0.515, -0.236] [-0.579, -0.171]
Palindrome 0.9158 0.1009 9.0718 0.0000 0.0000 *** 0.2142 0.1721 0.2528 0.1517 0.2694 [0.718, 1.114] [0.627, 1.205]
Prime 1.2090 0.0751 16.1027 0.0000 0.0000 *** 0.2701 0.2430 0.2951 0.2299 0.3060 [1.062, 1.356] [0.994, 1.424]
Higher Sum Digits -1.6857 0.1516 -11.1170 0.0000 0.0000 *** -0.3437 -0.3790 -0.3004 -0.3929 -0.2775 [-1.983, -1.389] [-2.120, -1.251]
Asc Digit Order 0.4966 0.0720 6.8944 0.0000 0.0000 *** 0.1217 0.0879 0.1543 0.0721 0.1689 [0.355, 0.638] [0.290, 0.703]
Desc Digit Order -0.4260 0.0697 -6.1083 0.0000 0.0000 *** -0.1049 -0.1371 -0.0718 -0.1515 -0.0563 [-0.563, -0.289] [-0.626, -0.226]
Higher Order Mag -0.6621 0.2098 -3.1551 0.0016 0.0192 * -0.1597 -0.2452 -0.0624 -0.2796 -0.0152 [-1.073, -0.251] [-1.263, -0.061]
Mult 2 -0.1914 0.0578 -3.3087 0.0009 0.0112 * -0.0477 -0.0756 -0.0195 -0.0883 -0.0064 [-0.305, -0.078] [-0.357, -0.026]
Mult 5 0.1786 0.0831 2.1493 0.0316 0.3793 n.s. 0.0445 0.0039 0.0845 -0.0149 0.1027 [0.016, 0.341] [-0.059, 0.417]
Mult 10 -0.2937 0.1151 -2.5505 0.0108 0.1291 n.s. -0.0729 -0.1270 -0.0170 -0.1510 0.0091 [-0.519, -0.068] [-0.624, 0.036]
First Presented -2.6783 0.1641 -16.3202 0.0000 0.0000 *** -0.4357 -0.4526 -0.4135 -0.4589 -0.4010 [-3.000, -2.357] [-3.149, -2.208]

Dataset: Numbers | Model: Claude Haiku 4.5

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
Higher 5.9598 0.1452 41.0348 0.0000 0.0000 *** 0.4974 0.4966 0.4981 0.4961 0.4983 [5.675, 6.245] [5.544, 6.376]
Has Repeat Digits -0.5574 0.0749 -7.4460 0.0000 0.0000 *** -0.1359 -0.1691 -0.1013 -0.1839 -0.0849 [-0.704, -0.411] [-0.772, -0.343]
Palindrome 0.6372 0.1060 6.0133 0.0000 0.0000 *** 0.1541 0.1058 0.1995 0.0826 0.2193 [0.430, 0.845] [0.334, 0.941]
Prime 1.4604 0.0798 18.3092 0.0000 0.0000 *** 0.3116 0.2865 0.3343 0.2741 0.3441 [1.304, 1.617] [1.232, 1.689]
Higher Sum Digits -1.0329 0.1570 -6.5776 0.0000 0.0000 *** -0.2375 -0.2926 -0.1737 -0.3150 -0.1417 [-1.341, -0.725] [-1.483, -0.583]
Asc Digit Order 0.3381 0.0752 4.4970 0.0000 0.0001 *** 0.0837 0.0475 0.1190 0.0306 0.1350 [0.191, 0.485] [0.123, 0.554]
Desc Digit Order -0.0224 0.0734 -0.3050 0.7604 1.0000 n.s. -0.0056 -0.0415 0.0304 -0.0579 0.0469 [-0.166, 0.122] [-0.233, 0.188]
Higher Order Mag 0.2353 0.2198 1.0704 0.2844 1.0000 n.s. 0.0586 -0.0487 0.1606 -0.0974 0.2037 [-0.196, 0.666] [-0.395, 0.865]
Mult 2 -0.2043 0.0609 -3.3530 0.0008 0.0096 ** -0.0509 -0.0802 -0.0212 -0.0936 -0.0074 [-0.324, -0.085] [-0.379, -0.030]
Mult 5 -0.1469 0.0878 -1.6731 0.0943 1.0000 n.s. -0.0367 -0.0791 0.0063 -0.0983 0.0261 [-0.319, 0.025] [-0.399, 0.105]
Mult 10 -0.7951 0.1214 -6.5511 0.0000 0.0000 *** -0.1889 -0.2375 -0.1358 -0.2582 -0.1100 [-1.033, -0.557] [-1.143, -0.447]
First Presented -2.2196 0.1721 -12.8987 0.0000 0.0000 *** -0.4020 -0.4280 -0.3679 -0.4378 -0.3490 [-2.557, -1.882] [-2.713, -1.727]

Dataset: Numbers | Model: GPT-5 Mini

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
Higher 5.4123 0.1376 39.3378 0.0000 0.0000 *** 0.4956 0.4942 0.4966 0.4934 0.4970 [5.143, 5.682] [5.018, 5.807]
Has Repeat Digits -0.9603 0.0726 -13.2292 0.0000 0.0000 *** -0.2232 -0.2508 -0.1938 -0.2628 -0.1797 [-1.103, -0.818] [-1.168, -0.752]
Palindrome 1.5162 0.1034 14.6638 0.0000 0.0000 *** 0.3200 0.2881 0.3480 0.2720 0.3597 [1.314, 1.719] [1.220, 1.812]
Prime 2.2003 0.0791 27.8110 0.0000 0.0000 *** 0.4003 0.3855 0.4134 0.3780 0.4189 [2.045, 2.355] [1.974, 2.427]
Higher Sum Digits -1.1830 0.1488 -7.9526 0.0000 0.0000 *** -0.2655 -0.3137 -0.2092 -0.3333 -0.1806 [-1.474, -0.891] [-1.609, -0.757]
Asc Digit Order 0.4382 0.0723 6.0645 0.0000 0.0000 *** 0.1078 0.0736 0.1410 0.0575 0.1559 [0.297, 0.580] [0.231, 0.645]
Desc Digit Order 0.2362 0.0703 3.3611 0.0008 0.0093 ** 0.0588 0.0246 0.0924 0.0087 0.1077 [0.098, 0.374] [0.035, 0.438]
Higher Order Mag -1.9825 0.2123 -9.3390 0.0000 0.0000 *** -0.3789 -0.4167 -0.3273 -0.4303 -0.2981 [-2.399, -1.566] [-2.591, -1.374]
Mult 2 -0.2782 0.0577 -4.8177 0.0000 0.0000 *** -0.0691 -0.0966 -0.0412 -0.1091 -0.0282 [-0.391, -0.165] [-0.444, -0.113]
Mult 5 -0.1155 0.0844 -1.3691 0.1710 1.0000 n.s. -0.0288 -0.0698 0.0125 -0.0884 0.0315 [-0.281, 0.050] [-0.357, 0.126]
Mult 10 -0.7642 0.1155 -6.6157 0.0000 0.0000 *** -0.1823 -0.2292 -0.1313 -0.2493 -0.1066 [-0.991, -0.538] [-1.095, -0.433]
First Presented -2.3541 0.1629 -14.4555 0.0000 0.0000 *** -0.4133 -0.4354 -0.3844 -0.4438 -0.3685 [-2.673, -2.035] [-2.821, -1.888]

Dataset: Numbers | Model: Qwen 3.5 35B

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
Higher 4.9958 0.1360 36.7284 0.0000 0.0000 *** 0.4933 0.4912 0.4948 0.4901 0.4954 [4.729, 5.262] [4.606, 5.385]
Has Repeat Digits -0.4555 0.0710 -6.4159 0.0000 0.0000 *** -0.1120 -0.1444 -0.0784 -0.1590 -0.0627 [-0.595, -0.316] [-0.659, -0.252]
Palindrome 1.9708 0.1041 18.9237 0.0000 0.0000 *** 0.3777 0.3540 0.3980 0.3419 0.4063 [1.767, 2.175] [1.672, 2.269]
Prime 2.3951 0.0780 30.7057 0.0000 0.0000 *** 0.4165 0.4040 0.4274 0.3977 0.4321 [2.242, 2.548] [2.172, 2.619]
Higher Sum Digits -1.5375 0.1475 -10.4204 0.0000 0.0000 *** -0.3231 -0.3614 -0.2770 -0.3766 -0.2530 [-1.827, -1.248] [-1.960, -1.115]
Asc Digit Order 0.5242 0.0717 7.3118 0.0000 0.0000 *** 0.1281 0.0948 0.1603 0.0790 0.1747 [0.384, 0.665] [0.319, 0.730]
Desc Digit Order -0.2919 0.0687 -4.2495 0.0000 0.0003 *** -0.0725 -0.1051 -0.0392 -0.1198 -0.0238 [-0.427, -0.157] [-0.489, -0.095]
Higher Order Mag -2.1878 0.2145 -10.2007 0.0000 0.0000 *** -0.3991 -0.4314 -0.3541 -0.4428 -0.3282 [-2.608, -1.767] [-2.802, -1.573]
Mult 2 -0.0244 0.0566 -0.4315 0.6661 1.0000 n.s. -0.0061 -0.0338 0.0216 -0.0465 0.0344 [-0.135, 0.087] [-0.187, 0.138]
Mult 5 0.3580 0.0821 4.3629 0.0000 0.0002 *** 0.0886 0.0491 0.1269 0.0307 0.1441 [0.197, 0.519] [0.123, 0.593]
Mult 10 0.1157 0.1128 1.0261 0.3049 1.0000 n.s. 0.0289 -0.0263 0.0834 -0.0517 0.1080 [-0.105, 0.337] [-0.207, 0.439]
First Presented -2.5823 0.1618 -15.9577 0.0000 0.0000 *** -0.4297 -0.4478 -0.4059 -0.4546 -0.3927 [-2.899, -2.265] [-3.046, -2.119]

Dataset: Numbers | Model: Random Pref Order 1

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
Higher 0.6862 0.1164 5.8940 0.0000 0.0000 *** 0.1651 0.1125 0.2139 0.0873 0.2349 [0.458, 0.914] [0.353, 1.020]
Has Repeat Digits -0.8362 0.0665 -12.5678 0.0000 0.0000 *** -0.1977 -0.2244 -0.1695 -0.2363 -0.1560 [-0.967, -0.706] [-1.027, -0.646]
Palindrome -0.3008 0.0939 -3.2018 0.0014 0.0164 * -0.0746 -0.1189 -0.0291 -0.1388 -0.0079 [-0.485, -0.117] [-0.570, -0.032]
Prime -0.6465 0.0680 -9.5019 0.0000 0.0000 *** -0.1562 -0.1856 -0.1255 -0.1988 -0.1110 [-0.780, -0.513] [-0.841, -0.452]
Higher Sum Digits -2.4242 0.1361 -17.8127 0.0000 0.0000 *** -0.4187 -0.4365 -0.3964 -0.4434 -0.3843 [-2.691, -2.157] [-2.814, -2.034]
Asc Digit Order -0.6335 0.0658 -9.6237 0.0000 0.0000 *** -0.1533 -0.1819 -0.1235 -0.1947 -0.1094 [-0.763, -0.505] [-0.822, -0.445]
Desc Digit Order 0.4631 0.0648 7.1517 0.0000 0.0000 *** 0.1137 0.0833 0.1434 0.0689 0.1567 [0.336, 0.590] [0.278, 0.649]
Higher Order Mag 1.7530 0.1923 9.1176 0.0000 0.0000 *** 0.3523 0.2984 0.3938 0.2689 0.4092 [1.376, 2.130] [1.202, 2.304]
Mult 2 -1.1821 0.0549 -21.5407 0.0000 0.0000 *** -0.2653 -0.2841 -0.2455 -0.2924 -0.2359 [-1.290, -1.075] [-1.339, -1.025]
Mult 5 0.2989 0.0776 3.8511 0.0001 0.0014 ** 0.0742 0.0366 0.1109 0.0191 0.1275 [0.147, 0.451] [0.077, 0.521]
Mult 10 -1.2820 0.1065 -12.0427 0.0000 0.0000 *** -0.2828 -0.3162 -0.2452 -0.3302 -0.2265 [-1.491, -1.073] [-1.587, -0.977]
First Presented 2.0520 0.1507 13.6203 0.0000 0.0000 *** 0.3861 0.3528 0.4127 0.3348 0.4230 [1.757, 2.347] [1.620, 2.484]

Dataset: Numbers | Model: Random Pref Order 2

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
Higher -0.9810 0.1174 -8.3538 0.0000 0.0000 *** -0.2273 -0.2705 -0.1794 -0.2888 -0.1558 [-1.211, -0.751] [-1.317, -0.645]
Has Repeat Digits -1.3738 0.0680 -20.2179 0.0000 0.0000 *** -0.2980 -0.3186 -0.2757 -0.3276 -0.2648 [-1.507, -1.241] [-1.569, -1.179]
Palindrome -0.0279 0.0969 -0.2875 0.7737 1.0000 n.s. -0.0070 -0.0542 0.0404 -0.0758 0.0621 [-0.218, 0.162] [-0.305, 0.250]
Prime -0.3174 0.0683 -4.6454 0.0000 0.0000 *** -0.0787 -0.1110 -0.0457 -0.1256 -0.0304 [-0.451, -0.184] [-0.513, -0.122]
Higher Sum Digits 2.7540 0.1366 20.1663 0.0000 0.0000 *** 0.4401 0.4232 0.4535 0.4139 0.4587 [2.486, 3.022] [2.363, 3.145]
Asc Digit Order -0.5336 0.0661 -8.0782 0.0000 0.0000 *** -0.1303 -0.1599 -0.0997 -0.1732 -0.0852 [-0.663, -0.404] [-0.723, -0.344]
Desc Digit Order 0.9587 0.0648 14.7857 0.0000 0.0000 *** 0.2229 0.1967 0.2476 0.1841 0.2585 [0.832, 1.086] [0.773, 1.144]
Higher Order Mag 2.2248 0.2023 10.9955 0.0000 0.0000 *** 0.4025 0.3615 0.4322 0.3382 0.4429 [1.828, 2.621] [1.645, 2.805]
Mult 2 -0.1140 0.0536 -2.1254 0.0336 0.4026 n.s. -0.0285 -0.0546 -0.0022 -0.0665 0.0099 [-0.219, -0.009] [-0.268, 0.040]
Mult 5 0.5921 0.0782 7.5727 0.0000 0.0000 *** 0.1438 0.1080 0.1782 0.0910 0.1934 [0.439, 0.745] [0.368, 0.816]
Mult 10 0.5423 0.1073 5.0531 0.0000 0.0000 *** 0.1323 0.0822 0.1797 0.0584 0.2005 [0.332, 0.753] [0.235, 0.850]
First Presented -1.8621 0.1529 -12.1821 0.0000 0.0000 *** -0.3655 -0.3968 -0.3267 -0.4089 -0.3060 [-2.162, -1.562] [-2.300, -1.424]

Dataset: Numbers | Model: Random Pref Order 3

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
Higher -1.2006 0.1168 -10.2774 0.0000 0.0000 *** -0.2686 -0.3068 -0.2255 -0.3228 -0.2039 [-1.430, -0.972] [-1.535, -0.866]
Has Repeat Digits -0.3174 0.0652 -4.8713 0.0000 0.0000 *** -0.0787 -0.1095 -0.0473 -0.1234 -0.0326 [-0.445, -0.190] [-0.504, -0.131]
Palindrome 1.1511 0.0948 12.1443 0.0000 0.0000 *** 0.2597 0.2242 0.2920 0.2067 0.3058 [0.965, 1.337] [0.880, 1.423]
Prime -0.2063 0.0671 -3.0744 0.0021 0.0253 * -0.0514 -0.0836 -0.0187 -0.0983 -0.0035 [-0.338, -0.075] [-0.398, -0.014]
Higher Sum Digits 0.1468 0.1343 1.0935 0.2742 1.0000 n.s. 0.0366 -0.0291 0.1011 -0.0592 0.1299 [-0.116, 0.410] [-0.238, 0.532]
Asc Digit Order -0.0363 0.0661 -0.5495 0.5827 1.0000 n.s. -0.0091 -0.0414 0.0233 -0.0562 0.0382 [-0.166, 0.093] [-0.226, 0.153]
Desc Digit Order 1.0637 0.0645 16.4834 0.0000 0.0000 *** 0.2434 0.2185 0.2668 0.2066 0.2771 [0.937, 1.190] [0.879, 1.249]
Higher Order Mag -1.2169 0.1966 -6.1897 0.0000 0.0000 *** -0.2715 -0.3323 -0.1967 -0.3557 -0.1578 [-1.602, -0.832] [-1.780, -0.654]
Mult 2 0.2229 0.0530 4.2054 0.0000 0.0003 *** 0.0555 0.0297 0.0810 0.0177 0.0926 [0.119, 0.327] [0.071, 0.375]
Mult 5 1.0912 0.0777 14.0396 0.0000 0.0000 *** 0.2486 0.2189 0.2762 0.2044 0.2882 [0.939, 1.244] [0.869, 1.314]
Mult 10 -1.7849 0.1070 -16.6745 0.0000 0.0000 *** -0.3563 -0.3802 -0.3285 -0.3901 -0.3143 [-1.995, -1.575] [-2.092, -1.478]
First Presented 0.5434 0.1516 3.5851 0.0003 0.0040 ** 0.1326 0.0613 0.1985 0.0272 0.2266 [0.246, 0.840] [0.109, 0.978]

Dataset: Names | Model: Gemini 2.5 Flash

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
First Alphabetically 0.5843 0.0391 14.9459 0.0000 0.0000 *** 0.1421 0.1243 0.1595 0.1153 0.1680 [0.508, 0.661] [0.470, 0.699]
Longer Length -0.3205 0.1615 -1.9849 0.0472 0.7073 n.s. -0.0795 -0.1541 -0.0010 -0.1888 0.0383 [-0.637, -0.004] [-0.795, 0.153]
Higher Vowel Frac 0.8737 0.1274 6.8555 0.0000 0.0000 *** 0.2055 0.1511 0.2546 0.1224 0.2769 [0.624, 1.124] [0.500, 1.248]
Higher Unique Char Frac -0.5521 0.1300 -4.2453 0.0000 0.0003 *** -0.1346 -0.1915 -0.0738 -0.2178 -0.0425 [-0.807, -0.297] [-0.934, -0.170]
Starting Vowel 0.2941 0.0908 3.2398 0.0012 0.0179 * 0.0730 0.0290 0.1159 0.0069 0.1366 [0.116, 0.472] [0.028, 0.561]
Ending Vowel 1.0348 0.0704 14.6992 0.0000 0.0000 *** 0.2378 0.2103 0.2637 0.1960 0.2758 [0.897, 1.173] [0.828, 1.241]
Higher Syllable Heur -0.4271 0.1714 -2.4919 0.0127 0.1906 n.s. -0.1052 -0.1820 -0.0228 -0.2171 0.0190 [-0.763, -0.091] [-0.930, 0.076]
Higher Bouba Frac 1.6951 0.1119 15.1460 0.0000 0.0000 *** 0.3449 0.3139 0.3715 0.2968 0.3832 [1.476, 1.914] [1.367, 2.024]
Higher Rare Letter Frac 4.0042 0.2899 13.8124 0.0000 0.0000 *** 0.4821 0.4688 0.4898 0.4590 0.4923 [3.436, 4.572] [3.153, 4.855]
Higher Frac of 1960 Babies -1.9725 0.3441 -5.7319 0.0000 0.0000 *** -0.3779 -0.4338 -0.2855 -0.4518 -0.2236 [-2.647, -1.298] [-2.983, -0.962]
Higher Frac of 1980 Babies 1.6718 0.1982 8.4368 0.0000 0.0000 *** 0.3418 0.2830 0.3870 0.2484 0.4049 [1.283, 2.060] [1.090, 2.253]
Higher Frac of 2000 Babies 1.2461 0.1557 8.0017 0.0000 0.0000 *** 0.2766 0.2193 0.3251 0.1876 0.3459 [0.941, 1.551] [0.789, 1.703]
Higher Frac of 2020 Babies 6.8005 0.2110 32.2269 0.0000 0.0000 *** 0.4989 0.4983 0.4993 0.4979 0.4994 [6.387, 7.214] [6.181, 7.420]
Higher Census Freq 0.2194 0.4509 0.4866 0.6266 1.0000 n.s. 0.0546 -0.1602 0.2509 -0.2510 0.3239 [-0.664, 1.103] [-1.104, 1.543]
First Presented -7.4784 0.2263 -33.0532 0.0000 0.0000 *** -0.4994 -0.4996 -0.4991 -0.4997 -0.4989 [-7.922, -7.035] [-8.142, -6.814]

Dataset: Names | Model: Gemini 3 Flash

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
First Alphabetically 0.1436 0.0433 3.3161 0.0009 0.0137 * 0.0358 0.0147 0.0569 0.0041 0.0672 [0.059, 0.228] [0.016, 0.271]
Longer Length 2.9711 0.1817 16.3550 0.0000 0.0000 *** 0.4513 0.4318 0.4653 0.4197 0.4708 [2.615, 3.327] [2.438, 3.504]
Higher Vowel Frac 2.1454 0.1413 15.1813 0.0000 0.0000 *** 0.3952 0.3663 0.4185 0.3495 0.4283 [1.868, 2.422] [1.731, 2.560]
Higher Unique Char Frac -0.6361 0.1377 -4.6187 0.0000 0.0001 *** -0.1539 -0.2122 -0.0905 -0.2389 -0.0577 [-0.906, -0.366] [-1.040, -0.232]
Starting Vowel -0.0508 0.0992 -0.5123 0.6084 1.0000 n.s. -0.0127 -0.0610 0.0358 -0.0847 0.0598 [-0.245, 0.144] [-0.342, 0.240]
Ending Vowel 0.8292 0.0775 10.6984 0.0000 0.0000 *** 0.1962 0.1631 0.2273 0.1460 0.2421 [0.677, 0.981] [0.602, 1.057]
Higher Syllable Heur -0.2180 0.1900 -1.1475 0.2512 1.0000 n.s. -0.0543 -0.1434 0.0385 -0.1847 0.0841 [-0.590, 0.154] [-0.776, 0.340]
Higher Bouba Frac 1.9206 0.1229 15.6332 0.0000 0.0000 *** 0.3722 0.3429 0.3967 0.3263 0.4073 [1.680, 2.161] [1.560, 2.281]
Higher Rare Letter Frac 7.4658 0.4740 15.7504 0.0000 0.0000 *** 0.4994 0.4986 0.4998 0.4977 0.4999 [6.537, 8.395] [6.074, 8.857]
Higher Frac of 1960 Babies -0.5154 0.3548 -1.4526 0.1463 1.0000 n.s. -0.1261 -0.2704 0.0449 -0.3259 0.1285 [-1.211, 0.180] [-1.557, 0.526]
Higher Frac of 1980 Babies 0.6379 0.2134 2.9896 0.0028 0.0419 * 0.1543 0.0547 0.2419 0.0029 0.2797 [0.220, 1.056] [0.012, 1.264]
Higher Frac of 2000 Babies 1.8989 0.1792 10.5968 0.0000 0.0000 *** 0.3698 0.3246 0.4047 0.2979 0.4187 [1.548, 2.250] [1.373, 2.425]
Higher Frac of 2020 Babies 12.0342 0.2941 40.9249 0.0000 0.0000 *** 0.5000 0.5000 0.5000 0.5000 0.5000 [11.458, 12.611] [11.171, 12.897]
Higher Census Freq -3.3299 0.4681 -7.1129 0.0000 0.0000 *** -0.4654 -0.4859 -0.4178 -0.4910 -0.3761 [-4.247, -2.412] [-4.704, -1.956]
First Presented -12.5708 0.3131 -40.1541 0.0000 0.0000 *** -0.5000 -0.5000 -0.5000 -0.5000 -0.5000 [-13.184, -11.957] [-13.490, -11.652]

Dataset: Names | Model: Claude Haiku 4.5

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
First Alphabetically -0.1761 0.0512 -3.4395 0.0006 0.0087 ** -0.0439 -0.0687 -0.0189 -0.0809 -0.0065 [-0.277, -0.076] [-0.326, -0.026]
Longer Length 2.2935 0.2126 10.7871 0.0000 0.0000 *** 0.4083 0.3672 0.4376 0.3415 0.4487 [1.877, 2.710] [1.669, 2.918]
Higher Vowel Frac 1.7469 0.1682 10.3826 0.0000 0.0000 *** 0.3516 0.3049 0.3886 0.2778 0.4038 [1.417, 2.077] [1.253, 2.241]
Higher Unique Char Frac -1.4928 0.1692 -8.8247 0.0000 0.0000 *** -0.3165 -0.3611 -0.2616 -0.3797 -0.2303 [-1.824, -1.161] [-1.989, -0.996]
Starting Vowel -0.0008 0.1215 -0.0062 0.9951 1.0000 n.s. -0.0002 -0.0595 0.0591 -0.0884 0.0881 [-0.239, 0.237] [-0.357, 0.356]
Ending Vowel -0.5409 0.0905 -5.9761 0.0000 0.0000 *** -0.1320 -0.1722 -0.0899 -0.1914 -0.0684 [-0.718, -0.363] [-0.807, -0.275]
Higher Syllable Heur -3.8017 0.2305 -16.4954 0.0000 0.0000 *** -0.4782 -0.4860 -0.4661 -0.4888 -0.4579 [-4.253, -3.350] [-4.478, -3.125]
Higher Bouba Frac -0.8907 0.1503 -5.9254 0.0000 0.0000 *** -0.2090 -0.2659 -0.1448 -0.2912 -0.1105 [-1.185, -0.596] [-1.332, -0.449]
Higher Rare Letter Frac 2.3778 0.3130 7.5971 0.0000 0.0000 *** 0.4151 0.3538 0.4522 0.3114 0.4643 [1.764, 2.991] [1.459, 3.297]
Higher Frac of 1960 Babies -5.3386 0.4276 -12.4850 0.0000 0.0000 *** -0.4952 -0.4979 -0.4890 -0.4986 -0.4834 [-6.177, -4.500] [-6.594, -4.083]
Higher Frac of 1980 Babies 0.1180 0.2521 0.4682 0.6396 1.0000 n.s. 0.0295 -0.0929 0.1484 -0.1506 0.2022 [-0.376, 0.612] [-0.622, 0.858]
Higher Frac of 2000 Babies 1.6065 0.2019 7.9589 0.0000 0.0000 *** 0.3329 0.2705 0.3810 0.2338 0.4002 [1.211, 2.002] [1.014, 2.199]
Higher Frac of 2020 Babies 9.9910 0.2745 36.3914 0.0000 0.0000 *** 0.5000 0.4999 0.5000 0.4999 0.5000 [9.453, 10.529] [9.185, 10.797]
Higher Census Freq 1.4958 0.5466 2.7367 0.0062 0.0931 n.s. 0.3170 0.1046 0.4287 -0.0271 0.4569 [0.425, 2.567] [-0.109, 3.100]
First Presented -1.4638 0.2632 -5.5620 0.0000 0.0000 *** -0.3121 -0.3786 -0.2207 -0.4035 -0.1663 [-1.980, -0.948] [-2.236, -0.691]

Dataset: Names | Model: GPT-5 Mini

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
First Alphabetically 0.1252 0.0406 3.0816 0.0021 0.0309 * 0.0313 0.0114 0.0510 0.0015 0.0608 [0.046, 0.205] [0.006, 0.245]
Longer Length 0.7122 0.1702 4.1836 0.0000 0.0004 *** 0.1709 0.0935 0.2400 0.0529 0.2706 [0.379, 1.046] [0.213, 1.212]
Higher Vowel Frac 2.1396 0.1339 15.9827 0.0000 0.0000 *** 0.3947 0.3673 0.4170 0.3515 0.4264 [1.877, 2.402] [1.747, 2.533]
Higher Unique Char Frac -0.6353 0.1304 -4.8712 0.0000 0.0000 *** -0.1537 -0.2091 -0.0938 -0.2346 -0.0628 [-0.891, -0.380] [-1.018, -0.252]
Starting Vowel -0.1170 0.0962 -1.2172 0.2235 1.0000 n.s. -0.0292 -0.0758 0.0178 -0.0985 0.0412 [-0.305, 0.071] [-0.399, 0.165]
Ending Vowel 1.4692 0.0743 19.7855 0.0000 0.0000 *** 0.3129 0.2898 0.3341 0.2775 0.3439 [1.324, 1.615] [1.251, 1.687]
Higher Syllable Heur 1.4036 0.1779 7.8903 0.0000 0.0000 *** 0.3028 0.2417 0.3522 0.2071 0.3728 [1.055, 1.752] [0.881, 1.926]
Higher Bouba Frac 2.2955 0.1169 19.6401 0.0000 0.0000 *** 0.4085 0.3876 0.4258 0.3757 0.4333 [2.066, 2.525] [1.952, 2.639]
Higher Rare Letter Frac 5.7250 0.3765 15.2039 0.0000 0.0000 *** 0.4967 0.4932 0.4984 0.4902 0.4989 [4.987, 6.463] [4.620, 6.830]
Higher Frac of 1960 Babies -0.2901 0.3434 -0.8448 0.3982 1.0000 n.s. -0.0720 -0.2238 0.0946 -0.2855 0.1721 [-0.963, 0.383] [-1.298, 0.718]
Higher Frac of 1980 Babies -2.0630 0.2078 -9.9296 0.0000 0.0000 *** -0.3873 -0.4220 -0.3397 -0.4354 -0.3105 [-2.470, -1.656] [-2.673, -1.453]
Higher Frac of 2000 Babies 1.2617 0.1701 7.4170 0.0000 0.0000 *** 0.2793 0.2167 0.3313 0.1819 0.3533 [0.928, 1.595] [0.762, 1.761]
Higher Frac of 2020 Babies 8.3294 0.2473 33.6813 0.0000 0.0000 *** 0.4998 0.4996 0.4999 0.4995 0.4999 [7.845, 8.814] [7.604, 9.055]
Higher Census Freq -2.2471 0.4504 -4.9888 0.0000 0.0000 *** -0.4044 -0.4581 -0.2965 -0.4726 -0.2161 [-3.130, -1.364] [-3.569, -0.925]
First Presented -8.9938 0.2599 -34.6100 0.0000 0.0000 *** -0.4999 -0.4999 -0.4998 -0.4999 -0.4997 [-9.503, -8.485] [-9.757, -8.231]

Dataset: Names | Model: Qwen 3.5 35B

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
First Alphabetically 0.1723 0.0401 4.2990 0.0000 0.0003 *** 0.0430 0.0234 0.0624 0.0137 0.0720 [0.094, 0.251] [0.055, 0.290]
Longer Length 0.3218 0.1647 1.9543 0.0507 0.7601 n.s. 0.0798 -0.0002 0.1558 -0.0403 0.1911 [-0.001, 0.645] [-0.162, 0.805]
Higher Vowel Frac 2.1121 0.1327 15.9139 0.0000 0.0000 *** 0.3921 0.3644 0.4147 0.3485 0.4243 [1.852, 2.372] [1.723, 2.502]
Higher Unique Char Frac -0.0917 0.1318 -0.6958 0.4865 1.0000 n.s. -0.0229 -0.0866 0.0415 -0.1174 0.0732 [-0.350, 0.167] [-0.478, 0.295]
Starting Vowel 0.1068 0.0930 1.1487 0.2507 1.0000 n.s. 0.0267 -0.0189 0.0718 -0.0414 0.0938 [-0.075, 0.289] [-0.166, 0.380]
Ending Vowel 0.1401 0.0719 1.9482 0.0514 0.7708 n.s. 0.0350 -0.0002 0.0698 -0.0177 0.0869 [-0.001, 0.281] [-0.071, 0.351]
Higher Syllable Heur -0.2098 0.1756 -1.1947 0.2322 1.0000 n.s. -0.0523 -0.1351 0.0335 -0.1738 0.0758 [-0.554, 0.134] [-0.725, 0.306]
Higher Bouba Frac 1.0794 0.1158 9.3179 0.0000 0.0000 *** 0.2464 0.2011 0.2869 0.1769 0.3052 [0.852, 1.306] [0.739, 1.419]
Higher Rare Letter Frac 1.7996 0.2486 7.2384 0.0000 0.0000 *** 0.3581 0.2879 0.4078 0.2446 0.4262 [1.312, 2.287] [1.070, 2.529]
Higher Frac of 1960 Babies -1.7022 0.3413 -4.9877 0.0000 0.0000 *** -0.3458 -0.4146 -0.2376 -0.4373 -0.1683 [-2.371, -1.033] [-2.704, -0.700]
Higher Frac of 1980 Babies 1.4942 0.2004 7.4551 0.0000 0.0000 *** 0.3167 0.2505 0.3684 0.2122 0.3892 [1.101, 1.887] [0.906, 2.083]
Higher Frac of 2000 Babies 1.3522 0.1589 8.5074 0.0000 0.0000 *** 0.2945 0.2390 0.3407 0.2080 0.3604 [1.041, 1.664] [0.886, 1.819]
Higher Frac of 2020 Babies 8.6993 0.2407 36.1373 0.0000 0.0000 *** 0.4998 0.4997 0.4999 0.4997 0.4999 [8.228, 9.171] [7.993, 9.406]
Higher Census Freq -1.7330 0.4464 -3.8823 0.0001 0.0016 ** -0.3498 -0.4314 -0.2023 -0.4545 -0.1041 [-2.608, -0.858] [-3.043, -0.423]
First Presented -6.3651 0.2196 -28.9885 0.0000 0.0000 *** -0.4983 -0.4989 -0.4974 -0.4991 -0.4967 [-6.795, -5.935] [-7.010, -5.721]

Dataset: Names | Model: Random Pref Order 1

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
First Alphabetically 0.2774 0.0341 8.1335 0.0000 0.0000 *** 0.0689 0.0524 0.0852 0.0442 0.0933 [0.211, 0.344] [0.177, 0.377]
Longer Length -0.2692 0.1408 -1.9121 0.0559 0.8379 n.s. -0.0669 -0.1330 0.0017 -0.1643 0.0359 [-0.545, 0.007] [-0.682, 0.144]
Higher Vowel Frac 0.6094 0.1109 5.4945 0.0000 0.0000 *** 0.1478 0.0968 0.1957 0.0705 0.2181 [0.392, 0.827] [0.284, 0.935]
Higher Unique Char Frac 0.3377 0.1118 3.0203 0.0025 0.0379 * 0.0836 0.0296 0.1357 0.0024 0.1606 [0.119, 0.557] [0.010, 0.666]
Starting Vowel -1.2292 0.0806 -15.2544 0.0000 0.0000 *** -0.2737 -0.3001 -0.2448 -0.3124 -0.2296 [-1.387, -1.071] [-1.466, -0.993]
Ending Vowel 0.0054 0.0593 0.0916 0.9270 1.0000 n.s. 0.0014 -0.0277 0.0304 -0.0421 0.0448 [-0.111, 0.122] [-0.169, 0.180]
Higher Syllable Heur 2.0655 0.1500 13.7696 0.0000 0.0000 *** 0.3875 0.3546 0.4137 0.3355 0.4245 [1.772, 2.360] [1.625, 2.506]
Higher Bouba Frac 1.1788 0.0991 11.9003 0.0000 0.0000 *** 0.2647 0.2280 0.2979 0.2085 0.3130 [0.985, 1.373] [0.888, 1.470]
Higher Rare Letter Frac -0.3743 0.2092 -1.7891 0.0736 1.0000 n.s. -0.0925 -0.1866 0.0089 -0.2287 0.0597 [-0.784, 0.036] [-0.988, 0.240]
Higher Frac of 1960 Babies -2.6165 0.2960 -8.8391 0.0000 0.0000 *** -0.4319 -0.4607 -0.3846 -0.4703 -0.3517 [-3.197, -2.036] [-3.485, -1.748]
Higher Frac of 1980 Babies 0.1391 0.1718 0.8093 0.4183 1.0000 n.s. 0.0347 -0.0493 0.1168 -0.0903 0.1555 [-0.198, 0.476] [-0.365, 0.643]
Higher Frac of 2000 Babies -1.7804 0.1364 -13.0487 0.0000 0.0000 *** -0.3557 -0.3857 -0.3195 -0.3985 -0.2990 [-2.048, -1.513] [-2.181, -1.380]
Higher Frac of 2020 Babies 1.5242 0.1555 9.8031 0.0000 0.0000 *** 0.3212 0.2720 0.3616 0.2442 0.3787 [1.219, 1.829] [1.068, 1.981]
Higher Census Freq 2.7138 0.3874 7.0047 0.0000 0.0000 *** 0.4378 0.3759 0.4699 0.3287 0.4792 [1.954, 3.473] [1.577, 3.851]
First Presented -1.2909 0.1757 -7.3462 0.0000 0.0000 *** -0.2843 -0.3369 -0.2204 -0.3590 -0.1846 [-1.635, -0.946] [-1.807, -0.775]

Dataset: Names | Model: Random Pref Order 2

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
First Alphabetically -0.1278 0.0353 -3.6234 0.0003 0.0044 ** -0.0319 -0.0491 -0.0147 -0.0576 -0.0061 [-0.197, -0.059] [-0.231, -0.024]
Longer Length 1.1314 0.1469 7.7032 0.0000 0.0000 *** 0.2561 0.1992 0.3052 0.1682 0.3267 [0.844, 1.419] [0.700, 1.562]
Higher Vowel Frac -1.4712 0.1141 -12.8948 0.0000 0.0000 *** -0.3132 -0.3448 -0.2769 -0.3589 -0.2570 [-1.695, -1.248] [-1.806, -1.136]
Higher Unique Char Frac -1.1863 0.1157 -10.2538 0.0000 0.0000 *** -0.2661 -0.3042 -0.2230 -0.3214 -0.1999 [-1.413, -0.960] [-1.526, -0.847]
Starting Vowel -0.2270 0.0825 -2.7518 0.0059 0.0889 n.s. -0.0565 -0.0960 -0.0163 -0.1152 0.0038 [-0.389, -0.065] [-0.469, 0.015]
Ending Vowel 0.1614 0.0611 2.6420 0.0082 0.1236 n.s. 0.0403 0.0104 0.0698 -0.0045 0.0843 [0.042, 0.281] [-0.018, 0.341]
Higher Syllable Heur -2.2179 0.1562 -14.2012 0.0000 0.0000 *** -0.4018 -0.4258 -0.3712 -0.4356 -0.3531 [-2.524, -1.912] [-2.676, -1.760]
Higher Bouba Frac 0.0636 0.1017 0.6259 0.5314 1.0000 n.s. 0.0159 -0.0339 0.0653 -0.0584 0.0895 [-0.136, 0.263] [-0.235, 0.362]
Higher Rare Letter Frac -0.2042 0.2119 -0.9639 0.3351 1.0000 n.s. -0.0509 -0.1501 0.0526 -0.1955 0.1029 [-0.619, 0.211] [-0.826, 0.418]
Higher Frac of 1960 Babies -3.4226 0.2990 -11.4465 0.0000 0.0000 *** -0.4684 -0.4822 -0.4446 -0.4866 -0.4272 [-4.009, -2.837] [-4.300, -2.545]
Higher Frac of 1980 Babies 1.5013 0.1784 8.4150 0.0000 0.0000 *** 0.3178 0.2598 0.3642 0.2266 0.3834 [1.152, 1.851] [0.978, 2.025]
Higher Frac of 2000 Babies 1.8705 0.1422 13.1512 0.0000 0.0000 *** 0.3665 0.3309 0.3956 0.3105 0.4079 [1.592, 2.149] [1.453, 2.288]
Higher Frac of 2020 Babies -2.1577 0.1616 -13.3529 0.0000 0.0000 *** -0.3964 -0.4223 -0.3631 -0.4329 -0.3434 [-2.474, -1.841] [-2.632, -1.683]
Higher Census Freq 1.9350 0.3875 4.9940 0.0000 0.0000 *** 0.3738 0.2642 0.4367 0.1895 0.4557 [1.176, 2.694] [0.798, 3.072]
First Presented 2.1758 0.1801 12.0793 0.0000 0.0000 *** 0.3981 0.3609 0.4261 0.3385 0.4373 [1.823, 2.529] [1.647, 2.705]

Dataset: Names | Model: Random Pref Order 3

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
First Alphabetically 0.3017 0.0352 8.5629 0.0000 0.0000 *** 0.0748 0.0579 0.0916 0.0494 0.0999 [0.233, 0.371] [0.198, 0.405]
Longer Length -2.1230 0.1462 -14.5236 0.0000 0.0000 *** -0.3931 -0.4175 -0.3625 -0.4277 -0.3447 [-2.409, -1.836] [-2.552, -1.694]
Higher Vowel Frac -1.7911 0.1151 -15.5653 0.0000 0.0000 *** -0.3571 -0.3825 -0.3272 -0.3937 -0.3105 [-2.017, -1.566] [-2.129, -1.453]
Higher Unique Char Frac -1.0228 0.1175 -8.7030 0.0000 0.0000 *** -0.2355 -0.2778 -0.1884 -0.2970 -0.1633 [-1.253, -0.792] [-1.368, -0.678]
Starting Vowel -1.1782 0.0824 -14.2981 0.0000 0.0000 *** -0.2646 -0.2924 -0.2343 -0.3053 -0.2184 [-1.340, -1.017] [-1.420, -0.936]
Ending Vowel 0.9644 0.0620 15.5565 0.0000 0.0000 *** 0.2240 0.1991 0.2476 0.1862 0.2588 [0.843, 1.086] [0.782, 1.146]
Higher Syllable Heur 0.0986 0.1535 0.6425 0.5205 1.0000 n.s. 0.0246 -0.0504 0.0986 -0.0871 0.1339 [-0.202, 0.399] [-0.352, 0.549]
Higher Bouba Frac 1.5460 0.1027 15.0584 0.0000 0.0000 *** 0.3243 0.2933 0.3516 0.2764 0.3638 [1.345, 1.747] [1.245, 1.847]
Higher Rare Letter Frac -0.2739 0.2137 -1.2817 0.2000 1.0000 n.s. -0.0681 -0.1666 0.0362 -0.2112 0.0874 [-0.693, 0.145] [-0.901, 0.353]
Higher Frac of 1960 Babies -2.6537 0.2957 -8.9752 0.0000 0.0000 *** -0.4342 -0.4621 -0.3884 -0.4713 -0.3564 [-3.233, -2.074] [-3.522, -1.786]
Higher Frac of 1980 Babies 2.7079 0.1781 15.2086 0.0000 0.0000 *** 0.4375 0.4136 0.4551 0.3989 0.4620 [2.359, 3.057] [2.185, 3.231]
Higher Frac of 2000 Babies 1.7673 0.1426 12.3900 0.0000 0.0000 *** 0.3541 0.3157 0.3856 0.2939 0.3990 [1.488, 2.047] [1.349, 2.186]
Higher Frac of 2020 Babies -1.0001 0.1576 -6.3453 0.0000 0.0000 *** -0.2311 -0.2873 -0.1662 -0.3119 -0.1312 [-1.309, -0.691] [-1.463, -0.537]
Higher Census Freq -1.7724 0.3840 -4.6161 0.0000 0.0001 *** -0.3548 -0.4259 -0.2349 -0.4478 -0.1560 [-2.525, -1.020] [-2.899, -0.645]
First Presented 2.2146 0.1815 12.2006 0.0000 0.0000 *** 0.4016 0.3652 0.4289 0.3431 0.4398 [1.859, 2.570] [1.682, 2.747]
2. Cities Benchmark (Click to Expand)

Paper Replica Figure 1: Cities Logistic Regression Plot

Paper Replica Figure 1 for Cities

Paper Replica Table 8: Cities Win Fractions (Color-coded by distance from 0.5)

MetricGemini 2.5 FlashGemini 3 FlashClaude Haiku 4.5GPT-5 MiniQwen 3.5 35BRandom Pref Order 1Random Pref Order 2Random Pref Order 3
Starting Vowel (32%)0.640.640.530.64nan0.610.560.47
Longer Length0.430.39*0.420.41*nan0.530.42*0.52
Higher Bouba Frac0.570.60*0.570.61**nan0.490.520.52
Higher Syllable Heur0.480.450.430.48nan0.480.38**0.55
Higher Vowel Frac0.58*0.60*0.530.60*nan0.450.540.49
Ending Vowel (39%)0.550.560.500.56nan0.530.440.46
Higher Rare Letter Frac (14%)0.570.540.540.59nan0.500.500.50
Generated Earlier Within Own Category0.530.560.520.54nan0.510.440.40**
First Generated0.440.420.490.48nan0.510.570.53
First Alphabetically0.530.520.490.50nan0.510.540.53
Higher Unique Char Frac0.480.510.540.51nan0.520.540.58*
First Presented0.510.47**0.510.39***nan0.500.500.50
Regression Coefficient & 95% CI Summary (Cities)

Values shown: Probability - 0.5 [95% CI Lower, 95% CI Upper] Significance Tag

Model Gemini 2.5 Flash Gemini 3 Flash Claude Haiku 4.5 GPT-5 Mini Random Pref Order 1 Random Pref Order 2 Random Pref Order 3
Metric
Ending Vowel +0.018 [-0.007, +0.043] n.s. +0.025 [-0.002, +0.051] n.s. -0.089 [-0.113, -0.065] *** +0.042 [+0.015, +0.068] * +0.132 [+0.108, +0.154] *** -0.085 [-0.109, -0.060] *** -0.076 [-0.101, -0.051] ***
First Alphabetically +0.038 [+0.023, +0.053] *** +0.044 [+0.027, +0.060] *** -0.011 [-0.026, +0.004] n.s. +0.008 [-0.008, +0.025] n.s. -0.009 [-0.024, +0.006] n.s. +0.143 [+0.128, +0.157] *** +0.083 [+0.067, +0.098] ***
First Generated +0.156 [+0.124, +0.185] *** +0.282 [+0.257, +0.306] *** -0.042 [-0.076, -0.009] n.s. -0.127 [-0.160, -0.093] *** -0.013 [-0.047, +0.020] n.s. -0.356 [-0.372, -0.337] *** -0.218 [-0.245, -0.188] ***
First Presented -0.397 [-0.430, -0.351] *** -0.235 [-0.312, -0.140] *** -0.302 [-0.358, -0.232] *** -0.378 [-0.416, -0.326] *** -0.047 [-0.135, +0.044] n.s. +0.491 [+0.486, +0.494] *** -0.476 [-0.484, -0.465] ***
Generated Earlier Within Own Category -0.124 [-0.153, -0.095] *** -0.248 [-0.272, -0.222] *** -0.004 [-0.035, +0.026] n.s. -0.124 [-0.154, -0.092] *** -0.108 [-0.137, -0.079] *** -0.029 [-0.061, +0.002] n.s. +0.407 [+0.396, +0.418] ***
Higher Bouba Frac +0.258 [+0.223, +0.291] *** +0.348 [+0.321, +0.372] *** +0.273 [+0.240, +0.303] *** +0.365 [+0.341, +0.387] *** -0.036 [-0.080, +0.009] n.s. +0.004 [-0.042, +0.050] n.s. -0.003 [-0.051, +0.046] n.s.
Higher Rare Letter Frac +0.473 [+0.449, +0.486] *** +0.460 [+0.424, +0.480] *** +0.473 [+0.452, +0.485] *** +0.454 [+0.417, +0.474] *** +0.066 [-0.061, +0.185] n.s. -0.342 [-0.400, -0.261] *** +0.452 [+0.418, +0.472] ***
Higher Syllable Heur +0.374 [+0.331, +0.407] *** +0.466 [+0.453, +0.476] *** +0.037 [-0.046, +0.117] n.s. +0.479 [+0.470, +0.485] *** -0.280 [-0.331, -0.218] *** -0.381 [-0.413, -0.341] *** +0.428 [+0.402, +0.448] ***
Higher Unique Char Frac -0.152 [-0.199, -0.102] *** -0.160 [-0.209, -0.106] *** +0.017 [-0.034, +0.068] n.s. -0.073 [-0.127, -0.017] n.s. -0.030 [-0.080, +0.021] n.s. -0.310 [-0.340, -0.276] *** +0.428 [+0.412, +0.442] ***
Higher Vowel Frac +0.097 [+0.044, +0.147] ** -0.124 [-0.174, -0.070] *** +0.071 [+0.019, +0.121] n.s. +0.075 [+0.020, +0.129] n.s. -0.244 [-0.282, -0.202] *** -0.184 [-0.229, -0.135] *** -0.028 [-0.083, +0.027] n.s.
Longer Length -0.447 [-0.465, -0.421] *** -0.498 [-0.499, -0.497] *** -0.331 [-0.382, -0.263] *** -0.495 [-0.497, -0.493] *** +0.287 [+0.209, +0.349] *** -0.404 [-0.436, -0.360] *** -0.251 [-0.324, -0.160] ***
Starting Vowel +0.124 [+0.096, +0.151] *** +0.126 [+0.096, +0.154] *** -0.052 [-0.080, -0.023] ** +0.084 [+0.053, +0.114] *** +0.300 [+0.281, +0.318] *** +0.120 [+0.092, +0.148] *** -0.193 [-0.219, -0.167] ***
Detailed Model-by-Model Regression Tables & Bonferroni p-values (Cities)

Model: Gemini 2.5 Flash

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
First Alphabetically 0.1524 0.0317 4.8117 0.0000 0.0000 *** 0.0380 0.0226 0.0534 0.0154 0.0605 [0.090, 0.214] [0.062, 0.243]
Longer Length -2.8850 0.2198 -13.1280 0.0000 0.0000 *** -0.4471 -0.4650 -0.4209 -0.4711 -0.4051 [-3.316, -2.454] [-3.515, -2.255]
Higher Vowel Frac 0.3912 0.1088 3.5952 0.0003 0.0039 ** 0.0966 0.0444 0.1467 0.0198 0.1689 [0.178, 0.604] [0.079, 0.703]
Higher Unique Char Frac -0.6268 0.1091 -5.7477 0.0000 0.0000 *** -0.1518 -0.1986 -0.1018 -0.2190 -0.0779 [-0.841, -0.413] [-0.939, -0.314]
Starting Vowel 0.5057 0.0603 8.3899 0.0000 0.0000 *** 0.1238 0.0957 0.1511 0.0825 0.1634 [0.388, 0.624] [0.333, 0.678]
Ending Vowel 0.0725 0.0515 1.4070 0.1594 1.0000 n.s. 0.0181 -0.0071 0.0432 -0.0188 0.0548 [-0.028, 0.173] [-0.075, 0.220]
Higher Syllable Heur 1.9355 0.1731 11.1784 0.0000 0.0000 *** 0.3739 0.3315 0.4068 0.3084 0.4192 [1.596, 2.275] [1.439, 2.432]
Higher Bouba Frac 1.1434 0.0951 12.0245 0.0000 0.0000 *** 0.2583 0.2225 0.2908 0.2049 0.3047 [0.957, 1.330] [0.871, 1.416]
Higher Rare Letter Frac 3.5934 0.3384 10.6196 0.0000 0.0000 *** 0.4732 0.4493 0.4860 0.4324 0.4897 [2.930, 4.257] [2.624, 4.563]
First Generated 0.6433 0.0692 9.2914 0.0000 0.0000 *** 0.1555 0.1243 0.1855 0.1094 0.1988 [0.508, 0.779] [0.445, 0.842]
Generated Earlier Within Own Category -0.5076 0.0635 -7.9966 0.0000 0.0000 *** -0.1243 -0.1530 -0.0946 -0.1659 -0.0807 [-0.632, -0.383] [-0.690, -0.326]
First Presented -2.1663 0.2165 -10.0038 0.0000 0.0000 *** -0.3972 -0.4303 -0.3509 -0.4420 -0.3243 [-2.591, -1.742] [-2.787, -1.546]

Model: Gemini 3 Flash

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
First Alphabetically 0.1750 0.0333 5.2503 0.0000 0.0000 *** 0.0436 0.0274 0.0598 0.0199 0.0672 [0.110, 0.240] [0.079, 0.270]
Longer Length -6.3582 0.2362 -26.9210 0.0000 0.0000 *** -0.4983 -0.4989 -0.4973 -0.4991 -0.4966 [-6.821, -5.895] [-7.035, -5.681]
Higher Vowel Frac -0.5049 0.1141 -4.4260 0.0000 0.0001 *** -0.1236 -0.1745 -0.0699 -0.1967 -0.0444 [-0.728, -0.281] [-0.832, -0.178]
Higher Unique Char Frac -0.6616 0.1179 -5.6099 0.0000 0.0000 *** -0.1596 -0.2095 -0.1060 -0.2310 -0.0802 [-0.893, -0.430] [-0.999, -0.324]
Starting Vowel 0.5144 0.0633 8.1318 0.0000 0.0000 *** 0.1258 0.0964 0.1544 0.0825 0.1672 [0.390, 0.638] [0.333, 0.696]
Ending Vowel 0.0985 0.0540 1.8240 0.0682 0.8178 n.s. 0.0246 -0.0018 0.0509 -0.0141 0.0630 [-0.007, 0.204] [-0.056, 0.253]
Higher Syllable Heur 3.3575 0.1825 18.3963 0.0000 0.0000 *** 0.4664 0.4526 0.4762 0.4445 0.4798 [3.000, 3.715] [2.835, 3.880]
Higher Bouba Frac 1.7189 0.1006 17.0921 0.0000 0.0000 *** 0.3480 0.3208 0.3717 0.3070 0.3815 [1.522, 1.916] [1.431, 2.007]
Higher Rare Letter Frac 3.1898 0.3510 9.0869 0.0000 0.0000 *** 0.4604 0.4243 0.4797 0.3988 0.4852 [2.502, 3.878] [2.184, 4.196]
First Generated 1.2799 0.0723 17.6946 0.0000 0.0000 *** 0.2824 0.2573 0.3056 0.2451 0.3156 [1.138, 1.422] [1.073, 1.487]
Generated Earlier Within Own Category -1.0870 0.0671 -16.1942 0.0000 0.0000 *** -0.2478 -0.2718 -0.2222 -0.2823 -0.2098 [-1.219, -0.955] [-1.279, -0.895]
First Presented -1.0186 0.2253 -4.5213 0.0000 0.0001 *** -0.2347 -0.3116 -0.1404 -0.3408 -0.0922 [-1.460, -0.577] [-1.664, -0.373]

Model: Claude Haiku 4.5

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
First Alphabetically -0.0445 0.0311 -1.4320 0.1521 1.0000 n.s. -0.0111 -0.0263 0.0041 -0.0333 0.0111 [-0.105, 0.016] [-0.134, 0.045]
Longer Length -1.5919 0.2143 -7.4286 0.0000 0.0000 *** -0.3309 -0.3820 -0.2635 -0.4008 -0.2267 [-2.012, -1.172] [-2.206, -0.978]
Higher Vowel Frac 0.2853 0.1073 2.6591 0.0078 0.0940 n.s. 0.0708 0.0187 0.1214 -0.0055 0.1440 [0.075, 0.496] [-0.022, 0.593]
Higher Unique Char Frac 0.0696 0.1051 0.6620 0.5080 1.0000 n.s. 0.0174 -0.0341 0.0685 -0.0576 0.0916 [-0.136, 0.276] [-0.232, 0.371]
Starting Vowel -0.2085 0.0590 -3.5349 0.0004 0.0049 ** -0.0519 -0.0803 -0.0232 -0.0933 -0.0099 [-0.324, -0.093] [-0.377, -0.039]
Ending Vowel -0.3592 0.0505 -7.1129 0.0000 0.0000 *** -0.0888 -0.1126 -0.0647 -0.1234 -0.0534 [-0.458, -0.260] [-0.504, -0.215]
Higher Syllable Heur 0.1466 0.1681 0.8724 0.3830 1.0000 n.s. 0.0366 -0.0456 0.1168 -0.0830 0.1521 [-0.183, 0.476] [-0.335, 0.628]
Higher Bouba Frac 1.2255 0.0928 13.2100 0.0000 0.0000 *** 0.2730 0.2396 0.3034 0.2231 0.3163 [1.044, 1.407] [0.960, 1.491]
Higher Rare Letter Frac 3.5944 0.3088 11.6417 0.0000 0.0000 *** 0.4733 0.4521 0.4852 0.4376 0.4888 [2.989, 4.200] [2.710, 4.479]
First Generated -0.1703 0.0685 -2.4880 0.0128 0.1541 n.s. -0.0425 -0.0756 -0.0090 -0.0906 0.0065 [-0.305, -0.036] [-0.367, 0.026]
Generated Earlier Within Own Category -0.0168 0.0623 -0.2691 0.7878 1.0000 n.s. -0.0042 -0.0347 0.0263 -0.0487 0.0404 [-0.139, 0.105] [-0.195, 0.162]
First Presented -1.4019 0.2026 -6.9195 0.0000 0.0000 *** -0.3025 -0.3580 -0.2320 -0.3789 -0.1945 [-1.799, -1.005] [-1.982, -0.821]

Model: GPT-5 Mini

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
First Alphabetically 0.0327 0.0334 0.9780 0.3281 1.0000 n.s. 0.0082 -0.0082 0.0245 -0.0158 0.0321 [-0.033, 0.098] [-0.063, 0.128]
Longer Length -5.3701 0.2343 -22.9169 0.0000 0.0000 *** -0.4954 -0.4971 -0.4927 -0.4976 -0.4910 [-5.829, -4.911] [-6.042, -4.699]
Higher Vowel Frac 0.3043 0.1149 2.6475 0.0081 0.0973 n.s. 0.0755 0.0197 0.1294 -0.0063 0.1533 [0.079, 0.530] [-0.025, 0.634]
Higher Unique Char Frac -0.2926 0.1154 -2.5350 0.0112 0.1349 n.s. -0.0726 -0.1269 -0.0166 -0.1510 0.0095 [-0.519, -0.066] [-0.623, 0.038]
Starting Vowel 0.3384 0.0637 5.3092 0.0000 0.0000 *** 0.0838 0.0532 0.1138 0.0389 0.1274 [0.213, 0.463] [0.156, 0.521]
Ending Vowel 0.1675 0.0541 3.0931 0.0020 0.0238 * 0.0418 0.0153 0.0680 0.0031 0.0799 [0.061, 0.274] [0.012, 0.323]
Higher Syllable Heur 3.8445 0.1846 20.8265 0.0000 0.0000 *** 0.4791 0.4702 0.4853 0.4650 0.4875 [3.483, 4.206] [3.316, 4.373]
Higher Bouba Frac 1.8617 0.1009 18.4567 0.0000 0.0000 *** 0.3655 0.3408 0.3869 0.3282 0.3957 [1.664, 2.059] [1.573, 2.151]
Higher Rare Letter Frac 3.0248 0.3153 9.5949 0.0000 0.0000 *** 0.4537 0.4174 0.4745 0.3930 0.4807 [2.407, 3.643] [2.122, 3.928]
First Generated -0.5195 0.0725 -7.1647 0.0000 0.0000 *** -0.1270 -0.1596 -0.0932 -0.1742 -0.0773 [-0.662, -0.377] [-0.727, -0.312]
Generated Earlier Within Own Category -0.5048 0.0671 -7.5213 0.0000 0.0000 *** -0.1236 -0.1539 -0.0922 -0.1675 -0.0775 [-0.636, -0.373] [-0.697, -0.312]
First Presented -1.9761 0.2121 -9.3168 0.0000 0.0000 *** -0.3783 -0.4162 -0.3264 -0.4298 -0.2971 [-2.392, -1.560] [-2.584, -1.368]

Model: Random Pref Order 1

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
First Alphabetically -0.0358 0.0310 -1.1544 0.2483 1.0000 n.s. -0.0090 -0.0241 0.0062 -0.0311 0.0133 [-0.097, 0.025] [-0.125, 0.053]
Longer Length 1.3081 0.2126 6.1521 0.0000 0.0000 *** 0.2872 0.2092 0.3488 0.1679 0.3718 [0.891, 1.725] [0.699, 1.917]
Higher Vowel Frac -1.0676 0.1072 -9.9594 0.0000 0.0000 *** -0.2441 -0.2821 -0.2021 -0.2981 -0.1815 [-1.278, -0.857] [-1.375, -0.760]
Higher Unique Char Frac -0.1202 0.1038 -1.1580 0.2468 1.0000 n.s. -0.0300 -0.0802 0.0208 -0.1029 0.0442 [-0.324, 0.083] [-0.418, 0.177]
Starting Vowel 1.3870 0.0604 22.9764 0.0000 0.0000 *** 0.3001 0.2805 0.3184 0.2710 0.3263 [1.269, 1.505] [1.214, 1.560]
Ending Vowel 0.5387 0.0506 10.6463 0.0000 0.0000 *** 0.1315 0.1081 0.1543 0.0972 0.1646 [0.440, 0.638] [0.394, 0.684]
Higher Syllable Heur -1.2640 0.1681 -7.5202 0.0000 0.0000 *** -0.2797 -0.3311 -0.2180 -0.3514 -0.1862 [-1.593, -0.935] [-1.746, -0.782]
Higher Bouba Frac -0.1437 0.0916 -1.5682 0.1168 1.0000 n.s. -0.0359 -0.0801 0.0090 -0.1002 0.0297 [-0.323, 0.036] [-0.406, 0.119]
Higher Rare Letter Frac 0.2670 0.2607 1.0241 0.3058 1.0000 n.s. 0.0663 -0.0607 0.1852 -0.1177 0.2338 [-0.244, 0.778] [-0.480, 1.014]
First Generated -0.0540 0.0680 -0.7941 0.4271 1.0000 n.s. -0.0135 -0.0467 0.0198 -0.0619 0.0351 [-0.187, 0.079] [-0.249, 0.141]
Generated Earlier Within Own Category -0.4392 0.0625 -7.0253 0.0000 0.0000 *** -0.1081 -0.1369 -0.0785 -0.1499 -0.0647 [-0.562, -0.317] [-0.618, -0.260]
First Presented -0.1881 0.1856 -1.0132 0.3109 1.0000 n.s. -0.0469 -0.1346 0.0438 -0.1726 0.0851 [-0.552, 0.176] [-0.720, 0.344]

Model: Random Pref Order 2

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
First Alphabetically 0.5863 0.0323 18.1616 0.0000 0.0000 *** 0.1425 0.1279 0.1569 0.1210 0.1635 [0.523, 0.650] [0.494, 0.679]
Longer Length -2.2477 0.2222 -10.1151 0.0000 0.0000 *** -0.4045 -0.4360 -0.3596 -0.4471 -0.3336 [-2.683, -1.812] [-2.884, -1.611]
Higher Vowel Frac -0.7728 0.1110 -6.9655 0.0000 0.0000 *** -0.1841 -0.2291 -0.1354 -0.2485 -0.1118 [-0.990, -0.555] [-1.091, -0.455]
Higher Unique Char Frac -1.4502 0.1074 -13.5074 0.0000 0.0000 *** -0.3100 -0.3403 -0.2755 -0.3529 -0.2582 [-1.661, -1.240] [-1.758, -1.143]
Starting Vowel 0.4905 0.0601 8.1556 0.0000 0.0000 *** 0.1202 0.0921 0.1476 0.0789 0.1599 [0.373, 0.608] [0.318, 0.663]
Ending Vowel -0.3434 0.0518 -6.6290 0.0000 0.0000 *** -0.0850 -0.1094 -0.0602 -0.1205 -0.0486 [-0.445, -0.242] [-0.492, -0.195]
Higher Syllable Heur -2.0059 0.1748 -11.4765 0.0000 0.0000 *** -0.3814 -0.4128 -0.3407 -0.4246 -0.3183 [-2.348, -1.663] [-2.507, -1.505]
Higher Bouba Frac 0.0167 0.0948 0.1764 0.8600 1.0000 n.s. 0.0042 -0.0422 0.0505 -0.0634 0.0716 [-0.169, 0.203] [-0.255, 0.288]
Higher Rare Letter Frac -1.6752 0.2648 -6.3274 0.0000 0.0000 *** -0.3423 -0.3997 -0.2607 -0.4194 -0.2144 [-2.194, -1.156] [-2.434, -0.917]
First Generated -1.7797 0.0718 -24.7729 0.0000 0.0000 *** -0.3557 -0.3722 -0.3374 -0.3793 -0.3283 [-1.920, -1.639] [-1.986, -1.574]
Generated Earlier Within Own Category -0.1167 0.0645 -1.8092 0.0704 0.8451 n.s. -0.0292 -0.0605 0.0024 -0.0748 0.0170 [-0.243, 0.010] [-0.302, 0.068]
First Presented 4.6491 0.1951 23.8350 0.0000 0.0000 *** 0.4905 0.4862 0.4935 0.4835 0.4946 [4.267, 5.031] [4.090, 5.208]

Model: Random Pref Order 3

Metric Log-Odds Beta Std Error z-score Raw p-val Bonferroni p-val Significance Prob Coeff (-0.5) 95% CI Lower (Prob) 95% CI Upper (Prob) Bonf CI Lower (Prob) Bonf CI Upper (Prob) 95% CI (Log-Odds) Bonf CI (Log-Odds)
First Alphabetically 0.3341 0.0324 10.2981 0.0000 0.0000 *** 0.0828 0.0672 0.0981 0.0600 0.1052 [0.271, 0.398] [0.241, 0.427]
Longer Length -1.1040 0.2237 -4.9341 0.0000 0.0000 *** -0.2510 -0.3238 -0.1605 -0.3513 -0.1137 [-1.543, -0.665] [-1.745, -0.463]
Higher Vowel Frac -0.1141 0.1125 -1.0135 0.3108 1.0000 n.s. -0.0285 -0.0829 0.0266 -0.1074 0.0519 [-0.335, 0.107] [-0.437, 0.208]
Higher Unique Char Frac 2.5637 0.1143 22.4227 0.0000 0.0000 *** 0.4285 0.4121 0.4420 0.4034 0.4474 [2.340, 2.788] [2.236, 2.891]
Starting Vowel -0.8153 0.0626 -13.0271 0.0000 0.0000 *** -0.1932 -0.2187 -0.1666 -0.2300 -0.1538 [-0.938, -0.693] [-0.995, -0.636]
Ending Vowel -0.3078 0.0528 -5.8274 0.0000 0.0000 *** -0.0763 -0.1014 -0.0509 -0.1128 -0.0390 [-0.411, -0.204] [-0.459, -0.156]
Higher Syllable Heur 2.5622 0.1758 14.5779 0.0000 0.0000 *** 0.4284 0.4018 0.4482 0.3868 0.4555 [2.218, 2.907] [2.059, 3.066]
Higher Bouba Frac -0.0104 0.0986 -0.1055 0.9160 1.0000 n.s. -0.0026 -0.0508 0.0456 -0.0727 0.0676 [-0.204, 0.183] [-0.293, 0.272]
Higher Rare Letter Frac 2.9814 0.2885 10.3356 0.0000 0.0000 *** 0.4517 0.4180 0.4720 0.3961 0.4783 [2.416, 3.547] [2.155, 3.808]
First Generated -0.9334 0.0716 -13.0340 0.0000 0.0000 *** -0.2178 -0.2453 -0.1885 -0.2574 -0.1744 [-1.074, -0.793] [-1.139, -0.728]
Generated Earlier Within Own Category 2.2819 0.0678 33.6734 0.0000 0.0000 *** 0.4074 0.3956 0.4179 0.3897 0.4224 [2.149, 2.415] [2.088, 2.476]
First Presented -3.7192 0.2030 -18.3195 0.0000 0.0000 *** -0.4763 -0.4840 -0.4652 -0.4866 -0.4584 [-4.117, -3.321] [-4.301, -3.138]
3. Mixed-Effects Model (GLMM) & Variance Decomposition (Click to Expand)

Rebuttal Context & Variance Disentanglement

"The design is fully crossed — every unique pair is evaluated under all 50 prompts and both orders — so per-prompt win rates are computed over an identical set of pairs, and pair-to-pair differences cannot masquerade as prompt effects. The remaining question is prompt effects vs. sampling noise, which is what the Levene's test addresses (rejecting the sampling-only null at p<0.001 for all 21 distributions). A mixed-effects model with crossed prompt and pair random effects is indeed the cleaner formulation..."

Variance Component Breakdown (Crossed Random Effects)

Total per-prompt win rate variance is decomposed into: Prompt Variance (\(\sigma^2_{\text{prompt}}\)), Item-Pair Variance (\(\sigma^2_{\text{pair}}\)), and Sampling / Residual Noise. Levene's \(W\) statistic tests whether prompt variance significantly exceeds pure sampling noise (\(H_0: \sigma^2_{\text{prompt}} = 0\)).

Dataset Model Prompt Var (σ²_prompt) Pair Var (σ²_pair) Sampling / Noise Var Levene W Stat Levene p-val Levene Test
Numbers Gemini 2.5 Flash 0.0061 (10.2%) 0.0026 (4.4%) 0.0506 (85.4%) 120.82 1.56e-25 *** (p<0.001)
Numbers Gemini 3 Flash 0.0014 (3.6%) 0.0052 (13.4%) 0.0322 (83.0%) 0.24 6.22e-01 n.s.
Numbers Claude Haiku 4.5 0.0101 (19.5%) 0.0051 (9.9%) 0.0364 (70.5%) 200.09 6.01e-39 *** (p<0.001)
Numbers GPT-5 Mini 0.0043 (8.6%) 0.0020 (4.1%) 0.0434 (87.3%) 7.36 6.87e-03 *
Numbers Qwen 3.5 35B 0.0025 (4.5%) 0.0026 (4.6%) 0.0506 (90.9%) 18.85 1.68e-05 *** (p<0.001)
Names Gemini 2.5 Flash 0.0031 (4.8%) 0.0021 (3.2%) 0.0594 (92.0%) 27.99 1.76e-07 *** (p<0.001)
Names Gemini 3 Flash 0.0019 (5.0%) 0.0061 (16.4%) 0.0292 (78.6%) 3.01 8.33e-02 n.s.
Names Claude Haiku 4.5 0.0040 (6.7%) 0.0256 (42.2%) 0.0310 (51.1%) 19.02 1.54e-05 *** (p<0.001)
Names GPT-5 Mini 0.0047 (9.5%) 0.0047 (9.5%) 0.0398 (81.0%) 5.94 1.51e-02 *
Names Qwen 3.5 35B 0.0043 (7.2%) 0.0055 (9.3%) 0.0497 (83.5%) 55.27 4.07e-13 *** (p<0.001)
Cities Gemini 2.5 Flash 0.0019 (3.0%) 0.0036 (5.6%) 0.0585 (91.4%) 11.25 8.51e-04 *** (p<0.001)
Cities Gemini 3 Flash 0.0005 (1.3%) 0.0113 (27.7%) 0.0290 (71.0%) 7.45 6.56e-03 *
Cities Claude Haiku 4.5 0.0110 (16.2%) 0.0116 (17.0%) 0.0454 (66.8%) 116.17 1.07e-24 *** (p<0.001)
Cities GPT-5 Mini 0.0041 (7.1%) 0.0063 (11.0%) 0.0468 (81.9%) 16.84 4.68e-05 *** (p<0.001)
4. Monte Carlo Permutation Test & Empirical Histograms (N=25,000) (Click to Expand)

Monte Carlo Permutation Framework (N = 25,000 Shuffles)

To test whether an observed feature bias could be an artifact of some unrelated, hidden entity preference, we performed a Monte Carlo permutation test. In each of the \(N = 25,000\) shuffles, we generate a random "fake" preference ordering over all entities, resolve every matchup according to that ordering, and compute the resulting feature win rate.

This shows what win rates can arise purely from arbitrary underlying entity preferences. Because we test \(m = 183\) features across datasets and models, a Bonferroni correction requires a significance threshold of \(0.05 / 183 \approx 0.00027\). Using 25,000 shuffles provides a minimum detectable p-value of \(p_{\text{min}} = \frac{1}{25,001} \approx 4.0 \times 10^{-5}\), giving us enough resolution to reliably test significance across all 183 comparisons.

Each plot below shows the null distribution of feature win rates across 25,000 random shuffles (grey histogram and navy curve), the 95% confidence interval (\([Q_{2.5\%}, Q_{97.5\%]}\), grey shading centered at 50%), and the observed win rate for each model family (vertical lines).

1. Numbers Benchmark: All 11 Features (N = 25,000 Shuffles)

Numbers Benchmark Monte Carlo Histograms

2. Names Benchmark: All 14 Features (N = 25,000 Shuffles)

Names Benchmark Monte Carlo Histograms

3. Cities Benchmark: All 11 Features (N = 25,000 Shuffles)

Cities Benchmark Monte Carlo Histograms