Summary

This document contains two new tables which build off the configuration 3 table from previous updates. These new configurations (4 and 5) include:

  1. Configuration 4 - this uses the vote_share instead of a parties win-record in a particular state in all other years
  2. Configuration 5 - this uses the vote_share_ratio of the individuals own/opponent party

Table 1 - Configuration 4

Replacing win_rate with vote_share in the election lm. This now includes:

  1. Party (limited to Democrat vs. Republican)
  2. Vote-share
  3. Party * Vote-share

Vote-share is computed as: “average vote share of my party in this state over all years excluding the current”

Table 01 - Version 04 - Election Regressions
Fit measured in adjusted R squared and AUC
Model Configuration Adjusted R Squared ROC AUC
Single Variable Model
Election LM 0.0108 0.5728622
Lower 95% C.I. 0.0023 0.5376222
Upper 95% C.I. 0.0255 0.6081021
Sex 0.0153 0.5604767
0.0047 0.5314525
0.0315 0.5895009
Skine-Tone −0.0049 0.5454082
−0.0013 0.5106945
0.0252 0.5801219
MTurk Features 0.0017 0.5433649
−0.0016 0.5077725
0.0153 0.5789574
P_hat_cnn 0.0972 0.6814615
0.0675 0.6487163
0.1294 0.7142067
Combined Variable Model
Election LM + P_hat_cnn 0.1044 0.6861957
0.0755 0.6536344
0.1388 0.7187569
Election LM + Sex 0.0229 0.5967729
0.0091 0.5618596
0.0418 0.6316862
Election LM + Sex + P_hat_cnn 0.1065 0.6888393
0.0803 0.6563431
0.1421 0.7213355
Election LM + Sex + Skin-Tone 0.0173 0.6147882
0.0164 0.5801846
0.0572 0.6493917
Election LM + Sex + Skin-Tone + P_hat_cnn 0.1023 0.7019433
0.0868 0.6699021
0.1563 0.7339845
Election LM + Sex + Skin-Tone + MTurk 0.0179 0.6196128
0.0191 0.5851869
0.0627 0.6540388
Election LM + Sex + Skin-Tone + MTurk + P_hat_cnn 0.1032 0.7052443
0.0945 0.6733634
0.1628 0.7371253

Table 1 - Configuration 5

Replacing win_rate with vote_share_ratio in the election lm. This now includes:

  1. Party (limited to Democrat vs. Republican)
  2. Vote-share-ratio
  3. Party * Vote-share-ratio

Vote share ratio is computed as: “average vote share of my party in this state over all years excluding the current / opponents parties vote share in this state over all years excluding the current”. This works as we are limiting ourselves to races with two parties !

Table 01 - Version 05 - Election Regressions
Fit measured in adjusted R squared and AUC
Model Configuration Adjusted R Squared ROC AUC
Single Variable Model
Election LM 0.0104 0.5718189
Lower 95% C.I. 0.0022 0.5365509
Upper 95% C.I. 0.0250 0.6070869
Sex 0.0153 0.5604767
0.0049 0.5314525
0.0309 0.5895009
Skine-Tone −0.0049 0.5454082
−0.0009 0.5106945
0.0267 0.5801219
MTurk Features 0.0017 0.5433649
−0.0014 0.5077725
0.0157 0.5789574
P_hat_cnn 0.0972 0.6814615
0.0695 0.6487163
0.1289 0.7142067
Combined Variable Model
Election LM + P_hat_cnn 0.1038 0.6856720
0.0760 0.6531037
0.1378 0.7182404
Election LM + Sex 0.0225 0.5969973
0.0096 0.5620844
0.0418 0.6319102
Election LM + Sex + P_hat_cnn 0.1059 0.6885440
0.0781 0.6560338
0.1391 0.7210542
Election LM + Sex + Skin-Tone 0.0170 0.6164495
0.0172 0.5818997
0.0582 0.6509994
Election LM + Sex + Skin-Tone + P_hat_cnn 0.1018 0.7012661
0.0896 0.6691915
0.1534 0.7333407
Election LM + Sex + Skin-Tone + MTurk 0.0177 0.6186759
0.0188 0.5842512
0.0646 0.6531005
Election LM + Sex + Skin-Tone + MTurk + P_hat_cnn 0.1028 0.7051144
0.0936 0.6732405
0.1572 0.7369884