This markdown contains table 01s for elections with updated MTurk labels. These now come from a side-by-side survey which is much closer to the survey conducted in Todorov et.al. We also updated the CNN to take care of the trainval-test split errors we found last week. All other variable definitions are the same as before. There are two tables here:
Table 01 Configuration 4 (using own-party-vote-share-jacknife in the election_lm) with updated Mturk features in the same model
Table 01 Configuration 4 but with each MTurk feature entred as a univariate model to check for each individually
Before jumping into regression output, I will put all definitions of important terms here:
Vote-Share : In all tables below, when vote-share is a LHS variable this means the vote-share of a specific candidate election. I.e. how much of the total vote did a specific candidate get in their election.
Own-Party-Vote-Share-Jacknife : This is a feature and computed as the “average vote share of my party in this state over all years excluding the current”
Replacing own_party_win_rate with own_party_vote_share_jacknife in the election lm. This now includes:
Own Party Vote-share-Jacknife 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 | Election Outcome | Vote Share | |
| Adjusted R Squared | ROC AUC | Adjusted R squared | |
| Single Variable Model | |||
| Election LM | 0.0667 | 0.6421 | 0.0742 |
| Lower 95% C.I. | 0.0468 | 0.6115 | 0.0537 |
| Upper 95% C.I. | 0.0920 | 0.6726 | 0.0986 |
| Sex | 0.0037 | 0.5183 | 0.0048 |
| 0.0005 | 0.4962 | 0.0007 | |
| 0.0098 | 0.5404 | 0.0127 | |
| Skine-Tone | −0.0077 | 0.5412 | −0.0022 |
| −0.0032 | 0.5098 | 0.0009 | |
| 0.0172 | 0.5726 | 0.0262 | |
| MTurk Features | 0.0127 | 0.5581 | −0.0003 |
| 0.0057 | 0.5264 | −0.0018 | |
| 0.0290 | 0.5898 | 0.0098 | |
| P_hat_cnn | 0.1011 | 0.6858 | 0.0804 |
| 0.0774 | 0.6566 | 0.0578 | |
| 0.1289 | 0.7149 | 0.1076 | |
| Combined Variable Model | |||
| Election LM + P_hat_cnn | 0.1579 | 0.7323 | 0.1469 |
| 0.1299 | 0.7047 | 0.1172 | |
| 0.1915 | 0.7598 | 0.1783 | |
| Election LM + Sex | 0.0683 | 0.6443 | 0.0780 |
| 0.0476 | 0.6139 | 0.0563 | |
| 0.0933 | 0.6748 | 0.1040 | |
| Election LM + Sex + P_hat_cnn | 0.1594 | 0.7342 | 0.1504 |
| 0.1317 | 0.7068 | 0.1242 | |
| 0.1932 | 0.7616 | 0.1845 | |
| Election LM + Sex + Skin-Tone | 0.0613 | 0.6577 | 0.0758 |
| 0.0532 | 0.6276 | 0.0644 | |
| 0.1013 | 0.6878 | 0.1185 | |
| Election LM + Sex + Skin-Tone + P_hat_cnn | 0.1540 | 0.7394 | 0.1492 |
| 0.1353 | 0.7123 | 0.1333 | |
| 0.1987 | 0.7666 | 0.1998 | |
| Election LM + Sex + Skin-Tone + MTurk | 0.0731 | 0.6699 | 0.0742 |
| 0.0652 | 0.6403 | 0.0674 | |
| 0.1172 | 0.6995 | 0.1222 | |
| Election LM + Sex + Skin-Tone + MTurk + P_hat_cnn | 0.1575 | 0.7432 | 0.1465 |
| 0.1424 | 0.7161 | 0.1300 | |
| 0.2047 | 0.7702 | 0.1964 | |
In this table I decide to include the different MTurk features individually in the single variable model section. We can see that the significant signal is coming from Competence and Trustworthiness. Of note here is that Competence is the feature isolated as significant in Todorov et.al (2015). This is very exciting news, since our CNN is still able to isolate significant signal above these features.
| Table 01 - Version 04 - Election Regressions - Individual MTurk Features | |||
|---|---|---|---|
| Fit measured in adjusted R squared and AUC | |||
| Model Configuration | Election Outcome | Vote Share | |
| Adjusted R Squared | ROC AUC | Adjusted R squared | |
| Single Variable Model | |||
| Election LM | 0.0667 | 0.6421 | 0.0742 |
| Lower 95% C.I. | 0.0463 | 0.6115 | 0.0529 |
| Upper 95% C.I. | 0.0888 | 0.6726 | 0.1003 |
| Sex | 0.0037 | 0.5183 | 0.0048 |
| 0.0007 | 0.4962 | 0.0006 | |
| 0.0099 | 0.5404 | 0.0127 | |
| Skine-Tone | −0.0077 | 0.5412 | −0.0022 |
| −0.0038 | 0.5098 | 0.0004 | |
| 0.0162 | 0.5726 | 0.0262 | |
| Attractiveness | −0.0005 | 0.5143 | −0.0007 |
| −0.0008 | 0.4824 | −0.0008 | |
| 0.0035 | 0.5463 | 0.0022 | |
| Competence | 0.0027 | 0.5368 | 0.0008 |
| −0.0006 | 0.5050 | −0.0008 | |
| 0.0105 | 0.5687 | 0.0066 | |
| Dominance | −0.0008 | 0.4912 | −0.0008 |
| −0.0008 | 0.4592 | −0.0008 | |
| 0.0021 | 0.5232 | 0.0021 | |
| Trustworthiness | 0.0051 | 0.5425 | 0.0006 |
| 0.0001 | 0.5107 | −0.0008 | |
| 0.0152 | 0.5743 | 0.0060 | |
| P_hat_cnn | 0.1011 | 0.6858 | 0.0804 |
| 0.0750 | 0.6566 | 0.0596 | |
| 0.1302 | 0.7149 | 0.1085 | |
| Combined Variable Model | |||
| Election LM + P_hat_cnn | 0.1579 | 0.7323 | 0.1469 |
| 0.1277 | 0.7047 | 0.1189 | |
| 0.1891 | 0.7598 | 0.1800 | |
| Election LM + Sex | 0.0683 | 0.6443 | 0.0780 |
| 0.0490 | 0.6139 | 0.0570 | |
| 0.0946 | 0.6748 | 0.1036 | |
| Election LM + Sex + P_hat_cnn | 0.1594 | 0.7342 | 0.1504 |
| 0.1324 | 0.7068 | 0.1230 | |
| 0.1910 | 0.7616 | 0.1837 | |
| Election LM + Sex + Skin-Tone | 0.0613 | 0.6577 | 0.0758 |
| 0.0523 | 0.6276 | 0.0643 | |
| 0.1027 | 0.6878 | 0.1195 | |
| Election LM + Sex + Skin-Tone + P_hat_cnn | 0.1540 | 0.7394 | 0.1492 |
| 0.1369 | 0.7123 | 0.1306 | |
| 0.2012 | 0.7666 | 0.1996 | |
| Election LM + Sex + Skin-Tone + MTurk | 0.0731 | 0.6699 | 0.0742 |
| 0.0662 | 0.6403 | 0.0680 | |
| 0.1185 | 0.6995 | 0.1195 | |
| Election LM + Sex + Skin-Tone + MTurk + P_hat_cnn | 0.1575 | 0.7432 | 0.1465 |
| 0.1421 | 0.7161 | 0.1298 | |
| 0.2057 | 0.7702 | 0.1963 | |