AK 2024 General — logit
Likely-voter model for AK 2024 GENR, training target election held Nov 5, 2024.
Methodology
Likely-voter model for 2024 GENR, fitted as a logistic regression on voter-history and demographic features from the AK Division of Elections statewide voter file. Trained on n=470,586 eligible voters (target positive rate 57.6%); test ROC AUC 0.8548; Brier 0.1523. Features were standardized (zero-mean, unit-variance). Top positive coefficients: voted_2022_primary (+0.638), voted_2022_general (+0.603), n_prior_generals (+0.279), voted_2020_primary (+0.249), voted_2020_general (+0.235). Top negative coefficients: years_since_last_vote (-7.427), share_polls (-6.130), share_absentee (-4.194), share_mail (-3.224), share_questioned (-2.497).
Standardized coefficients (sorted by |magnitude|)
| Rank | Feature | coef (std-scaled) |
|---|---|---|
| 1 | years_since_last_vote | -7.4267 |
| 2 | share_polls | -6.1298 |
| 3 | share_absentee | -4.1943 |
| 4 | share_mail | -3.2239 |
| 5 | share_questioned | -2.4971 |
| 6 | share_early | -2.3204 |
| 7 | voted_2022_primary | 0.6385 |
| 8 | voted_2022_general | 0.6028 |
| 9 | n_prior_primaries | -0.3822 |
| 10 | n_prior_generals | 0.2791 |
| 11 | voted_2020_primary | 0.2494 |
| 12 | voted_2020_general | 0.2354 |
| 13 | years_at_district | -0.1504 |
| 14 | party_U | -0.1335 |
| 15 | n_prior_municipal | 0.1269 |
| 16 | party_R | 0.1253 |
| 17 | age_at_election | -0.1177 |
| 18 | ever_voted_primary | -0.1001 |
| 19 | n_prior_special | 0.0970 |
| 20 | n_prior_elections | 0.0869 |
| 21 | voted_2018_primary | 0.0858 |
| 22 | years_registered | -0.0778 |
| 23 | pra_flag_int | 0.0668 |
| 24 | party_D | 0.0407 |
| 25 | gender_M | -0.0307 |
| 26 | gender_F | 0.0303 |
| 27 | ever_voted_municipal | 0.0275 |
| 28 | party_OTHER | -0.0263 |
| 29 | voted_2016_general | 0.0233 |
| 30 | party_N | 0.0193 |
| 31 | party_G | -0.0172 |
| 32 | party_L | -0.0170 |
| 33 | gender_U | 0.0066 |
| 34 | voted_2018_general | -0.0057 |
| 35 | ever_voted_general | 0.0052 |
Intercept: 0.5864. Coefficients are in standardized units (features were zero-mean / unit-variance scaled before fit).
Calibration — predicted vs. actual turnout by decile
| Decile (10 = highest LV) | n | Pred. p | Actual turnout |
|---|---|---|---|
| 10 | 58824 | 0.960 | 0.947 |
| 9 | 58823 | 0.938 | 0.944 |
| 8 | 58823 | 0.905 | 0.911 |
| 7 | 58824 | 0.834 | 0.826 |
| 6 | 58823 | 0.656 | 0.634 |
| 5 | 58823 | 0.443 | 0.465 |
| 4 | 58824 | 0.339 | 0.424 |
| 3 | 58823 | 0.272 | 0.303 |
| 2 | 58823 | 0.235 | 0.178 |
| 1 | 58823 | 0.180 | 0.128 |
Eligible scored voters bucketed into 10 equal-size deciles by predicted probability. A well-calibrated model has predicted p ≈ actual turnout in every decile.
Hyperparameters
{'C': 1.0, 'seed': 42, 'max_iter': 200, 'test_size': 0.2}
Trained on AK Division of Elections statewide voter file snapshot AK DoE 2025-05 at 2026-05-23 04:36 UTC.