This article introduces a new supervised machine learning approach that links campaign contributions directly to roll call voting behavior. Unlike unsupervised methods, it maps donation patterns onto actual legislative votes.
Methods & Data
The study uses advanced supervised algorithms rather than statistical factor analysis of giving data.
Key Findings
* Supervised models exceed traditional ideology measures in predicting future votes.
* Fundraising before election performance is a strong predictor of post-election behavior.
* Pre-office fundraising forecasts match the accuracy of early voting records for new legislators.
Political Significance
The findings resolve debate about campaign contributions' ability to distinguish between ideologically similar elected officials. These results demonstrate contribution data's independent power in predicting partisan alignment and policy outcomes beyond party membership during USA election cycles.
| Inferring-Roll Call Scores from Campaign Contributions Using Supervised Machine Learning was authored by Adam Bonica. It was published by Wiley in AJPS in 2018. |
