Key Articles That Built the Model

  • Summary of the 1998 Minnesota Gubernatorial Race

  • Analysis of the New Jersey Primary Ballot Line Phenomena

  • Analysis of the Power of Political Endorsements

  • Analysis of the Power of Party Endorsements on the Ballot Line

  • More Analysis of the Power of the Party and Endorsements

  • Analysis of the Power of Celebrity Endorsements

  • Analysis of the Power of a Trump Endorsement

  • Explanation of how the old 538 model worked

  • Analysis of the importance of scandal in elections

  • Analysis of the “invisible primary” in primary elections

Notes on the model

  1. Questions are open to interpretation. How does one determine if someone has “It-Factor,” for example? This will inevitably yield individual bias - however, this is true of any predictive model.

  2. In the case of multi-way races, you must do each candidate head-to-head to get an accurate result, as there is no built-in function for this right now. I encourage putting those head to head results in a Bradley-Terry model for an accurate analysis.