This repository provides our model forecast and outputs for 2022 U.S. governor elections.
Our forecast is inspired by previous models on forecasting presidential elections 1,2. The model assumes each state has day-to-day changes that vary a set of factors: polls, pollsters, polling bias in each state, polling methods, and voter types. It starts with the last election outcomes and moves forwards to predict vote share between the two major parties. We use polling data from Fivethirtyeight and convert it to two-party vote share.
We organize these in the folder forecasts as follows:
forecast-governors.Rmd: the main source file
forecast-governors.md: the github version of the main source file (you can click on this file to see the code rendered on Github)
forecast-governors_files: the files needed to render the md on Github
model-output folder: it contains the election day prediction every time we run the model
stan folder: it holds the stan source file and the compiled stan model
data folder: it holds two other datasets we use: cook report and Fivethirtyeight's prediction on Sep 1