The House Always Wins: House Effects in Polling
In 2007--2008, I collected some data to estimate house 'effects', and to explain them. For instance, some variation across polling houses likely stems from the whether or not a 'don't know' option is provided, the length of the polling period, whether or not the polling house uses a likely or registered voter screen, the question text, etc.
This repository includes some of the data I collected, and may over time include some of the analyses I conducted.
Rasmussen Vs. the Rest
G. W. Bush Approval
2008 Election Cycle, McCain Support
- G. W. Bush Approval Ratings
- From PollingReport.com among other other places.
- [2008 Elections polling data](data/collapsed polls_pollster.csv)
- From pollster.com
- Rasmussen Vs. Rest
- Kalman Filtering
Contribute to the project
If you see an inconsistency in the data, or have a suggestion, or some data that you would like to contribute to the project, please create a pull request or open an issue.
Released under CC BY 2.0.