Looking at the data

Everyone has an opinion about politics and thoughts about what each party is working on. JustTheData focuses on what the numbers have to say.

The primary goal of this project is to introduce current data to the conversation about Canadian political rhetoric and media's coverage of that rhetoric.  This website is a work in progress and will begin with data obtained from Hansard for the 41st and 42nd parliament as well as media's online coverage of the 42nd and 43rd election. Other data will become available as it is created.

Creating data on political rhetoric and media coverage requires gaining an understanding of the language and messaging used by Canadian political parties as well as the issues important to each party. This is done by mining Hansard, party platforms and Twitter feeds. Creating data on media coverage requires understanding how media covers those issues, primarily through Twitter but eventually through other mediums as well.  Data for media will initially be obtained by scraping Twitter feeds of media outlets and may eventually be expanded to more traditional mediums like web pages and print. The outcome is data on political issues and messaging as well as a media index scoring.  

Media index scoring will then be used to measure the impact of media’s coverage on the voting intentions of Canadians using statistical analysis.  

This project does not rely on surveys.  Instead, data is derived from a variety of sources including Hansard, Twitter and Facebook. Unlike studies by American academics in the early 2000’s, who often required an equal balance of Democratic and Republican research assistants to ensure the neutrality of their studies, this project removes the potential for bias by relying heavily on text mining packages to identify important issues and key phrases used by political parties. Once these issues and phrases have been identified, simple coding can be used to determine the overlap in political news coverage.  Finally, several packages to measure tone are used, including a machine learning tool developed for online use.  The measure of tone does not require any human intervention beyond the coding required to develop and make use of the tone analysis package.  Methodology for deriving data can be found in the methodology section.

Methodology will be made publicly available along with data files. All presentation will be free from personal opinion and conjecture. That is, you'll get just the data.