Canadian Voting Index–An Introduction


Partisan Voting Index is a concept developed by American political scientist Charlie Cook. While Mr. Cook would definitely not be happy me stealing his concept of partisan voting index to Canada, it’s not going to be effective in a Canadian context anyway. Canada has three major political parties–the Conservatives (or the Tories), the Liberals, and the New Democrats (or the Socialists)–PVI fails to work in a three-party context. Therefore, an alternative method to measure political party support needs to be developed. Therefore, I am introducing the Canadian Voting Index, basing on a slight variation of Mr. Cook’s legendary work.


Canadian Voting Index measures each political party’s support level in a given geographic area, rather than a positive-negative number that measures two party’s support level in an area. Essentially, this is altering from Cook’s methodology, just converting a two-party index into a three-party index. Method of calculating CVI is identical to that of PVI, except in a three-party context.

CVI is calculated by:

\frac{\%_{r1}+\%_{r2}-\%_{n1}-\%_{n2}}{2} (1)

Where \%_{r1} denotes a party’s popular vote in a riding in election 1, \%_{r2} denotes that party’s popular vote in that riding in election 2, \%_{n1} denots that party’s nationwide popular vote in election 1, and \%_{n2} denotes that party’s nationwide popular vote in election 2.

For example, if political party B has 35% of votes in riding A in election 1, 33% of votes in riding A in election 2, 25% of national popular vote in election 1, and 27% of national popular vote in election 2, its CVI is:



Canadian Voting Index could be used to project a political party’s performance in a riding in an upcoming election. For instance, we can apply CVI number to a nationwide poll to project a party’s expected vote share in a riding. Note that CVI is not a comparison between political parties, as that comparison would have no practical application aside from building a statistical table.

Canadian Voting Index could also be used calculate a region’s electoral shift from one election to another, by comparing CVI from different elections.


Creating a CVI table would involve collecting 2011 and 2015 election results for all 338 electoral districts across Canada. 2011 election result needs to be in redistributed format to make such result most accurate. A table based on equation 1 then needs to be computed to output final CVI data for each riding.

Preliminary Data

A CVI table at the provincial level has already been completed. Preliminary findings include huge swings between one election to another in both the provinces and the ridings. For example, Gatineau swung 40 points toward the Liberals from 2011 to 2015. Based on the preliminary data, it might be difficult for a CVI-based model to accurately predict elections in Canada due to the existence of such swings.



Province NDP Cons. Lib.
British Columbia 4.05 2 -4.9
Alberta -10.95 27.4 -12.25
Sasketchewan 3.55 16.65 -12.95
Manitoba -5.35 9.65 1.4
Ontario -4.05 3.95 5.85
Quebec 9 -19.15 -4.25
New Brunswick -1.1 -1.15 7.9
Nova Scotia -1.8 -8.45 16.2
Prince Ed. Il. -9.45 -5.5 20.45
Newfoundland 1.65 -16.4 22
Yukon T. -8.2 -6.85 14.1
Northwest T. 13.15 -10.7 4.15
Nunavut T. -2.2 1.6 8.75

Table 1: Preliminary Data

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes:

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>