Interest of Canadians in Internet Voting (2004, 2006, 2008 and 2011) – Research Note
Elections Canada (EC) has an interest concerning Internet voting in view of its strategic objectives for research into new voting means, as well as provisions from section 18.1 of the Canada Elections Act (2001).Footnote 1
Since the 2004 general election, Elections Canada has been polling Canadians on Internet voting by means of questions provided to the Canadian Election Study (CES)Footnote 2, as well as its own post-election survey of electors. Furthermore, since 2008, its Survey of Candidates has also included a question on the principle of Internet voting.
The agency has used an incremental approach in developing survey questions on Internet voting. The questions have been gradually fine-tuned to obtain Canadian electors' opinions regarding the following three areas of interest:
- likelihood of voting over the Internet
- perception of risk associated with Internet voting
- principle of Internet voting
With respect to electors, this research note sets out the results obtained for all three areas of interest. The results also report on the impact of three socio-demographic variables: age group, education and employment status.Footnote 3
With respect to candidates, the analysis looks exclusively at the principle of Internet voting and takes into account the impact of having been elected or not.
The presence / absence of statistically significant association between our variables is calculated using the Pearson Chi-Square (X2).Footnote 4 Regarding the measurement to determine the strength and direction of these relationships, we use the Gamma (γ)Footnote 5 and Cramer's V.Footnote 6 The choice of these tools is based on the very nature of our dependent variables (i.e.: ordinal or nominal / dichotomous) and the objective is to identify avenues of research that could lead to the development of more advanced studies on the state of relations and inter-related interests of Canadians toward Internet voting.
Return to source of Footnote 1 Section 18.1 provides that "[t]he Chief Electoral Officer may carry out studies on voting, including studies respecting alternative voting means, and may devise and test an electronic voting process for future use in a general election or a by-election."
Return to source of Footnote 2 The CES is the only longitudinal study dealing with Canadian federal elections. Introduced in 1965, this university study consists of more than 300 questions related to various aspects of elections. Elections Canada has been involved with the CES since the 1997 general election.
Return to source of Footnote 3 It is worth noting that the choice of socio-demographic variables for this analysis is based on frequencies obtained in the different surveys selected for this research note. Some variables or categories of individuals (e.g. students, Aboriginal people and immigrants) were not included in the analysis because of an insufficient number of respondents.
Return to source of Footnote 4 Pearson's Chi Squared test (X2) can confirm the presence of a statistically significant association (i.e.: by rejecting the null hypothesis), generalizable to the entire Canadian population when "p" is less than 0.05.
Return to source of Footnote 5 Gamma (γ) ranges from – 1 (negative association) to 1 (positive association); 0 means no association. Regardless of the direction, the strength of the association is considered weak between 0 and 0.25; moderate between 0.25 and 0.50; strong between 0.50 and 0.75; and very strong between 0.75 and 1. Since Gamma is a measure of the Proportional Reduction of Error (PRE), the absolute value of Gamma makes it possible to indicate the extent to which knowing the independent variable reduces the error that would be made in predicting the values of the dependent variable compared with those obtained by chance (e.g. heads or tails).
Return to source of Footnote 6 Cramer's V indicates a weak association between 0 to 0.10; moderate association between 0.10 and 0.25; strong association when above 0.25 and perfect association when it reaches 1.0. The Cramer's V does not indicate the direction of the association since the variable being assessed is nominal (i.e.: there is no natural order between categories).