Absentee Online Voters in the Northwest Territories:Attitudes and Impacts on Participation

Abstract

Despite being deployed in Canadian municipal elections since 2003, online ballots were not used in binding elections at higher levels of government until the Northwest Territories’ adoption of online voting for absentee voters in its territorial elections in 2019 and 2023. Municipal and Indigenous use of online voting in Canada are well studied, but implementation at higher orders of government have not yet been examined. Drawing on an original data set of online voters in the 2023 Northwest Territories territorial election, we examine who votes online in higher order elections, attitudes towards the voting mode, and its impact on engagement. Throughout our analysis, we simultaneously compare these data to original data from online voter exit surveys conducted during the 2022 Ontario municipal elections. We find that uncommitted voters outside of Yellowknife would not have voted without the online option. Similarly, for municipal voters, we find that age and past voting record correlate with whether the online option influenced electors to cast a ballot.

1 Introduction

For more than two decades, Canada has been a hotspot for online voting use. But, until recently, online voting activity, and research studying it, has solely focused on community elections in Canada: municipally and among Indigenous communities. In the past several years, however, higher orders of government, including some territorial and provincial election agencies, have either adopted online voting for special groups of voters (e.g., absentee voters) or undertaken plans to do so (e.g., for military service) [1]. In 2022, for example, the Yukon introduced online voting in its school board elections, while in 2019 and 2023, the Northwest Territories (NWT) offered online ballots to absentee voters in its territorial elections. Using online voting in higher order elections prompt questions about its efficacy and uptake beyond the municipal arena, including whether voters report similar attitudes towards voting online. The unique geography and infrastructure of the NWT also raise questions about the potential of online voting to engage electors who might not otherwise have had the capacity or willingness to vote. While studies of online voting in Canada have examined the attitudes of voters, candidates, and administrators towards the local online voting experience [1 − 4], there has yet to be an examination of online voter attitudes in a higher-level government election or a comparison of provincial/territorial voter attitudes to municipal voters. Additionally, while research has probed whether online voting engages less committed voters [5] it has not done so in the context of such a geographically dispersed territory.

Unlike other Canadian provinces and territories, the NWT is an interesting case given the rural and remote character of some parts of the territory, which can be characterized by weaker broadband and physical infrastructure [6]. The territory can often encounter challenges delivering election materials, including paper ballots, to communities. Several communities can only be accessed by airplanes, and some are inaccessible when river crossings are not solidly frozen in the fall months. In addition, the northern location of the territory means it can experience unpredictable weather, including heavy fog and blizzards, which can complicate ballot and election delivery. Relying on mail voting as an alternative can also prove unreliable given delays in mail delivery due to inclement weather and lack of road infrastructure. In fact, the territories are the only locale in Canada where the national postal service does not guarantee delivery times. This unique rurality presents a challenge to delivering both paper-based and online elections.

To understand voter perceptions of online voting in the context of a higher-level (territorial) election and its potential impact on engagement, this article draws upon a unique data set of persons who registered to vote online as part of the absentee voter program in the 2023 NWT territorial election. This is the first time such data has been collected in Canada, thereby contributing significantly to the study of online voting implementation in the country. We consider who is applying to vote online, including age, geographic location, and digital profile. We also look at respondent rationales for casting an online ballot and attitudes towards the voting mode. Finally, we consider online voting’s potential to engage electors. As a point of comparison, we draw upon original data collected from online voters during the 2022 Ontario municipal elections where voting by internet is open to the general population in select municipalities. Online voter attitudes in municipal elections are well studied [1, 5, 7] and this comparison allows us to establish whether there are differences in voter perceptions territorially and municipally.

Drawing upon this data, the article answers three important questions: (1) who is voting online in the NWT, (2) how do online voter attitudes compare in municipal and territorial elections, and (3) what factors predict non-voting without an online option? Answering these questions is imperative to understanding the effects that order of government may have on who votes online, voters’ attitudes towards the franchise when it presents novel opportunities to access the ballot box, and whether it can encourage voting in rural and/or northern contexts.

2 Literature Review

2.1 Online Voter Characteristics and Reasons for Use

The introduction of online voting in Canada was followed by a push to understand the characteristics of its users. Little has been written on the demographic characteristics of “an online voter” in Canada [8], with notable exception of Goodman et al.’s (2024) recent study of Ontario municipal elections, which finds that online voters tend to be older homeowners with higher incomes who often sit left-of-center on the political spectrum. Interestingly, these findings mirror the characteristics associated with other convenience voting methods [9], including mail-in and telephone voting.

In studies of online voting in Canada, including in municipal [8] and Indigenous voting contexts [10, 11], convenience is often cited as the primary reason for voting online. This implies that ease of voting holds considerable importance for uptake because the online option can reduce the perceived costs of casting a ballot [12, 13]. This is particularly relevant in NWT, because the territory’s rurality can present serious obstacles to accessing ballots. As a consequence, NWT voters may place a heightened importance on ballot accessibility or convenience when deciding to vote.

In terms of attitudes, studies have found that online voting in Canada has, generally, had a positive effect on satisfaction with local democracy [1], and that it correlates with strong support for use at all three levels of government. In fact, studies have shown that municipal online voters report being likely to recommend the voting mode to others and have few concerns with its security [3, 15]. Voters’ satisfaction is relevant because governments typically hesitate to enact changes without public backing, as they are expected to consider voter preferences and have, historically, been responsive to them [14]. For online voting to be adopted more broadly [15], it is reasonable to believe that electors must look favourably upon the franchise.

2.2 Engagement

There are mixed results regarding the effect of internet voting on turnout in the literature. Some studies find little to no effect [16 − 18] while others paint a more optimistic picture of the voting mode as a possible solution to counter declining turnout [13, 19]. This optimism stems from its ability to be conducted remotely, reducing the need for physical presence at polling stations [13], and its support of equality in the voting process by enabling older individuals or individuals with disabilities to vote independently [20].

Within this vein, there is a second strand of research on internet voting and engagement focused on who decides to vote by internet. On the one hand, there are studies which argue that voting online is a tool to make voting easier for the already engaged [21]. Conversely, however, studies across multiple jurisdictions over time point to the ability of online voting to bring less committed voters into the voting process [5, 7, 22, 23]. Studies of Switzerland [22, 24], Estonia [23, 25], and Canada [5, 7] all document the ability of internet ballots to encourage voting among abstainers, occasional voters, or both. The theoretical reasoning behind both strands of research is similar: making voting easier and reducing the costs or barriers to accessing ballots can encourage certain voters to turnout [9, 19].

Since the perceived costs of voting are not the same for everyone, there are certain situations where we might expect internet voting to reduce voting costs more significantly. The time needed to travel to a polling location may be longer in rural areas where there are fewer polling locations. Likewise, in the case of the NWT, receiving and returning a ballot from a remote community with poor mail service may make online voting additionally appealing. Remote or convenience voting modes have been regarded in the literature as having the potential to make voting easier for groups of voters that face additional barriers to accessing the ballot box [27].

Looking to absentee voters - a group that often depends on remote voting - results have also been mixed. Work by Berinsky et al. (2001) [9], for example, suggests that alternative voting options for absentee voters, including vote-by-mail, do not improve voter turnout, but rather make voting more convenient for those already likely to vote. However, there is optimism regarding the potential of online voting to reduce barriers and positively affect participation for other special groups of voters. Germann (2021) [19], for example, finds that the extension of online voting increases expatriate turnout among Swiss voters. Likewise, in their assessment of internet voting deployment for citizens abroad in French consular elections, Dandoy and Kernalegenn (2021) [38] find that voting mode choice affects participation. These results are encouraging for deployments like in the NWT. Other Canadian research has shown that internet voting can be a tool to engage voters with less habitual voting histories [5, 7]. There is reasonable expectation to assume that a similar result could extend to the NWT where voting costs may be greater [12, 26] due to its remoteness and poor mail service.

Variables of Interest.

Based on the literature, several variables stand out as potential predictors of electors’ uptake of online voting. First, as noted above, voting history is a key variable of interest [5, 7]. Having a disability could also increase voting costs and therefore encourage voting by internet [20]. This supposition is supported by studies which document that internet voting can make the voting process more equal by canceling out mobility challenges or the need for assisted voting [28]. Third, rurality, and the perception of unreliable mail-service service, is another factor that may drive some voters to cast a ballot remotely online. Studies have shown that distance to the polls can have a marked impact on participation [29] and that convenience voting can be most helpful for voters on the margins [30].

Fourth, experience using the internet could affect an individual’s ability to cast an online ballot [11]. Finally, age is a key variable in studies of voting by internet that could predict whether voters would have cast a ballot with or without the voting mode [1, 7].

After looking at who is voting online and their attitudes towards the voting mode, we consider the effect of these variables on engagement, measured by self-reports that respondents would not have voted otherwise in the 2023 NWT territorial election. Where possible, we compare these results to Ontario municipal voters.

3 Background and Context

3.1 Northwest Territories: Context

Spanning over 1.3 million square kilometres of northern Canada, the Northwest Territories is home to 45,000 residents in 33 communities, a third of which do not have permanent roads, and are dependent upon air travel, river transport, or winter ice roads. Most post-secondary students must travel out of territory for schooling and an increasing number of residents remain but spend their winters in the south.

Many areas of the territory experience challenges with internet access and quality. According to the NWT Bureau of Statistics’ 2019 Community Survey, 16% of households reported not having home internet, either through a home internet service or a cellular data plan. More specifically, in the capital city of Yellowknife, where half of the territory’s population resides, 7% of households do not have home internet access. This number exceeded 50% of households in most small communities [31]. This is relevant given the expected influence of internet access, usage, and comfortability on an individuals’ likelihood of voting online [32].

The 2023 NWT territorial election was conducted in a unique context, following the wake of an exogenous shock wherein 70% of the territorial population was subject to a mandatory evacuation order due to widespread wildfires. With no legal authority to operate a polling station outside of the Northwest Territories, various alternatives, including an electronic absentee ballot were considered. Unclear whether communities would suffer further damage from the wildfires or the extent to which residents would be displaced, the Legislative Assembly passed An Act to Postpone Polling Day for the 2023 General Election on August 28, 2023 to delay the election past its planned date of October 3rd. The Chief Electoral Officer issued instructions on August 29, 2023 that the displacement or absence of a person from an electoral district as a result of the wildfire did not constitute a change to their place of ordinary residence, even if that residence no longer existed. Fortunately, all evacuation orders were lifted by mid-September. The postponed date of the election was set to November 14, 2023. Voters were informed of the available voting methods through both social and traditional media platforms. Post-secondary students received three emails over the course of the election notifying them of the option to vote online.

3.2 Legislative Context and History of Online Voting

Given Canada’s multi-level governance structure, the authority, decision-making capabilities, and existence of municipalities are determined by provincial and territorial laws. This includes the regulation of municipal elections, with each province and territory having its own legislation governing the process. This means that the legislative authority to trial and use online voting or any electronic voting mode in sub-national elections rests with provincial and territorial governments.

In Canada, only Nova Scotia and the NWT have legislated the option to use online voting for their provincial and territorial elections.¹ Shortly after NWT’s 2015 election, the then-Chief Electoral Officer recommended amending the Elections and Plebiscites Act, 2006 to permit electronic voting for special voting opportunities based on a belief that this is where voters wanted the future of elections to go [33]. The Act was amended in 2018 to allow the option of electronic voting for absentee voters.

3.3 Online Elections: Comparing NWT and Ontario

Online voting was first adopted for use by absentee voters in the 2019 NWT territorial general election. Elections Northwest Territories (Elections NWT) defines an absentee voter as a person who is registered to vote but is “unable to make it to an ordinary or advance poll” (Elections and Plebiscites Act, 2006). While absentee voters may apply to cast an online ballot because they are out of the territory entirely, they could also do so because they are outside of their district within the territory, or simply because they know in advance that they will be unable to attend a physical poll. Absentee ballots can be cast online or by mail. To cast an online ballot, voters were required to register with Elections NWT. Applications were approved if a voter was on the Register of Territorial Electors at the address that corresponded to their identification. In 2019, 489 votes were cast online, representing 3.7% of all ballots cast. This represents a sizable increase from when the absentee program offered the mail option only: 162 ballots were requested in 2011 and 244 in 2015. In 2023, online ballots accounted for 3.5% of all votes cast (375 total votes). There were 395 accepted applications, and 19 cancelled applications due to acclamations in those ridings.

In Ontario, by comparison, online voting is open to all voters in the select municipalities whose local councils adopt the voting mode. In the 2022 municipal elections, for example, online ballots were deployed in 222 of 414 municipalities that ran local elections. Decisions regarding the types of voting methods offered (for example, whether online ballots are offered alongside other voting modes or on their own) are at the discretion of local governments. As such, some cities decide to require online voter registration, while others do not. Because of this autonomy some cities have used online voting for over two decades, while for others it is a new addition. Thus, diffusion of online voting is more widespread across Ontario and uptake is greater given that it is offered to a much larger segment of the voting population. In terms of uptake, among municipalities that offered both online and paper voting in the 2018 Ontario local elections, approximately 57% of all votes were cast electronically [1] while in the case of NWT use represented 3.5% of all ballots cast.

While differences in online voting availability, diffusion and uptake make the samples distinct, the NWT represents the first point of comparison to understand more about online voter attitudes and engagement in higher order elections and how this matches up against municipal data.

4 Methods

The data informing this article come from a survey of persons registered to vote online in the 2023 NWT territorial election. A total of 395 electors registered to vote online (of 20,550 eligible electors) with 375 casting an online ballot. The survey invitation was sent to all persons who registered to vote online, regardless of how they voted.

The survey was programmed and distributed via the Qualtrics interface in English and French and was administered via email between April 9 and April 30, 2024.² An initial invitation was sent out in week one, with three reminder messages following in each of the subsequent weeks. A total of 392 valid email invitations were sent (three bounced back) and 171 persons completed the survey for a response rate of 44%. Respondents ranged in age from 18 to 78 years old, with a mean age of 48 years. There is a similar age distribution among the full population of online voters, which has a mean age of 44 years (see Fig. 1 below), albeit those aged 18 to 24 and 55 to 64 are slightly overrepresented in our sample. Respondents reported on average having “some university” education and an annual household income between $110,001 and $150,000. Geographically, there was representation from across the territory, with 67% respondents residing in the capital city, Yellowknife. About 25% of respondents reside in communities that are regional centres (Hay River, Inuvik, Fort Smith), and 9% in smaller, more rural communities. This geographic breakdown is closely aligned with the population of online voters in the territory (58.4% from Yellowknife, 27.7% in regional centres, and 13.9% from small communities).

We compare the responses of territorial voters with a dataset of municipal voters programmed and administered by Schlesinger Group Ltd during the 2022 Ontario municipal elections. Like the NWT survey, respondents self-selected to take part in the Ontario survey, but were recruited immediately following their online voting experience (rather than several months after voting, as was the case in NWT). Once voters had cast a ballot and exited the online voting system, they were prompted with a thank you screen that invited them to take part in a survey about their voting experience.³ Surveys were available for completion during the voting period of each municipality, which, in total, lasted from September 26 to October 24, 2022 across the 24 municipalities that took part in the study. The survey was completed by 29,284 voters for a response rate of 16%. In terms of geography and size, participating municipalities were located across the province: 42% are considered small with fewer than 10,000 electors, 46% are medium sized, with electorates between 10,000 and 99,000, and 13% are large municipalities, with more than 100,000 electors.

Both the NWT and Ontario surveys were written and conducted in similar ways. Recruitment for both surveys was conducted online and Letters of Consent were compiled based on the same template. Both surveys were programmed using online platforms and had similar question sequencing with many identical questions. These included questions probing perceived satisfaction with voting online, reason for use, concerns, desirability to see the voting mode used in the future, trust, perceived risk, digital access and literacy, voting history, and standard socio-demographic items. The NWT survey also asked whether respondents had voted online in the 2023 territorial election since the survey invitation was extended to everyone who registered but may not have voted online. As such differences in data collection across the samples was minimal, although the memory of municipal voters may have been fresher.

Differences in response rates could be attributed to the fact that NWT respondents were sent an initial request and three reminders while Ontario municipal respondents only had one opportunity to complete a survey. The NWT response rate after the initial message was 19%, which is closed to the 16% in Ontario. Overall, we believe these differences are minimal and that the samples are appropriate for comparison.

In terms of limitations, all data is self-selected, and most respondents are voters (all Ontario municipal respondents and nearly all NWT respondents). While not necessarily representative of the total populations, there are similarities in the age and geographic location of NWT voters as described above. Comparable data is not available for Ontario, however, past comparisons suggest the age distributions of exit surveys and actual online voters are similar [3]. Additionally, it is possible that past turnout was over-reported based on social desirability [34]. This should be considered when interpreting the findings below.

5 Results

5.1 Who is Voting Online as an Absentee Voter in NWT?

First, we consider who votes online in a territorial election by looking at age and geography. We then turn to survey items on digital literacy - an important consideration given the rurality of the territory and the digital infrastructure challenges faced by some residents.⁴ Note here that digital literacy refers to voters’ capacity – or perceived capacity – to use online systems and/or digital technologies, including computers and other internet-enabled devices and platforms. Data on age and geographic location was collected by Elections NWT for all online voters. Figure 1 compares the percentages of all NWT voters (online and paper) in the 2023 election with online voters and NWT survey respondents by age group. Overall, we see that proportionately young people aged 18 to 34 were more likely to cast an online ballot than a paper one. Likewise, voters aged 45 + were proportionately less likely to opt for the online option compared to the traditional paper ballot. Interestingly, voter data from Ontario municipal elections overtime has shown, to the contrary, that older voters aged 50 + are the most likely users of online ballots [1, 3, 7]. In the case of absentee voters, however, it could be that younger people away at school or middle-aged voters traveling for work or busy with their families may make better use of online voting than when it is offered to the entire electorate. The limited post-secondary opportunities in the NWT support this line of reasoning. Overall, persons of all ages are voting online as absentee voters in the NWT: the youngest online voter in the 2023 election was 18, while the oldest was 95.

In terms of geography, there was a fairly even distribution of ballots cast across the territory: 36.4% of all votes were cast in Yellowknife districts, 30.4% in regional centres, and 33.2% in small communities. By comparison, 58.4% of online ballots were cast for Yellowknife districts, 27.7% for regional centre districts, and 13.9% for small community districts. This tells us that online ballots were used by voters across communities of varying urbanity and infrastructure, although, the voting mode was most heavily drawn upon by residents of the capital city. This is unsurprising given the improved digital infrastructure and access to travel in Yellowknife. Overall, persons of all ages and geographies in the territory voted online as absentee voters.

Fig. 1.

Percentage of all NWT voters, online voters, and survey respondents by age group

To understand who is choosing to cast an online absentee vote in NWT, we also look at measures of digital literacy, including home internet access, frequency of use, and perceived ability to use the internet. This allows us to understand whether a respondent has direct access to the internet and the quality of that connection, their experience being online, and their confidence using it. We compare these results to responses to the same questions from the Ontario municipal survey.

Despite expecting these items to be lower for NWT voters given the rurality of the territory and challenges with digital infrastructure, we find the opposite is true. All NWT respondents report having an internet connection at home (compared to 1% of Ontario municipal respondents who do not). Furthermore, territorial respondents report having more stable cable connections (81.8% NWT compared to 58.3% in Ontario). With respect to access, 99.4% of NWT respondents in our sample reported using the internet daily compared to 94.8% of municipal respondents.⁵ Conducting a cross-tabulation, Kendall’s tau-b reveals that the difference between community size/urbanity and internet use is significant (tau-b = 0.177, sig. = 0.036). Finally, when asked about their ability to use the internet, a greater percentage of NWT respondents describe their ability as ‘very good’ (66.1%) or ‘good’ (28.5%) compared to Ontario municipal voters (61% and 27.5% respectively).

Taken together, NWT respondents report having better access, more frequent use, and greater confidence in their ability to use the internet than Ontario’s municipal voters, even when considering responses from voters in Ontario’s most urban and tech connected cities. These findings are contrary to the digital profile of the territory [6] and point to online voting attracting absentee voters who are more digitally connected. Online voters outside of Yellowknife do report being less digitally connected, but these differences are marginal at best (e.g., using the internet once a day instead of multiple times daily). It could be that while online voting is a much-needed measure of accessibility for territorial voters, it better reduces accessibility barriers for residents who are already comfortable using technology. While it is possible that less digitally skilled absentee voters could opt for a mail ballot option, data shows this is not the case. Only 48 mail-in ballots were requested by absentee voters, 35 of which were from incarcerated electors who did not have the online voting option. Applications for mail-in ballots have consistently dropped since the introduction of online voting [35], and the small number being applied for, and cast, suggest that less connected residents are not opting for this alternative. Future studies could examine this phenomenon further. It could be that some very rural voters - who have both poor mail delivery and connectivity - simply opt not to vote.

5.2 How Do Territorial Online Voters Feel About Online Ballots? How Does This Compare to Municipal Online Voters?

Respondents of both the NWT and Ontario surveys were asked to identify the main reason they chose to vote online. Note that some categories were collapsed or treated as missing data for comparison.⁶ The striking difference between NWT and Ontario respondents is that Ontario municipal online voters were more likely to have said they voted online for convenience (80.2% municipal versus 28.3% NWT), whereas territorial voters were far more likely have cast an online ballot based on accessibility (6.3% municipal compared to 57.8% NWT) (see Fig. 2). This finding is distinct from all previous data obtained from municipalities and First Nations in Canada [8, 11, 36]. While online voters in municipalities and First Nations likewise experience challenges related to rurality and remoteness, it is possible that these issues are not as severe as they are in NWT. This difference could also be attributed to the fact that online voting in NWT was offered to absentee voters only, while deployments in municipalities and First Nations have been extended to the entire voting population. Twenty-one NWT residents (of 395) applied for absentee ballots outside of Canada.⁷

Aside from differences between convenience and access, there are similarities across the other rationales cited for voting online. ‘Positive past experience’ with the voting mode is slightly higher for municipal voters. This is to be expected given the longstanding use of online voting in Ontario municipalities. Similar percentages of respondents chose ‘found it interesting’ and ‘privacy’ as their primary reason for use. ‘Other’ was modestly higher among territorial respondents, however all but one of these responses pertain to accessibility. These included being outside of the territory or having medical or mobility issues.

Fig. 2.

Main reason for voting online.

Respondents were also asked to identify their top concern with voting online. This question gives us a sense of respondents’ support for the voting mode. Presumably, a high percentage of certain concerns, like security and fraud, could contribute to weaker support for online ballots. The largest proportion of NWT respondents report having no concerns with voting online (37.7%), followed by security (31.1%), ‘other’ (9.6%)⁸ and fraud (8.4%) (see Fig. 3). Lack of internet access only accounted for 5.4% of concerns. On the one hand, this is surprising given the digital infrastructure challenges faced by the territory. But, on the other hand, perhaps it is to be expected given that 67% of respondents reside in in the capital city of Yellowknife, which has high levels of home internet access. Looking at respondents’ geography and their reported top concern, we see that lack of internet access is proportionately identified as a greater issue in regional centres and small communities, albeit only slightly. In fact, differences in concerns based on rurality were minimal overall, though respondents from regional centres and smaller communities were more likely to be unconcerned. Finally, items such as privacy (5.4%) and fear of online voting replacing voting traditions (2.4%) occupied an equally small share of responses.

Fig. 3.

Top concern with online voting.

Ontario municipal online voters, by comparison, had fewer concerns with online voting. Municipal respondents were more likely to say that they do not have concerns about casting a ballot online (47.7%), and fewer respondents identified security as their top concern (25.2%). Most other items were similar across the two samples, apart from fraud, which a greater percentage of municipal voters selected as their top concern (11.7% compared to 8.4% among NWT respondents). The fact that NWT respondents were more likely to report concerns could be related to the newness of online voting in territorial elections. Similar trends in responses are found among municipal respondents whose cities adopted online voting for the first or second time. Often, as the voting mode is made available overtime, concerns tend to lessen.

Considering satisfaction and willingness to recommend online voting to others, we find similarities across territorial and municipal samples (see Table 1). Respondents from both groups report satisfaction with the online voting process (96.4% grouping together ‘very’ and ‘fairly’ satisfied categories). A slightly greater percentage of municipal online voters say they are ‘very satisfied’ compared to territorial voters; however, it is not uncommon for online voters to report slightly lower satisfaction in the first couple of elections it is offered. Similar patterns have been noted among first- or second-time user municipalities [7]. Willingness to recommend online voting to others is similarly strong: 98.2% of territorial respondents indicate they are ‘definitely’ or ‘probably’ likely to recommend online voting to others, compared to 95% of municipal voters. A slightly greater percentage of municipal online voters say they would ‘definitely’ recommend online voting. Again, it is not uncommon to see the percentage of respondents who would recommend online voting grow the longer it is offered. These results suggest that NWT voters are satisfied and feel somewhat favorable towards online voting, if they are willing to endorse its future use.

An interesting difference between the samples, however, is the percentage of voters who say they would not have voted otherwise had online ballots not been offered. Territorial voters are much less certain than Ontario municipal voters to say that they would have voted without the online option. Grouping together ‘probably no’ and ‘definitely no’ responses, we see that 42.8% of NWT online voters say they likely would not have voted without online voting compared to 13% of municipal voters. Earlier studies examining community-level elections in Canada have found that online voting can encourage a modest proportion of electors to cast a ballot [5, 7] by reducing the costs of voting [13]. Studies examining the City of Markham, Ontario showed that, in 2003, this percentage was as high as 25% and then decreased in each subsequent election, leveling off at around 10% to 13% [37]. Likewise, research with First Nations showed that 9% to 13% of respondents said they would not have voted without an online option [11]. The high percentage of NWT online voters that reported they may not have voted otherwise could be explained by the fact that respondents had to apply to vote online because they could not make it to an ordinary or advance poll. These responses suggest that extending this option to voters who need it can positively affect the participation of some. We explore this further by conducting a logistic regression.

5.3 Can Online Voting Engage Some Voters?

We performed a logistic regression to better understand the variables that may predict respondents saying whether they would/would not have voted without the online option. Our dependent variable is whether respondents would have voted without online voting (definitely no/probably no = 1, definitely yes/probably yes = 0). Our independent variables include age, disability, living outside of Yellowknife (NWT) or rurality (municipal sample), voting record, and frequency of internet use.⁹ While not displayed here, we also carried out a confirmatory OLS regression. Table 1 presents the logistic regression results and includes two models: one for the NWT and a second drawing on the Ontario municipal data.¹⁰

Our results show that, for the NWT, living outside of Yellowknife and having an uncommitted voting record (reporting not voting in all elections) predict the likelihood of not having voted without the online option. More specifically, looking at the odds ratios, we see that if a voter is from outside of Yellowknife, there is a 128% increase in the odds that they would not have voted had online voting been unavailable to them. With respect to reporting an uncommitted voting record, the ability to predict not having voted without online voting is even greater. There is a 219% increase in the odds that a voter would not have cast a ballot otherwise.

Table 1. Factors predicting reported not voting without the online option.

The fact that voters outside of Yellowknife have a greater likelihood of not voting without the online option may be explained by the fact that residents outside of the capital may have even less faith in the postal service. While some regional centres are relatively urban, they do not have the same infrastructure and services as Yellowknife. Certainly, for smaller communities, some of which are without main roads, timely access to ballots and the ability to have them promptly returned to be counted in the final tally may be a greater concern. A history of uncommitted voting among these respondents could reflect the additional barriers some face casting an absentee ballot. Online voting could reduce some of the perceived costs of casting an absentee ballot, making participation easier and therefore enabling persons who may not have voted otherwise [12, 13].

Conducting the same analysis with the Ontario municipal data for comparison, except that we use rural as the reference category instead of living outside of Yellowknife, we see a similarly strong result for past voting record. There is a 270% increase in the odds that voters with uncommitted voting records will vote because of the online option. While rurality is not significant in the municipal model, age is. We can interpret this to mean that for each additional year older a person is, the odds of having not voted without the online option decrease by 15.8%. This is an interesting result that future research could probe further. While older voters are the primary users of online voting municipally, young people typically report less committed voting histories. The reduced costs of casting an online ballot compared to a paper one (e.g., not having to travel to a poll location) could again reduce the net effort of casting a ballot and thus make voting more feasible.

Overall, while there are differences in the results across the samples - living outside of Yellowknife matters in the NWT sample, whereas age is significant in the Ontario municipal sample - reported voting history increases the odds of not having voted without online voting in both cases. This points to the potential of online voting to facilitate voter participation in both the context of absentee voters territorially and the full electorate municipally. While these findings do not directly comment on voter turnout per se, they are aligned with previous studies which show that online voting increases turnout among special groups of voters (i.e. expatriates and voters abroad) [26, 38] and that habitual voters are less likely to vote online [6].

6 Conclusion

This article has examined the first-ever collected data from online voters in a territorial election in Canada. Where possible, comparisons are made to data obtained from a survey of Ontario voters in the province’s 2022 municipal elections. The study produces three main insights. First, it provides an understanding of who is opting to cast an online absentee ballot in territorial elections. This includes residents of all ages from a range of communities. A greater percentage of online voters are from the capital city of Yellowknife and proportionately young people aged 18 to 34 and 55 to 64 are more likely users of the voting mode. Absentee online voters in NWT have strong digital literacy¹¹ measured by reported access, frequency of use, and self-reported confidence in their ability to use the internet. Their digital profiles are stronger than municipal online voters despite living in one of the most rural and remote locations of Canada, where many areas have poor digital infrastructure.

Second, we find that territorial respondents are relatively supportive of online voting. They report being satisfied with the online voting process, are willing to recommend it to others, and while they have some concerns related to security and fraud, the largest proportion of respondents say they have no concerns. However, distinct from the Ontario municipal sample and previous findings on online voting use in local government [4, 9] and First Nations elections [14, 36], NWT respondents are much more likely to cite accessibility as their reason for voting online and correspondingly, are much less likely to attribute their use to convenience. In addition to differences in the rationale for use, another key distinction is the percentage of respondents who claim that they would not have voted if online voting was not an option. Previous research has found that, on average, about 10 to 13% of respondents indicate that they would not have voted otherwise. In the NWT sample, 43% say that they would not have voted without it. This suggests online voting has the potential to increase civic engagement among absentee voters in the territory.

To examine this further, we looked at which variables could predict this sentiment. We find that uncommitted voting records and living outside of Yellowknife predict the likelihood that respondents would not have voted otherwise. The Ontario municipal sample also finds that an uncommitted voting record is a significant predictor as well as age. This finding is consistent with previous research on municipal voters in Canada [6].

Taken together, these findings point to online voting being a positive voting channel for electors seeking to cast an absentee ballot in the NWT, especially since it can increase the likelihood of voting among residents whose districts are outside the capital and who may not have voted otherwise. Despite many differences between the territory and the centrally located province of Ontario, findings continue to point to support for online voting in areas it is used across Canada.

This study suggests several directions for future research. First, studies could examine why the digital profiles of online voters in the NWT are so strong despite having some of the weakest infrastructure and costliest internet services in the country. Where are the absentee voters with weaker digital access and skills? Are some potential voters being excluded based on internet availability, cost, or literacy? If so, what measures could be taken to enhance access? A second avenue for future research is to further investigate the extent to which online voting may motivate certain young people to vote. While research on online voting in Norway [39] and voting machines in the Netherlands [40] have documented that young people are less likely to cast an electronic ballot, does this differ based on voting record or intention to vote? Third, studies could examine the challenges of voting online in rural and remote communities. This could be especially interesting where online voting is not limited to certain subsets of the population. Finally, future work could consider the impact exogenous shocks such as the wildfires on online voting uptake [5].

Practitioners and policymakers consulting this work should not underestimate the potential of online voting to engage less committed voters outside of urban centres. As a best practice, election agencies and administrators should be open to working with academics to collect data on future deployments. Collecting data and probing the aforementioned questions could improve voter access to digital voting modes.