Travel Cost on Election Day and Voter-Turnout in Chile: Exploring University Students’ Willingness to Cast a Hypothetical Online Vote

Abstract

Plain Language Summary

An i-voting system may promote electoral participation by reducing travel cost and time to polling places, especially among youth who are more accustomed to technological changes. We study the linkage between travel costs on election day and voter turnout by comparing past electoral participation with stated voting behavior in a hypothetical i-voting system. The main contribution of our paper is threefold. First, it expands the scarce empirical literature that addresses the impact of travel costs on voter turnout. Second, it provides an alternative formalization that links travel costs on election day to voter turnout, which explains how i-voting can reduce travel cost and voting time, becoming a convenient voting channel. And third, it proposes a novel empirical strategy that accounts for underlying factors that simultaneously determine the voting decision in paper-ballot and i-voting elections. Given that our data were collected using an online questionnaire, we also conducted a robustness check to examine if our findings are in line with those reported in national surveys. Binary and bivariate probit estimates show that conventional turnout probability among university students is negatively affected by travel costs on election day. And vehicle availability seems to be important to explain past voting behavior, positively affecting voter turnout. Interestingly, whereas political interest and democracy valuation still augment the probability of voter turnout with i-voting, travel-to-polling-station costs are not statistically relevant.

Keywords

voting behavior travel cost internet voting e-democracy Latin America

Introduction

There is a wide scholarly consensus that the act of voting should be simple and avoid any hurdle or inconvenience that can generate disincentives for voter turnout (Highton, 1997, 2004; Stone, 1985). In that regard, well-established democracies have reformed their electoral systems in order to simplify the act of voting, diminish related costs, and promote electoral engagement (Haspel & Knotts, 2005; Highton, 1997, 2004; Stone, 1985). For instance, the United States has promoted mail-in registration, postponed the closing date, set automatic registration through driver’s licenses, created absentee voting and in-person early voting, and adopted Election Day vote centers (EDVC) in the most populated counties (Highton & Wolfinger, 1998; Knack, 1995; Stein & Vonnahme, 2008; Stone, 1985). But even with those changes, the positive effect on turnout rates has been lower than expected, especially among young people and minorities (Avery & Peffley, 2005; Berinsky, 2005; Brians & Grofman, 1999; Nagler, 1991; Oliver, 1996; Rosenstone & Wolfinger, 1978; Stein, 1998; Stein & Vonnahme, 2008).

More recently, remote electronic voting in the form of online voting, also known as internet voting, has been interpreted as a safe voting mechanism in the face of contagion risks posed by the COVID-19 pandemic (Ehin et al., 2022; James, 2021; Krimmer et al., 2021a; Musiał-Karg & Kapsa, 2021). Internet voting¹ (i-voting) involves diversifying voting channels, captivating those who are unable or unwilling to reach a polling place on election day, which is generally distant from residence (Ehin et al., 2022; Mellon et al., 2017). Moreover, attitudes toward technological change and the internet seem to be critical for the satisfactory implementation of i-voting (Nemeslaki et al., 2016; Powell et al., 2012; Vassil et al., 2016). Therefore, i-voting could be thought of as a plausible way to encourage electoral participation among youth, a population segment with higher levels of both technology acceptance and political apathy (Nemeslaki et al., 2016).

Convenience voting reforms has been emulated by emerging democracies to mitigate an increasing electoral apathy, but those countries have experienced the opposite effect. For instance, in 2012, Chile replaced mandatory voting with voluntary voting and automatic enrolment. Since then, voter turnout has declined sharply, particularly with young potential voters. According to Servicio Electoral (2020), overall turnout remained below 50% for all elections between 2012 and 2017, with electoral engagement rates among youth (ages 18–29) ranging between 23% and 37%.² To a certain extent, this can be explained by the increasing political disaffection reported among youth before the electoral reform was imposed (Acuña-Duarte, 2017; Carlin, 2006; Contreras & Navia, 2013; Corvalán & Cox, 2013; Toro, 2007). An alternative explanation stems from the old electoral system, which imposed a cost (in this case a fine) to those who abstained from voting. This cost was removed after the voluntary election system was put into place. It is interesting, then, to explore to what extent these and other voting costs are driving the low voter turnout rates observed among Chilean youth. Promoting a stronger electoral engagement among young population is crucial to safeguard the sustainability of the electoral roll, encourage the emergence of novel and high-quality politicians, and reinforce the foundations of regime legitimacy in the long-term. We are particularly interested in alternative actions that could reduce costs associated with voting but that do not imply changing the implicit voting bureaucracy. One specific issue we emphasize relates to the role of distance and transportation costs to reaching polling places in explaining this low electoral participation. Variation in this type of costs are not necessarily linked to a kind of “desirable” cost that involve an extra effort time, potentially suggesting more knowledgeable or motived citizens, but this type of costs respond mainly to income constraints that are faced by more vulnerable people (e.g., low income, young, remote areas, commuting workers). Then, lower turnout rates reported for this population group would reduce its representativeness in the political decisions process, relegating their demands and preferences. The latter is not associated with lack of citizens’ education or motivation, but this is simply because of insufficient access and opportunities that an important portion of the electorate faces to afford transportation costs on election day. Therefore, our concern is not only overall turnout, but also make political election more inclusive, especially among more vulnerable populations. We argue that i-voting may reduce this cost barrier and make electoral engagement more affordable. In addition, i-voting could be in line with new paradigms in democracy that promote reforms that give citizens more direct participation in policy decisions, for example, via referendums (Landwehr & Harms, 2020; Seyd et al., 2018). The success of participatory democracy, however, is conditioned on greater political participation, as the more individuals participate the better able they become to seeking higher quality in efficacy and depth (Pateman, 2012).

This paper attempts to advance in the literature by assessing the effect of travel-to-polling-station costs on electoral engagement among Chilean youth in a hypothetical i-voting system. We expect that the i-voting system will increase political participation by reducing travel-to-polling-station costs. This is not an obvious expectation in countries with high political disaffection, especially among youth. The contribution of the paper can be condensed in three main aspects. First, to our knowledge, the impact of travel costs on voter turnout has been scarcely addressed in the empirical literature (see, for instance, Bhatti, 2012; Dyck & Gimpel, 2005; Haspel & Knotts, 2005). Second, this paper provides an alternative formalization that links travel costs on election day to voter turnout by extending the costly voting model outlined by Acuña-Duarte (2017). In doing so, a simple model is proposed to explain voting behavior and how travel costs and journey time interact with placement decision of a new or existing polling station. This model helps to explain how i-voting can reduce travel cost and voting time, and then become a convenient mechanism to encourage turnout rates. Finally, we contribute to the literature by proposing a novel empirical strategy that accounts for the fact that past individual voting decisions and their preferences for voting under a hypothetical i-voting system are linked. To deal with this issue, we estimate an ordered bivariate probit model to account for underlying factors that simultaneously determine these decisions and preferences. Thus, we compare estimates of voter turnout in past paper-ballot elections with stated online voting behavior to explore differences in the relative importance of travel cost variables on electoral participation. Data were collected using an online questionnaire that was disseminated among Chilean university students. We focus on university students because they belong to a population segment that has shown more political apathy and lower turnout rates in the last decade than any other age group, and we expect young people to more readily accept an innovative voting system that requires the use of information technology. Furthermore, university students are more likely to experience higher voting costs as many in this age bracket move away from home for academic reasons.

The rest of the paper is organized as follows: Section “Literature Review” reviews the literature related to travel costs on election day and i-voting. Section “Theoretical Model” outlines the theoretical model that formalizes the linkage between travel costs on election day and voter turnout. Section “Materials and Methods” describes the empirical strategy and data. The main findings are discussed in Section “Results and Discussion,” and a robustness check is provided in Section “Robustness.” Section “Conclusion” concludes the manuscript.

Literature Review

Few studies have paid attention to reducing transportation cost on election day or assessing the mechanism behind how voting sites are selected in order to increase voter turnout. Using voter data from Clark County, Nevada, USA, Dyck and Gimpel (2005) assessed the effect of distance to the polling place on the probability of casting a ballot through three available voting options: absentee voting by mail, in-person early voting, and traditional voting. They found evidence that a larger distance from home to early/traditional voting sites increases the likelihood of choosing absentee voting by mail. Similarly, Haspel and Knotts (2005) used the 2001 mayoral elections held in Atlanta, Georgia, USA to examine the effect of distance and precinct placement on voter turnout. Their results confirmed that changes in the geographical position of a voting site or the distance a citizen must travel on election day have a significant impact on turnout probability. Brady and McNulty (2011) used the modifications to voting places that occurred in 2003 in Los Angeles County, California, USA to identify changes in voting costs. They focused on two costs in particular: a (higher or lower) transportation cost to a new polling place and a search cost for finding the updated voting site. They found that voting costs are negatively and significantly related to voter turnout, and that the search effect was relatively more important than the transportation effect. In addition, the effect of polling-station placement on voter turnout has been addressed in Scandinavian countries. Using individual-level data from the 2009 Danish mayoral election, Bhatti (2012) found that a larger distance from home to polling-places disincentivizes electoral participation; however, this negative effect was mitigated due to car availability. The author also highlighted that increasing the number of polling stations in Denmark would be costly for local governments, especially if the latter experiences financial distress. Thus, the empirical evidence of the effect of transportation costs on turnout is limited and, to our knowledge, scarce in developing contexts. This paper provides additional empirical evidence on settings where economic, social, and political institutions are still underdeveloped, and therefore, transportation costs can be even more relevant to explain electoral engagement.

More recently, i-voting constitutes a convenient alternative that can reduce transportation costs on election day. Unlike the traditional electronic voting (e-voting) system, which requires an electronic device (e.g., an electronic machine voting) to be available at a public facility, i-voting involves online voting through the internet (Nemeslaki et al., 2016). Because of this feature, attitude toward Information and Communication Technologies (ICT) is a key factor for its successful implementation. In fact, Powell et al. (2012) provided evidence that higher levels of trust in the internet and government would increase voting intention among young and senior voters in the Midwestern US. Furthermore, the authors showed that higher computer anxiety reduces the willingness to vote online in both age groups. Nemeslaki et al. (2016) found that the higher expectations for ease-of-use and convenience of an i-voting system would encourage voter turnout among Hungarian higher-education students. Their findings also confirmed the relevance of internet trust on increasing i-voting intent.

Although several countries have desisted after initial trials (e.g., the United States, the United Kingdom), i-voting has proven itself an effective mechanism for increasing voter turnout and a more cost-efficient voting channel in Estonia (Ehin et al., 2022; Krimmer et al., 2021b; Vassil et al., 2016). Estonia was an early adopter of i-voting, which has been implemented for local, national, and Europe-wide elections. According to Vassil et al. (2016), the share of online votes in Estonia increased 15-fold between 2005 and 2015, reaching nearly a third of the voters in the 2015 national election. The authors confirmed that there is a diffusing effect of i-voting on turnout, in which effects are not immediate but need at least three election cycles to manifest. In contrast to Estonia, Swiss i-voting exhibited only intermittent use after its application. Germann and Serdült (2017) found evidence that i-voting was not able to increase turnout rates at the county-level in the cantons of Geneva and Zurich, Switzerland. The authors remarked that postal voting, a convenient voting mode, was available before i-voting was implemented.

i-Voting has also been used in non-electoral processes, for example, participatory budgeting. Mellon et al. (2017) examined the experience of Rio Grande do Sul, Brazil, on a participatory budgeting process held in 2014 through paper ballot and online voting. They found evidence that the probability of using i-voting increased among people with higher levels of education and internet usage. Nevertheless, offline and online voters did not differ in terms of voting behavior. Authors argued that this was because participatory budgeting discussion was isolated from political ideologies. Even though there is previous literature supporting i-voting as an effective mechanism for increasing voter turnout, there is less efforts in the literature to explore the underlying mechanisms that explain this increase in electoral participation. This paper provides both theoretical and empirical evidence of how the importance of transportation vanishes to explain voting under a hypothetical i-voting system. To our knowledge, this is novel in the literature.

There are some security issues related to i-voting, such as a higher risk of election fraud, potential cyber-attacks carried out by malicious users, or an increasing vulnerability in voter anonymity.³ With an i-voting system, someone could buy votes or pressure others to vote for a particular candidate, and can actually verify whether or not they did so, affecting the secrecy of voting (Jefferson et al., 2004; Lauer, 2004; Springall et al., 2014; Wolchok et al., 2012). Addressing these difficulties would undoubtedly increase the cost of election security, but this should not be an individual cost that reduce incentives to vote. Security costs must be funded by authorities jointly with the implementation of the i-voting system, and the former must not constitute a further burden for the electorate. Some incentives have been implemented to deal with these insecurity problems. In the Estonian case, electoral distortions were diminished by permitting voters to vote through the internet as many times as they desire, being just the final attempt officially recorded; and since 2021, internet suffrages are permitted to be annulled by in-person ballots (Ehin et al., 2022; Vinkel & Krimmer, 2017). In addition, the State Information System Authority is an Estonian institution responsible for i-voting security and e-governance solutions, which designed a trusted voting application where voters authenticate themselves using their ID-card and cast an encrypted suffrage. An independent Registration Service assigns a timestamp and a signature to each online vote, which can be decrypted and verified by Estonian voters using a mobile device. At the end of the election, suffrages are conveyed to an offline Ballot Processor unit before performing the vote-counting (see Ehin et al., 2022, for further details).

Theoretical Model

This section extends the model proposed by Acuña-Duarte (2017) in order to formalize the effect of travel costs on the voter-turnout decision, which is defined by the following assumptions: Assume that $n$ rational citizens live together in a democratic society, where voting is voluntary but costly. They have preferences for a consumption good, $x$ , leisure time, $l$ , and democracy, $d$ . As pointed out by Dyck and Gimpel (2005), both democracy and voting are treated as public goods because they cannot be traded or valued under a market structure. In turn, suppose that the supply of democracy is linked to electoral engagement; that is, a higher voter turnout will contribute to a greater legitimacy and stability of the political system. Therefore, we suppose the public good democracy, $d$ , to be a continuous and increasing function of voter turnout, $R$ ; that is to say, $d = f (R)$ .

Following Brady and McNulty (2011), suppose that there are more costs to voting than just registration. These costs can be divided into administrative costs (e.g., registration costs, fines for not voting) and travel cost on election day. Travel costs incur because the act of voting ( $v_{i}$ ) implies time ( $t$ ), which depends on the distance from home to the polling place ( $m$ ).

In addition, each citizen earns a nominal wage, $w$ , for allocating a fraction ( $h$ ) of his/her available time to the labor market. Time constraint for each citizen is then given by: $l_{i} + h_{i} + t (m) v_{i} = 1$ . Therefore, travel time on election day represents an additional opportunity cost since “it takes away from work or leisure time that could be spent on other activities” (Dyck & Gimpel, 2005, p. 535).

Since citizens are rational, they must solve the following problem:

max_{{x, l, v}} u (x_{i}, l_{i}, d)

(1)

Subject to:

x_{i} + (δ + τ) v_{i} \leq h_{i} ω

(2)

l_{i} + h_{i} + t (m) v_{i} = 1

(3)

R = \sum_{j = 1}^{n} v_{j} = v_{i} + \sum_{j \neq i}^{n} v_{j}

(4)

Where: $x_{i} \geq 0$ ; $l_{i}$ , $h_{i}$ , and $t_{i} \in [0, 1]$ ; $v_{i} \in {0, 1}$ ; $R \geq 0$ , and; $d = f (R) \geq 0$ . The budget constraint in real terms is depicted in equation (2), where $ω$ is the real wage, and the real administrative and travel costs of voting are denoted by $δ$ and $τ$ , respectively. The voting decision is represented by $v_{i}$ , so $v_{i} = 1$ if citizen $i$ decides to cast a vote and zero otherwise. Once the first-order conditions are obtained from the above utility maximization problem, the optimal solution can be calculated from the following equation:

\frac{\partial u}{\partial x_{i}} = \frac{1}{ω} \frac{\partial u}{\partial l_{i}} = \frac{1}{(t (m) ω + δ + τ)} \frac{\partial u}{\partial d} f^{'} (R)

(5)

Following Acuña-Duarte (2017), suppose that citizens’ preferences are represented by a Cobb-Douglas utility function: $u (x, l, d) = x_{i}^{ϕ_{1}} l_{i}^{ϕ_{2}} d^{ϕ_{3}}$ ; and the supply of democracy as a public good is given by: $d = f (R) = R^{2} / n$ . Therefore, the voting decision at the individual level is driven by the following reaction function:

\begin{matrix} v_{i} = \frac{2 ϕ_{3} ω}{(t (m) ω + δ + τ) (ϕ_{1} + ϕ_{2} + 2 ϕ_{3})} \\ - (\frac{ϕ_{1} + ϕ_{2}}{ϕ_{1} + ϕ_{2} + 2 ϕ_{3}}) \sum_{j \neq i}^{n} v_{j} \end{matrix}

(6)

For the sake of simplicity, assume that citizens have identical preferences, and the utility function is homogenous of degree one in consumption and leisure. Thus, the Nash equilibrium from equation (6) for any finite population, $n$ , is given by:

v^{*} = \frac{2 ϕ_{3} ω}{(t (m) ω + δ + τ) (n + 2 ϕ_{3})}

(7)

Under this particular equilibrium, the turnout rate will be equal to zero if nobody values democracy (i.e., $ϕ_{3} = 0$ ). In terms of public policy, the government could positively affect voting intent, $v^{*}$ , and turnout rate by reducing the three sources of voting costs included in the model. For instance, automatic enrollment would reduce administrative costs ( $δ$ ), whereas free public-transportation on election day would diminish travel cost ( $τ$ ). Notice that travel costs and journey time are not independent of the placement decision of a new or existing polling station because conventional voting is a time-consuming activity. Thus, i-voting emerges as a convenient mechanism to encourage turnout rates since it eradicates travel costs ( $τ$ ) and reduces voting time ( $t$ ) at its minimum value, making it even more advantageous than postal voting (Germann & Serdült, 2017).

Finally, the above conjectures can be formalized by the following first forward differences:

\frac{Δ v^{*}}{Δ δ} = - \frac{2 ϕ_{3} ω}{(t ω + δ + τ) (t ω + δ + Δ δ + τ) (n + 2 ϕ_{3})} < 0

(8)

\frac{Δ v^{*}}{Δ τ} = - \frac{2 ϕ_{3} ω}{(t ω + δ + τ) (t ω + δ + τ + Δ τ) (n + 2 ϕ_{3})} < 0

(9)

\frac{Δ v^{*}}{Δ t} = - \frac{2 ϕ_{3} ω^{2}}{(t ω + δ + τ) [(t + Δ t) ω + δ + τ] (n + 2 ϕ_{3})} < 0

(10)

\frac{Δ v^{*}}{Δ m} = - \frac{2 ϕ_{3} ω^{2} [t (m + Δ m) - t (m)]}{(t (m) ω + δ + τ) (t (m + Δ m) ω + δ + τ) (n + 2 ϕ_{3}) Δ m} < 0

(11)

Where: $Δ δ > 0$ , $Δ τ > 0$ , $Δ t > 0$ , and $Δ m > 0$ .

Materials and Methods

Data

Data were collected from an online survey conducted by the authors between January 25 and August 6, 2018. The survey was disseminated among Chilean higher-education students through mailing lists from a state university with campuses in Concepción and Chillán, two large urban centers located in the regions of Biobío and Ñuble, respectively, that concentrate a large number of higher education institutions and students from southern Chile. Students’ ages range from 18 to 29 years old.

The survey was composed of several modules with questions that gathered information about socioeconomic characteristics, location variables, labor/student status, household income, parental educational level, political preferences, and attitudes toward democracy. Civic and electoral engagement was addressed in a separate module, including queries related to civic participation activities, voter turnout in the past elections (mayoral, congressional, and presidential), reasons for voting abstention, perceived distance to the polling stations, transportation mode used on election day, journey time, and travel cost. Note that our online questionnaire made use of validated questions included in national surveys conducted in Chile by governmental institutions and foreign research institutes, such as the AmericasBarometer survey and the household survey of National Socioeconomic Characterization (known as CASEN), which guarantees comparability and bias mitigation. For instance, indigeneity and civic engagement were asked using question R3 and R6 from CASEN survey, whereas satisfaction with democracy and political interest were queried with Likert-type questions PN4 and POL1 from AmericasBarometer survey (for details, see LAPOP, 2014, 2017, 2019; Observatorio Social, 2017).

Finally, and in accordance with the stated-preferences approach, respondents were asked about their willingness to participate in future elections held using hypothetical i-voting. In particular, i-voting intention was framed through a binary question requested information about the willingness to participate in future elections held under i-voting in Chile once the corresponding security actions have been taken. We also explored if results change as asked about hypothetical i-voting turnout on several local, regional, and national elections.

A sample of 677 people was obtained, where 55.4% of respondents were women and 9.3% identified as Indigenous. In terms of the sample, a few aspects should be highlighted. First, most of the respondents were enrolled in a higher education institution and lived in the regions of Biobío and Ñuble. This is reasonable since the university hosting this study has campuses in both regions, which conjointly are the second most populated area in Chile. Second, although the respondents exhibited a relatively high level of political interest, they expressed low satisfaction with Chilean democracy and an evident disapproval of political parties. Third, there is a prominent gap in the turnout rate reported in the last two elections held in Chile; however, a plausible explanation is that respondents were not old enough to be enrolled in the electoral register in 2016. Fourth, 12.8% of respondents who abstained in the presidential/congressional election held in 2017 would vote in hypothetical i-voting. And fifth, more than 15% of the respondents were commuting voters: 12% of them needed transportation to another county and 3.6% to another region on election day.

Empirical Strategy

To assess the impact of travel-related costs to polling stations on voter turnout, we first model individual voting behavior through the estimation of a binary probit model, which is formalized as follows:

P (v_{i} = 1 | X, W, Z) = Φ (X β, W π, Z γ)

(12)

Where: $v_{i}$ is defined above as the voting decision and is represented by a dummy variable indicating voter turnout of respondent $i$ in two paper-ballot elections (i.e., 2016 Mayoral and 2017 Presidential elections) and future elections held using hypothetical i-voting.

$X$ is a vector of variables related to political attitudes, including valuation of democracy, satisfaction with democracy, and political interest, all measured on a 4-point Likert scale (e.g., none, a little, some, a lot). Civic engagement is defined as a dummy variable that takes the value of one if respondent i is engaged in volunteerism, charity, religious groups, or similar activities. Dummy variables for left-wing political orientation and apathy toward political parties are also included.

$W$ is a vector of variables linked to transportation on election day. It includes self-reported information about transportation costs in 2018 thousand Chilean pesos, journey time in minutes, and distance to the polling station.⁴ Following Haspel and Knotts (2005) and Bhatti (2012), dummy variables for vehicle availability and an interaction term with self-reported distance were also added. The reason behind the inclusion of the vehicle variable in our estimations has to do with the transportation cost argument. That is, owning a car can reduce the inconvenience of using the public transportation system where inflexible routs, deficient multi-modal networks, or high crime rates in public transport could be relevant issues on election day, especially in developing countries.

$Z$ is a vector of control variables: gender, indigeneity, parents’ educational level, and household income decile. $β$ , $π$ , and $γ$ are vectors of the parameters to be estimated, and $Φ$ is the normal cumulative density function.

Given that past and future electoral engagement are likely to be related, we also estimate a recursive bivariate probit model to assess whether voter turnout in paper-ballot elections is independent of the voter’s willingness to cast a hypothetical online vote or if previous turnout reinforces/weakens future voting behavior. This second empirical specification is formalized as follows:

v_{1}^{*} = X β_{1} + W π_{1} + Z γ_{1} + ε_{1}

(13)

v_{2}^{*} = X β_{2} + W π_{2} + Z γ_{2} + φ v_{1}^{*} + ε_{2}

(14)

v_{j} = {\begin{matrix} 1 & i f v_{j}^{*} > 0 \\ 0 & i f v_{j}^{*} \leq 0 \end{matrix}; j = 1, 2

(15)

Where: $v_{j}^{*}$ is an unobserved latent variable; $v_{1}$ represents voter turnout in the presidential/congressional election held in 2017, whereas $v_{2}$ indicates future participation in elections held with hypothetical i-voting. The error terms $ε_{1}$ and $ε_{2}$ are assumed to be normally distributed and their covariance is equal to $ρ$ . This parameter is not directly estimated, so its inverse hyperbolic tangent (athrho) is reported. Thus, if athrho is statistically significant, then past voter-turnout and stated i-voting behavior are interrelated decisions.

On the other hand, vectors $X$ , $W$ , and $Z$ are defined as above; and $β$ , $π$ , $γ$ , and $φ$ are the vectors of parameters, which have to be estimated.

Furthermore, the bivariate probit model allows us to predict the conditional probability of success and failure, that is, the likelihood of casting a hypothetical online vote, conditional on past voting behavior:

\begin{matrix} \Pr (i - voting = 1 | paper - ballot = 0) \\ = \frac{\Pr (i - voting = 1, paper - ballot = 0)}{\Pr (paper - ballot = 0)} \end{matrix}

(16)

\begin{matrix} \Pr (i - voting = 1 | paper - ballot = 1) \\ = \frac{\Pr (i - voting = 1, paper - ballot = 1)}{\Pr (paper - ballot = 1)} \end{matrix}

(17)

Finally, Table 1 shows the descriptive statistics of dependent variables and the set of regressors that were used in the empirical study.

Table 1.

Descriptive Statistics.

	Obs.	Mean	SD
Dependent variables
2016 Mayoral election (Turnout = 1)	514	0.677	0.468
2017 Presidential election (Turnout = 1)	633	0.853	0.354
i-voting election (Turnout = 1)	677	0.939	0.239
Political attitudes
Valuation of democracy	677	3.495	0.741
Satisfaction with democracy	677	2.180	0.699
Political interest	677	2.836	0.884
Apathy toward political parties (Yes = 1)	677	0.731	0.444
Left-wing ideology (Yes = 1)	677	0.257	0.437
Civic engagement (Yes = 1)	677	0.399	0.490
Travel costs on election day
Transportation cost (thousand Chilean pesos)	677	2.394	11.372
Journey time in minutes	677	52.669	185.086
Declared distance	650	3.832	0.914
Available vehicle (Yes = 1)	677	0.199	0.400
Control variables
Gender (Female = 1)	677	0.554	0.497
Indigeneity (Indigenous people = 1)	677	0.093	0.291
Parental educational level	677	5.384	1.819
Household income decile	677	3.682	2.425

Results and Discussion

In this section, we discuss the main results. We present estimations of the probability of voting in two recent ballot-paper elections held in Chile.⁵ This can be taken as evidence of real past behavior regarding political participation. Then, we replicate the same models under the assumption that voters can opt for online voting, which is a hypothetical scenario. Finally, we test if stated i-voting behavior is independent from past electoral engagement.

Table 2 shows probit estimations for the turnout probability in the 2016 Chile mayoral election. Results are consistent with past literature, with variables from the political dimension being more relevant statistically and in terms of magnitude than socioeconomic characteristics for explaining electoral participation. Specifically, our results indicate that if university students report higher valuation of democracy and more political interest, and when they are involved in civic participation activities, they are more likely to have participated in the last mayoral election held in Chile. The effect of household income on voting behavior is unclear in this model.

Table 2.

Marginal Effects at the Means After Binary Probit Estimation for Youth Voter Turnout, 2016 Chile Mayoral Election.

Dependent variable: Voter turnout (Yes = 1)	(1)	(2)	(3)	(4)	(5)
Political attitudes
Valuation of democracy	0.083***	0.081***	0.079**	0.080***	0.079**
	(0.031)	(0.031)	(0.031)	(0.031)	(0.031)
Satisfaction with democracy	0.041	0.043	0.045	0.043	0.043
	(0.031)	(0.031)	(0.031)	(0.031)	(0.031)
Political interest	0.097***	0.101***	0.100***	0.099***	0.098***
	(0.028)	(0.029)	(0.029)	(0.028)	(0.028)
Apathy toward political parties	−0.032	−0.030	−0.038	−0.041	−0.041
	(0.049)	(0.049)	(0.049)	(0.048)	(0.048)
Left-wing ideology	−0.007	−0.008	−0.007	−0.005	−0.005
	(0.052)	(0.052)	(0.052)	(0.052)	(0.051)
Civic engagement	0.076*	0.074*	0.072*	0.082*	0.082*
	(0.043)	(0.043)	(0.043)	(0.043)	(0.043)
Travel costs on election day
Transportation cost		−0.001
		(0.002)
Journey time			−0.0004***
			(0.000)
Declared distance				−0.044*	−0.050**
				(0.024)	(0.025)
Available vehicle
Unconditional effect:		0.073	0.067	0.076	0.086*
		(0.050)	(0.050)	(0.050)	(0.052)
Conditional effect on the distance to a polling place:
In the same street/block					0.135
					(0.112)
Less than 5 blocks, same county					0.129
					(0.103)
5 to 10 blocks, same county					0.111
					(0.077)
More than 10 blocks, same county					0.081
					(0.050)
Another county, same region					0.039
					(0.096)
Another region					−0.011
					(0.183)
Control variables
Gender	0.027	0.022	0.022	0.020	0.020
	(0.043)	(0.043)	(0.043)	(0.043)	(0.043)
Indigeneity	−0.098	−0.085	−0.076	−0.085	−0.084
	(0.074)	(0.075)	(0.075)	(0.075)	(0.075)
Parental educational level	0.004	0.002	0.001	0.003	0.003
	(0.013)	(0.013)	(0.013)	(0.013)	(0.013)
Household income	0.016*	0.015	0.015	0.015	0.015
	(0.010)	(0.010)	(0.010)	(0.010)	(0.010)
Observations	514	514	514	508	508
Log-likelihood	−300.5	−299.2	−296.4	−291.2	−291.1
Pseudo R²	0.071	0.075	0.083	0.080	0.080
χ² statistic	47.42	50.31	57.78	52.47	53.11

Note. Robust standard errors in parentheses.

*p < .1. **p < .05. ***p < .01.

Regarding travel costs on Election Day, results show a negative and significant relationship between journey time and declared distance on past electoral behavior, suggesting that travel cost matters in voter decisions. These results are consistent with our predictions outlined in the theoretical model (see equations (8)–(11)). Moreover, the positive effect of vehicle availability on voter turnout was only significant at the 10% level when it was allowed to interact with the distance to the polling station. And we did not find statistical significance for the transportation cost variable, although the former association is also negative.

Table 3 replicates the previous model for the 2017 Chile presidential election. Results are fundamentally the same as those found for the 2016 mayoral election model. While political dimension variables such as valuation of democracy and political interest are statically significant and positively associated with electoral behavior, most socioeconomic characteristics are not relevant for turnout decisions in presidential elections. Nevertheless, results indicate that university students with more educated parents are more likely to vote in Chile presidential elections.

Table 3.

Marginal Effects at the Means After Binary Probit Estimation for Youth Voter Turnout, 2017 Chile Presidential Election.

Dependent variable: Voter turnout (Yes = 1)	(1)	(2)	(3)	(4)	(5)
Political attitudes
Valuation of democracy	0.041**	0.041**	0.037**	0.029*	0.032*
	(0.018)	(0.018)	(0.017)	(0.017)	(0.018)
Satisfaction with democracy	0.021	0.022	0.027	0.018	0.019
	(0.020)	(0.021)	(0.020)	(0.019)	(0.019)
Political interest	0.063***	0.069***	0.061***	0.062***	0.065***
	(0.016)	(0.017)	(0.016)	(0.015)	(0.015)
Apathy toward political parties	−0.014	−0.017	−0.029	−0.020	−0.022
	(0.031)	(0.032)	(0.030)	(0.028)	(0.029)
Left-wing ideology	−0.012	−0.014	−0.015	−0.007	−0.011
	(0.035)	(0.037)	(0.035)	(0.033)	(0.034)
Civic engagement	−0.004	−0.011	−0.004	−0.002	−0.003
	(0.028)	(0.030)	(0.028)	(0.027)	(0.028)
Travel costs on election day
Transportation cost		−0.012***
		(0.003)
Journey time			−0.0002*
			(0.000)
Declared distance				−0.053***	−0.042**
				(0.014)	(0.017)
Available vehicle
Unconditional effect:		0.095***	0.082***	0.078***	0.062**
		(0.027)	(0.026)	(0.024)	(0.029)
Conditional effect on the distance to a polling place:
In the same street/block					−0.123
					(0.218)
Less than 5 blocks, same county					−0.069
					(0.126)
5 to 10 blocks, same county					−0.005
					(0.060)
More than 10 blocks, same county					0.076***
					(0.027)
Another county, same region					0.178***
					(0.044)
Another region					0.300***
					(0.079)
Control variables
Gender	−0.001	−0.007	−0.004	−0.001	−0.002
	(0.027)	(0.028)	(0.027)	(0.026)	(0.026)
Indigeneity	−0.052	−0.024	−0.032	−0.030	−0.028
	(0.049)	(0.047)	(0.045)	(0.044)	(0.046)
Parental educational level	0.030***	0.024***	0.026***	0.026***	0.028***
	(0.009)	(0.008)	(0.008)	(0.008)	(0.008)
Household income	−0.001	−0.001	−0.004	−0.003	−0.005
	(0.007)	(0.007)	(0.006)	(0.006)	(0.006)
Observations	633	633	633	625	625
Log-likelihood	−237.6	−218.3	−225.6	−213.6	−212.3
Pseudo R²	0.101	0.174	0.146	0.141	0.146
χ² statistic	43.45	55.16	46.57	48.43	50.54

Note. Robust standard errors in parentheses.

*p < .1. **p < .05. ***p < .01.

Similarly, travel cost variables such as journey time and distance to the voting place are also negatively and significantly associated with the probability of participating in the past presidential election. In this model, we found that transportation cost and the availability of a vehicle also matter. Having a vehicle may facilitate travel to the polling station if that place is located far from home. If this is true, we should observe that the availability of a vehicle matters more to the extent that people must travel longer distances for voting. Thus, we interacted vehicle availability with distance to the polling station. Results suggest that vehicle availability is important to explain past voting behavior when the polling station is more than 10 blocks away from home. The magnitude of that effect is greater when voters have to travel to another county or region to participate in the election. This finding has been reported in the literature before (Bhatti, 2012; Haspel & Knotts, 2005).

Tables 4 and Supplemental Tables SM1 to SM3 (see Supplemental Material) present the results of a hypothetical setting with i-voting. While Table 4 shows the estimates when asking about a general willingness to vote online, Supplemental Tables SM1 to SM3 are separated between municipal, presidential, parliamentary, and regional elections held using this hypothetical scenario. Each one reports how that variant of the model relates to transportation cost.

Table 4.

Marginal Effects at the Means After Binary Probit Estimation for Youth Voter Turnout, Election Held Using Hypothetical i-Voting.

Dependent variable: i-voting turnout (Yes = 1)	(1)	(2)	(3)	(4)	(5)
Political attitudes
Valuation of democracy	0.014	0.014	0.014	0.010	0.010
	(0.010)	(0.010)	(0.010)	(0.011)	(0.011)
Satisfaction with democracy	0.029**	0.029**	0.029**	0.029**	0.028**
	(0.012)	(0.012)	(0.012)	(0.012)	(0.012)
Political interest	0.022**	0.022**	0.022**	0.022**	0.021**
	(0.009)	(0.010)	(0.009)	(0.010)	(0.010)
Apathy toward political parties	−0.000	−0.000	−0.001	−0.001	−0.001
	(0.018)	(0.018)	(0.018)	(0.018)	(0.018)
Left-wing ideology	0.022	0.022	0.022	0.020	0.020
	(0.017)	(0.017)	(0.017)	(0.018)	(0.017)
Civic engagement	−0.010	−0.010	−0.010	−0.009	−0.009
	(0.015)	(0.015)	(0.015)	(0.016)	(0.015)
Travel costs on election day
Transportation cost		0.000
		(0.000)
Journey time			−0.000
			(0.000)
Declared distance				0.005	0.004
				(0.008)	(0.009)
Available vehicle
Unconditional effect:		0.003	0.003	0.005	0.007
		(0.018)	(0.018)	(0.019)	(0.019)
Conditional effect on the distance to a polling place:
In the same street/block					0.033
					(0.072)
Less than 5 blocks, same county					0.023
					(0.051)
5 to 10 blocks, same county					0.014
					(0.031)
More than 10 blocks, same county					0.005
					(0.018)
Another county, same region					−0.003
					(0.031)
Another region					−0.010
					(0.055)
Control variables
Gender	0.021	0.021	0.020	0.023	0.023
	(0.015)	(0.015)	(0.015)	(0.015)	(0.015)
Indigeneity	−0.066*	−0.066*	−0.064*	−0.069*	−0.070*
	(0.037)	(0.037)	(0.036)	(0.039)	(0.039)
Parental educational level	0.010**	0.010**	0.010**	0.010**	0.010**
	(0.005)	(0.005)	(0.005)	(0.005)	(0.005)
Household income	−0.002	−0.003	−0.003	−0.003	−0.003
	(0.003)	(0.003)	(0.003)	(0.003)	(0.003)
Observations	677	677	677	650	650
Log-likelihood	−135.0	−135.0	−135.0	−128.9	−128.8
Pseudo R²	0.127	0.127	0.127	0.110	0.110
χ² statistic	36.77	38.35	38.14	32.36	32.93

Note. Robust standard errors in parentheses.

*p < .1. **p < .05. ***p < .01.

Main findings regarding political dimension variables remain the same as previous models. That means that political interest and democratic values are themselves strong drivers of political participation and are not affected by changes in the voting system (Table 4). Socioeconomic characteristics continue being less relevant in explaining political participation in general, though these variables present statistical significance in some cases. While women and people with more educated parents are more likely to participate in a presidential election (columns 4–6; Supplemental Table SM1), a higher household income reduces the willingness to vote during the forthcoming regional governor elections (columns 4 and 6; Supplemental Table SM3). The latter suggests the existence of a class-biased electorate among Chilean youth, which has been previously discussed in the literature (see Acuña-Duarte, 2017; Corvalán & Cox, 2013).

Interestingly, travel cost variables are no longer statically significant in hypothetical online voting (columns 2–5, Table 4). That means that there is absence of evidence indicating that time, distance, and transport costs matter when people are not required to move to polling places to cast their ballots in elections. This finding holds across the different types of elections (see Supplemental Tables SM1–SM3). While we cannot argue in favor of a null effect of transport cost, that some people demonstrate different electoral behaviors depending on the institutional setting suggests that the importance of transport costs in explaining voting behavior can be conditioned on the voting system.

In addition, the availability of a vehicle continues to be an important factor in explaining voter turnout in hypothetical i-voting, particularly in mayoral, congressional, and presidential election (see Supplemental Tables SM1–SM3). A positive relationship between having a vehicle and voting may be reflective of reasons other than voting costs like, for example, household wealth (Fowler, 2013; Nadeau et al., 2019). One possible explanation is that the high-income people have a potentially higher tax exposure compared to low-income people, which may incentivize wealthier individuals to get more involved in politics and participle in the act of voting (Kasara & Suryanarayan, 2015).

Finally, Table 5 shows the results for the recursive bivariate probit model. The new evidence confirms what was previously reported for political and travel-cost variables. That is, voter turnout among Chilean youth is reinforced by a higher political interest and a positive attitude toward democracy, regardless of the voting system. Travel-time, distance, and transportation costs only matter in conventional elections where paper-based ballots were used (columns 1, 3, and 5, Table 5).

Table 5.

Estimates of the Recursive Bivariate Probit Model for Voter Turnout in Paper-Ballot and i-Voting Elections.

	Paper ballot	i-voting	Paper ballot	i-voting	Paper ballot	i-voting
	(1)	(2)	(3)	(4)	(5)	(6)
Past voting behavior
2017 paper-ballot election (Yes = 1)		−1.254***		−1.116***		−1.238***
		(0.201)		(0.222)		(0.426)
Political attitudes
Valuation of democracy	0.204**	0.206**	0.189**	0.190**	0.172**	0.161*
	(0.085)	(0.088)	(0.084)	(0.096)	(0.088)	(0.098)
Satisfaction with democracy	0.105	0.255**	0.136	0.304**	0.087	0.265**
	(0.103)	(0.118)	(0.100)	(0.119)	(0.107)	(0.133)
Political interest	0.360***	0.350***	0.314***	0.332***	0.346***	0.338***
	(0.088)	(0.088)	(0.083)	(0.095)	(0.090)	(0.092)
Apathy toward political parties	−0.095	−0.062	−0.166	−0.093	−0.140	−0.070
	(0.163)	(0.163)	(0.164)	(0.177)	(0.177)	(0.167)
Left-wing ideology	−0.076	0.106	−0.087	0.152	−0.047	0.134
	(0.174)	(0.184)	(0.167)	(0.195)	(0.174)	(0.215)
Civic engagement	−0.094	−0.108	−0.031	−0.112	−0.039	−0.075
	(0.160)	(0.145)	(0.145)	(0.155)	(0.167)	(0.161)
Travel costs on election day
Transportation cost	−0.066***	−0.006
	(0.015)	(0.005)
Journey time			−0.001**	−0.000
			(0.001)	(0.000)
Declared distance					−0.307***	−0.100
					(0.088)	(0.131)
Available vehicle
Unconditional effect:	0.666***	0.328	0.556***	0.268	0.591***	0.317
	(0.236)	(0.209)	(0.202)	(0.204)	(0.216)	(0.263)
Control variables
Gender	−0.050	0.119	−0.030	0.161	−0.003	0.150
	(0.147)	(0.148)	(0.141)	(0.149)	(0.155)	(0.187)
Indigeneity	−0.088	−0.376*	−0.131	−0.449**	−0.124	−0.423*
	(0.212)	(0.198)	(0.200)	(0.191)	(0.215)	(0.233)
Parental educational level	0.125***	0.138***	0.136***	0.140***	0.156***	0.149***
	(0.043)	(0.044)	(0.043)	(0.047)	(0.047)	(0.046)
Household income	−0.000	−0.022	−0.016	−0.031	−0.020	−0.030
	(0.034)	(0.030)	(0.032)	(0.032)	(0.033)	(0.031)
Constant	−1.267***	−0.544	−1.192***	−0.515	−0.066	−0.037
	(0.452)	(0.433)	(0.444)	(0.466)	(0.543)	(0.718)
athrho	1.673**		1.134***		1.486
	(0.781)		(0.326)		(1.368)
Conditional probabilities
Pr(i-voting = 1 \| paper-ballot = 0)
Sample average	0.309		0.523		0.356
Sample standard deviation	(0.272)		(0.208)		(0.251)
Pr(i-voting = 1 \| paper-ballot = 1)
Sample average	0.953		0.955		0.955
Sample standard deviation	(0.065)		(0.052)		(0.051)
Observations	633		633		625
Log-likelihood	−346.7		−355.1		−339.5
χ² statistic	152.55		115.41		132.18

Note. Robust standard errors in parentheses.

p < .1. **p < .05. ***p < .01.

These findings suggest that people care about the transport costs that they must incur when going to vote. Therefore, an i-voting system, by saving time and cost, may promote higher electoral participation, a voting behavior that has been observed in Estonia and Switzerland (Germann & Serdült, 2017; Vassil et al., 2016). Moreover, the evidence indicates that these voting decisions are not independent of each other (i.e., athrho is statistically significant at the 5% level) when transportation cost and journey time were used to measure travel costs on election day. Specifically, we found a negative association between voter turnout in paper-ballot elections and the likelihood of casting an online vote (columns 2, 4, and 6, Table 5). In other words, university students who did not vote in past elections are more likely to change their voting behavior and participate in an i-voting system. In the most favorable scenario, the estimates suggested that online voting would attract 52.3% of unengaged university students to the virtual polling stations (columns 3 and 4, Table 5).

Robustness

One concern that arises from our data is that turnout rates among university students (ages 18–29) computed from our online survey seem to be greater than those reported by official sources for Chilean youth (see Section “Introduction” and Table 1). The latter may have consequences over the interpretation and implications of our results. However, national surveys conducted in Chile reveal that youth vote is larger among higher-education students (HES). To illustrate, the participation rate among Chilean HES (ages 18–29) was 45.8% and 55.3% in the presidential elections held in 2013 and 2017, respectively (LAPOP, 2017, 2019); that is, nine percentage points over the national electoral participation rate reported in 2017 (Servicio Electoral, 2020).

In order to examine if our results are in line with those reported in national surveys, we used individual-level data from the 2014–2018/2019 waves of the AmericasBarometer survey conducted in Chile to estimate a pooled probit model for the likelihood of casting a vote in a presidential election. Given that the questions from our online questionnaire resemble those included in this national representative survey, most of our covariates were replicated. Although AmericasBarometer survey does not include questions related to travel costs on Election Day, question R5 allowed to identify the number of household vehicle holdings among respondents. In turn, question OCUP4A permitted to control by the HES status—for details, see LAPOP (2014, 2017, 2019). Results are shown in Table 6.

Table 6.

Marginal Effects at the Means After Pooled Probit Estimation for Voter Turnout in Presidential Elections Held in Chile.

Dependent variable: Voter turnout (Yes = 1)	Youth vote (ages 18–29)				Higher-education students (ages 18–29)
	(1)	(2)	(3)	(4)	(5)	(6)
Political attitudes
Valuation of democracy	0.201***	0.195***	0.198***	0.193***	0.248*	0.247*
	(0.068)	(0.068)	(0.068)	(0.068)	(0.147)	(0.147)
Political interest	0.363***	0.349***	0.360***	0.346***	0.201*	0.204*
	(0.057)	(0.058)	(0.057)	(0.058)	(0.119)	(0.119)
Apathy toward political parties	−0.232***	−0.238***	−0.229***	−0.236***	−0.206*	−0.209*
	(0.062)	(0.062)	(0.062)	(0.062)	(0.112)	(0.112)
Left-wing ideology	0.070*	0.065*	0.071*	0.066*	0.099	0.099
	(0.039)	(0.039)	(0.039)	(0.039)	(0.077)	(0.077)
Civic engagement	0.077**	0.087**	0.075**	0.085**	0.160**	0.157**
	(0.036)	(0.037)	(0.036)	(0.036)	(0.074)	(0.075)
Travel costs on election day
Vehicle availability (Yes = 1)	0.059*	0.054			0.001
	(0.035)	(0.036)			(0.075)
Number of household vehicle holding			0.046*	0.042*		−0.031
			(0.025)	(0.025)		(0.049)
Control variables
Gender	0.020	0.019	0.022	0.021	−0.042	−0.040
	(0.036)	(0.037)	(0.037)	(0.037)	(0.076)	(0.076)
Indigeneity	−0.047	−0.048	−0.044	−0.046	−0.063	−0.054
	(0.052)	(0.052)	(0.052)	(0.052)	(0.117)	(0.118)
Higher-education student (Yes = 1)		0.077*		0.077*
		(0.043)		(0.044)
Observations	868	868	868	868	200	200
Log-likelihood	−534.2	−532.5	−533.9	−532.2	−126.8	−126.6
Pseudo R²	0.099	0.102	0.100	0.103	0.084	0.085
χ² statistic	102.14	105.13	101.58	104.82	23.68	23.91

Note. Year dummies are unreported. Robust standard errors in parentheses.

*p < .1. **p < .05. ***p < .01.

The complementary evidence is consistent with previous findings related to voter turnout in past paper-ballot elections (see Tables 2 and 3). That is to say, conventional turnout probability among university students is positively affected by a higher political interest, valuation of democracy, and civic engagement (columns 5 and 6, Table 6). These findings are extendable to young population (ages 18–29), where the likelihood of casting a vote in a presidential election seems to be 7.7% higher among higher-education students (columns 2 and 4, Table 6).

Regarding the effect of travel costs on Election Day, pooled probit estimates suggest that voter turnout is larger among Chilean youth as the number of household vehicle holding increases (columns 1, 3, and 4, Table 6). As previously highlighted, vehicle availability seems to reduce time and the inconvenience of voting, positively affecting voter turnout among young population.

Conclusion

After implementing voluntary voting in 2012, Chile has experienced a significant reduction in turnout rates, especially among youth. The discussion on new public policy instruments to encourage political engagement is still amid debate. We argue that an i-voting system may promote electoral participation by reducing travel cost and time to polling places, especially among youth who are more accepting of technological changes and who, due to higher-education, are more likely to live away from home or where they are registered to vote. In that regard, this paper assessed the effect of transportation costs, journey time, and distance to a polling place on the electoral engagement of Chilean university students. We first outlined a theoretical model to understand the mechanisms through which time and travel costs may affect voting rates. We tested the predictions of our model by comparing past electoral behavior from the mayoral and presidential elections held in 2016 and 2017 in Chile, respectively, with responses to a hypothetical i-voting system. In doing so, we shed some light on the potential contribution of voting system reforms that promote i-voting.

The following insights can be highlighted from our findings: In line with previous literature, the likelihood of voting among youth is positively affected by a higher political interest, valuation of democracy, and parental educational level. Regarding the effect of travel costs on Election Day, vehicle availability seems to be relevant to explain past voting behavior, reducing the inconvenience of voting and increasing voter turnout. These results are robust as estimating a pooled probit model with individual-level data from a national representative survey conducted in Chile.

Moreover, conventional voter turnout is negatively affected by transportation cost, journey-time, and distance to the polling station. Although political interest and attitudes toward democracy still increase voter turnout probability with hypothetical internet voting, we did not find evidence of travel-to-polling-station costs being statistically relevant with i-voting. Furthermore, we explored whether stated i-voting behavior was influenced by past voter turnout in paper-ballot elections. The evidence indicates that both decisions are significantly linked and that university students not voting previously through conventional voting channels are more likely to change their electoral behavior and participate in an election with an i-voting system. Our results are robust among different hypothetical national, regional, and local elections held using i-voting.

Our findings shed some lights on important policy implications. We distinguish those more generic policy implications that can be extrapolated to other countries and other population segments different from youth that exhibit low turnout rates. First, our results suggest that a reduction in transportation costs would increase voter turnout. Then, free public transportation across the country on this particular day or increasing the number of polling places in order to reduce costs, time, and/or inconvenience of voting can be common policy recommendations to any country and population segments. It is important, however, discuss some issues that can emerge in the implementation of such policy. For example, in contexts where public transport is not fully funded by the government, the free busing policy will require a private-public agreement which can turn out to be complex to achieve. In addition, free busing fully funded by the government may be questioned for giving room to be used as political intervention by incumbents. On the other hand, travel cost variables may also be related to leisure opportunity costs, as Chilean elections are commonly held on weekends. Voters may be more prone to vote online because this alternative reduces restrictions for planning activities and enjoying leisure time with family and friends. This factor could be even more relevant among university students. Second, the discussion of i-voting systems as a secure and efficient voting mechanism becomes special relevant with the current COVID-19 pandemic, as decreasing exposure and contagion risks of the virus take special precedence. Internet voting ensures sanitary and hygienic conditions for voters. Since youth exhibit higher levels of technological acceptance and early adoption rates, young people would be more aware of the computer risks related to i-voting; therefore, policy makers should continue strengthening information security of e-government options in order to foster citizen trust on i-voting. Generally speaking, the possibility of offering i-voting as alternative to more conventional voting modes can avoid postponing key democratic elections during emergency situations. Moreover, this voting system will make electoral participation easier not only among youth but people with reduced mobility or with the responsibility of caring others. Note that i-voting also makes it suitable for high-crime environments (e.g., gang-controlled neighborhoods) where the exercise of civil rights has been undermined and the right to vote as a free citizen could be a utopia for its inhabitants, a phenomenon that has been previously highlighted for Latin America (for instance, see Berens & Dallendörfer, 2019; Córdova, 2019).

On the other hand, a couple of caveats deserve attention. Our data set was retrieved from an online questionnaire, which could be permeable to self-selection or under-coverage biases. Nevertheless, several indicators from our survey data (such as average age, indigeneity, satisfaction with democracy, or political leanings) are similar to those computed from national surveys conducted in Chile by international research institutes. In turn, the impact of attitudes toward ICTs and citizen trust in electronic participation and online government services were discarded from the analysis. However, the questions used to proxy the willingness to vote under an i-voting system was framed conditional on corresponding security actions have been taken. Certainly, addressing the above issues imposes a challenge for future research.

Finally, it is important to mention that an eventual implementation of i-voting should not be viewed as a panacea for eradicating abstention among youth, and the positive effects may not be immediate. As highlighted by Vassil et al. (2016), while i-voting is a more convenient voting system and an efficient channel of participation, but it will not address the underlying causes of youth apathy toward politics, political parties, and democracy. To deal with political disaffection, we must promote structural reforms that generate favorable attitudes toward democracy among youth.

Supplemental Material

sj-docx-1-sgo-10.1177_21582440241252057 – Supplemental material for Travel Cost on Election Day and Voter-Turnout in Chile: Exploring University Students’ Willingness to Cast a Hypothetical Online Vote

Supplemental material, sj-docx-1-sgo-10.1177_21582440241252057 for Travel Cost on Election Day and Voter-Turnout in Chile: Exploring University Students’ Willingness to Cast a Hypothetical Online Vote by Andrés A. Acuña-Duarte and César A. Salazar in SAGE Open

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by Universidad del Bío-Bío, Chile, under the research project “Análisis institucional y comportamiento político-electoral” [Grant Number 2130483 GI/EF].

ORCID iD

Andrés A. Acuña-Duarte

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon request due to privacy restrictions.

Supplemental Material

Supplemental material for this article is available online.

Notes

References

Acuña-Duarte

A. A.

(2017). Electoral apathy among Chilean youth: New evidence for the voter registration dilemma. Estudios Gerenciales, 33(145), 341–351. https://doi.org/10.1016/j.estger.2017.11.005

Avery

J. M.

Peffley

(2005). Voter registration requirements, voter turnout, and welfare eligibility policy: Class bias matters. State Politics & Policy Quarterly, 5(1), 47–67. https://doi.org/10.1177/153244000500500103

Benabdallah

Audras

Coudert

El Madhoun

Badra

(2022). Analysis of blockchain solutions for e-voting: A systematic literature review. IEEE Access, 10, 70746–70759. https://doi.org/10.1109/ACCESS.2022.3187688

Berens

Dallendörfer

(2019). Apathy or anger? How crime experience affects individual vote intention in Latin America and the Caribbean. Political Studies, 67(4), 1010–1033. https://doi.org/10.1177/0032321718819106

Berinsky

A. J.

(2005). The perverse consequences of electoral reform in the United States. American Politics Research, 33(4), 471–491. https://doi.org/10.1177/1532673X04269419

Bhatti

(2012). Distance and voting: Evidence from Danish municipalities. Scandinavian Political Studies, 35(2), 141–158. https://doi.org/10.1111/j.1467-9477.2011.00283.x

Brady

H. E.

McNulty

J. E.

(2011). Turning out to vote: The costs of finding and getting to the polling place. American Political Science Review, 105(1), 115–134. https://doi.org/10.1017/S0003055410000596

Brians

C. L.

Grofman

(1999). When registration barriers fall, who votes? An empirical test of a rational choice model. Public Choice, 99(1–2), 161–176. https://doi.org/10.1023/A:1018346602250

Carlin

R. E.

(2006). The decline of citizen participation in electoral politics in post-authoritarian Chile. Democratization, 13(4), 632–651. https://doi.org/10.1080/13510340600791921

10.

Contreras

Navia

(2013). Diferencias generacionales en la participación electoral en Chile, 1988–2010. Revista de Ciencia Política, 33(2), 419–441. https://doi.org/10.4067/S0718-090X2013000200001

11.

Córdova

(2019). Living in gang-controlled neighborhoods: Impacts on electoral and nonelectoral participation in El Salvador. Latin American Research Review, 54(1), 201–221. https://doi.org/10.25222/larr.387

12.

Corvalán

Cox

(2013). Class-biased electoral participation: The youth vote in Chile. Latin American Politics and Society, 55(3), 47–68. https://doi.org/10.1111/j.1548-2456.2013.00202.x

13.

Dyck

J. J.

Gimpel

J. G.

(2005). Distance, turnout, and the convenience of voting. Social Science Quarterly, 86(3), 531–548. https://doi.org/10.1111/j.0038-4941.2005.00316.x

14.

Ehin

Solvak

Willemson

Vinkel

(2022). Internet voting in Estonia 2005–2019: Evidence from eleven elections. Government Information Quarterly, 39(4), 101718. https://doi.org/10.1016/j.giq.2022.101718

15.

Fowler

(2013). Electoral and policy consequences of voter turnout: Evidence from compulsory voting in Australia. Quarterly Journal of Political Science, 8(2), 159–182. http://dx.doi.org/10.1561/100.00012055

16.

Germann

Serdült

(2017). Internet voting and turnout: Evidence from Switzerland. Electoral Studies, 47, 1–12. https://doi.org/10.1016/j.electstud.2017.03.001

17.

Haspel

Knotts

H. G.

(2005). Location, location, location: Precinct placement and the costs of voting. The Journal of Politics, 67(2), 560–573. https://doi.org/10.1111/j.1468-2508.2005.00329.x

18.

Highton

(1997). Easy registration and voter turnout. The Journal of Politics, 59(2), 565–575. https://doi.org/10.1017/S0022381600053585

19.

Highton

(2004). Voter registration and turnout in the United States. Perspectives on Politics, 2(3), 507–515. https://doi.org/10.1017/S1537592704040307

20.

Highton

Wolfinger

R. E.

(1998). Estimating the effects of the National Voter Registration Act of 1993. Political Behavior, 20(2), 79–104. https://doi.org/10.1023/A:1024851912336

21.

James

T. S.

(2021). New development: Running elections during a pandemic. Public Money & Management, 41(1), 65–68. https://doi.org/10.1080/09540962.2020.1783084

22.

Jefferson

Rubin

A. D.

Simons

Wagner

(2004). Analyzing internet voting security. Communications of the ACM, 47(10), 59–64. https://doi.org/10.1145/1022594.1022624

23.

Kasara

Suryanarayan

(2015). When do the rich vote less than the poor and why? Explaining turnout inequality across the world. American Journal of Political Science, 59(3), 613–627. https://doi.org/10.1111/ajps.12134

24.

Knack

(1995). Does “motor voter” work? Evidence from state-level data. The Journal of Politics, 57(3), 796–811. https://doi.org/10.2307/2960193

25.

Krimmer

Duenas-Cid

Krivonosova

(2021a). Debate: Safeguarding democracy during pandemics. Social distancing, postal, or internet voting—The good, the bad or the ugly? Public Money & Management, 41(1), 8–10. https://doi.org/10.1080/09540962.2020.1766222

26.

Krimmer

Duenas-Cid

Krivonosova

(2021b). New methodology for calculating cost-efficiency of different ways of voting: Is internet voting cheaper? Public Money & Management, 41(1), 17–26. https://doi.org/10.1080/09540962.2020.1732027

27.

Landwehr

Harms

(2020). Preferences for referenda: Intrinsic or instrumental? Evidence from a survey experiment. Political Studies, 68(4), 875–894. https://doi.org/10.1177/0032321719879619

28.

LAPOP. (2014). Chile 2014 AmericasBarometer Survey: Data files and notes. http://datasets.americasbarometer.org/database

29.

LAPOP. (2017). Chile 2016/17 AmericasBarometer Survey: Data files and notes. http://datasets.americasbarometer.org/database

30.

LAPOP. (2019). Chile 2018/19 AmericasBarometer Survey: Data files and notes. http://datasets.americasbarometer.org/database

31.

Lauer

T. W.

(2004). The risk of e-voting. Electronic Journal of E-government, 2(3), 177–186.

32.

Mellon

Peixoto

Sjoberg

F. M.

(2017). Does online voting change the outcome? Evidence from a multi-mode public policy referendum. Electoral Studies, 47, 13–24. https://doi.org/10.1016/j.electstud.2017.02.006

33.

Musiał-Karg

Kapsa

(2021). Debate: Voting challenges in a pandemic—Poland. Public Money & Management, 41(1), 6–8. https://doi.org/10.1080/09540962.2020.1809791

34.

Nadeau

Lewis-Beck

M. S.

Foucault

(2019). Wealth and voter turnout: Investigating twenty-eight democracies. Polity, 51(2), 261–287. https://doi.org/10.1086/701961

35.

Nagler

(1991). The effect of registration laws and education on U.S. voter turnout. American Political Science Review, 85(4), 1393–1405. https://doi.org/10.2307/1963952

36.

Nemeslaki

Aranyossy

Sasvári

(2016). Could on-line voting boost desire to vote? Technology acceptance perceptions of young Hungarian citizens. Government Information Quarterly, 33(4), 705–714. https://doi.org/10.1016/j.giq.2016.11.003

37.

Observatorio Social. (2017). Manual de trabajo de campo Encuesta CASEN 2017. http://observatorio.ministeriodesarrollosocial.gob.cl/encuesta-casen-2017

38.

Oliver

J. E.

(1996). The effects of eligibility restrictions and party activity on absentee voting and overall turnout. American Journal of Political Science, 40(2), 498–513. https://doi.org/10.2307/2111634

39.

Pateman

(2012). Participatory democracy revisited. Perspectives on Politics, 10(1), 7–19. https://doi.org/10.1017/S1537592711004877

40.

Powell

Williams

C. K.

Bock

D. B.

Doellman

Allen

(2012). e-Voting intent: A comparison of young and elderly voters. Government Information Quarterly, 29(3), 361–372. https://doi.org/10.1016/j.giq.2012.01.003

41.

Rosenstone

S. J.

Wolfinger

R. E.

(1978). The effect of registration laws on voter turnout. American Political Science Review, 72(1), 22–45. https://doi.org/10.2307/1953597

42.

Servicio Electoral. (2020). Estadísticas: Padrón electoral y registro electoral. Retrieved September 15, 2020, from https://www.servel.cl/estadisticas-2/

43.

Servicio Electoral. (2023). Centro de datos: Resultados electorales históricos—Plebiscito 2022. Retrieved March 17, 2023, from https://www.servel.cl/centro-de-datos/resultados-electorales-historicos-gw3/

44.

Seyd

Curtice

Rose

(2018). How might reform of the political system appeal to discontented citizens? The British Journal of Politics and International Relations, 20(2), 263–284. https://doi.org/10.1177/1369148117736189

45.

Springall

Finkenauer

Durumeric

Kitcat

Hursti

MacAlpine

Halderman

J. A.

(2014). Security analysis of the Estonian internet voting system. In Ahn

(Ed.), Proceedings of the 2014 ACM SIGSAC conference on computer and communications security (pp. 703–715). Association for Computing Machinery.

46.

Stein

R. M.

(1998). Introduction: Early voting. Public Opinion Quarterly, 62(1), 57–69. https://www.jstor.org/stable/2749718

47.

Stein

R. M.

Vonnahme

(2008). Engaging the unengaged voter: Vote centers and voter turnout. The Journal of Politics, 70(2), 487–497. https://doi.org/10.1017/S0022381608080456

48.

Stone

M. N.

(1985). Voter registration: Context and results. The Urban Lawyer, 17(3), 519–528. https://www.jstor.org/stable/27893313

49.

Toro

(2007). La inscripción electoral de los jóvenes en Chile. Factores de incidencia y aproximaciones al debate. In Públicos

Centro de Estudios

(Ed.), Modernización del Régimen Electoral Chileno (pp. 101–122). CEP Chile.

50.

Vassil

Solvak

Vinkel

Trechsel

A. H.

Alvarez

R. M.

(2016). The diffusion of internet voting. Usage patterns of internet voting in Estonia between 2005 and 2015. Government Information Quarterly, 33(3), 453–459. https://doi.org/10.1016/j.giq.2016.06.007

51.

Vinkel

Krimmer

(2017). The how and why to internet voting an attempt to explain E-Stonia. In Krimmer

Volkamer

Barrat

Benaloh

Goodman

Ryan

P. Y. A.

Teague

(Eds.), International joint conference on electronic voting (pp. 178–191). Springer. https://doi.org/10.1007/978-3-319-52240-1_11

52.

Wolchok

Wustrow

Isabel

Halderman

J. A.

(2012). Attacking the Washington, DC internet voting system. In Keromytis

A. D.

(Ed.), Financial cryptography and data security: FC 2012 (pp.114–128). Springer. https://doi.org/10.1007/978-3-642-32946-3_10

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.05 MB