Theoretical Background and Hypotheses

Over one hundred nations or territories now allow expatriates to participate in elections via absentee ballot (Lafleur, Reference Lafleur2013; Umpierrez de Reguero et al., Reference Umpierrez de Reguero, Finn and Peltoniemi2023; McCann et al., Reference McCann, Escobar and Arana2019; Wellman et al., Reference Wellman, Allen and Nyblade2023). This expansion of voting rights ensures that a great many migrants will be drawn in some fashion into native-country politics when elections take place there, casting absentee ballots or donating to political parties, talking about politics with friends and relatives from the native country, or simply paying close attention to the contest from the distance (Burgess and Tyburski, Reference Burgess and Tyburski2020; Paarlberg, Reference Paarlberg2020).

Focusing on this latter form of engagement, what are the implications of attention to native-country politics for immigrant incorporation in the country of residence? One plausible scenario is that when immigrants turn their attention towards elections in their native country, they become less open to taking part in the country where they have settled. Following from Tsuda (Reference Tsuda2012), we label this a “zero-sum” model of bi-nationalism (see also Black, Reference Black2011: 1171; Black, Reference Black1987; Peltoniemi Reference Peltoniemi2018). The key theoretical premise of this model is that in an immigrant's country of origin, campaigns and elections would likely tend to focus on topics that are salient for voters within that country, such as the job performance of incumbent leaders or the attractiveness of specific national policies and ideological positions. These considerations may have little bearing on political debates, ideologies and agendas in the settlement country (Zechmeister, Reference Zechmeister2006). It is a truism that involvement, including simply paying attention to politics, can be costly for individuals. Many people may turn away from politics because they lack the wherewithal to keep track of public affairs and take part (Verba et al., Reference Verba, Schlozman and Brady1995). For immigrants, the prospect of following politics in two distinctive national contexts could be daunting; an immigrant whose attention is directed towards campaigns and elections in the country of birth may naturally pull back from politics in the residential country due to these limitations.

It is possible, however, to imagine a contrasting dynamic, where remote attentiveness to country-of-origin elections promotes deeper engagement in the settlement country (Levitt, Reference Levitt, Foner, Rumbaut and Gold2000: 460; Tsuda, Reference Tsuda2012: 635).Footnote ¹ Various mechanisms could account for this. Attention to elections as a transnational citizen may reinforce a general sense of social obligation and civic duty or bolster beliefs about personal efficacy in politics (see, for example, Mansbridge, Reference Mansbridge, Elkin and Soltan1999; Superti, Reference Superti2023). These effects could in turn lead to higher levels of enthusiasm to take part in politics where one has settled. At the very least, this hypothesis posits that there is no inherent tension between democratic incorporation as both an immigrant and emigrant (cf. Finn, Reference Finn2020).

We explore these contrasting possibilities by focusing on a sizable immigrant population in a Western democracy that has received scant scholarly attention: Americans living in Canada. The United States has long considered itself a nation of immigrants, but it is also a nation of emigration. In a 2018 report, the Federal Voting Assistance Program estimated that nearly five million voting-age Americans live abroad. Canada is the leading destination country for American emigrants, with some 860,000 US citizens having relocated north of the border (FVAP, 2020). Within this population, the FVAP indicates that 516,309 were eligible to vote in the US—a far higher number than in any other country, and more than half again as many as in the UK, which has the second-largest number of American emigrants.

This focus on the American-born population in Canada is theoretically advantageous for several reasons. Most residents of Canada live close to the US-Canada border, and American elections receive extensive coverage in English-language Canadian media outlets (Matthews and Satzewich, Reference Matthews, Satzewich, Satzewich and Wong2006). In any US election cycle, there would be ample opportunities for Americans living north of the border to take part in a campaign. It would do little good to examine whether directing attention to elections in the country of origin undercuts or promotes involvement in the residential country for an immigrant group that would not have much potential to become informed about politics in the country of origin.Footnote ²

Another advantage of focusing the analysis on Americans in Canada is that the relationship between political attentiveness in the US context and engagement in the Canadian context—be it “zero-sum” or positively reinforcing—may be more readily gauged. The United States and Canada are similar in many respects, sometimes described by researchers and observers as “two peas in a pod” (Bloemraad, Reference Bloemraad2011). Both are first-world industrialized democracies where English is spoken most. Few Americans in Canada would be considered economic migrants who live on the margins of society. Rather, Americans who are most open to settling in Canada tend to have personal finances that are somewhat better than average (Marrow and Klekowski von Koppenfels, Reference Marrow and von Koppenfels2020). If attention to politics in the native country is compatible with or even boosts political incorporation in the settlement country, this effect should be most noticeable for an immigrant population that is not otherwise excluded from civic life because of social or economic marginalization, as many “South to North migrants” are.

The limited prior research on the American immigrant population in Canada suggests that migrants’ attention to native-country politics does not undercut political incorporation in the Canadian context. The participants in Croucher's (Reference Croucher2011) study were said to be “perfectly at ease with and fully entitled to possess and practice citizenship in a country where they did not reside … and to practice simultaneous membership” in two nation-states (126; see also Dashefsky and Woodrow-Lafield, Reference Dashefsky and Woodrow-Lafield1992 and Matthews and Satzewich, Reference Matthews, Satzewich, Satzewich and Wong2006). Yet while everyday life in Canada is in many respects similar to life in the United States, there is no denying that the two countries differ markedly in their party organizations, governing agendas, federal structure, and formal institutions of representation (Bilodeau et al., Reference Bilodeau, White and Nevitte2010). Such differences lend credence to the “zero-sum” view of bi-nationalism. Americans in Canada whose attention is drawn into US elections may well grow less enthusiastic about taking part in Canadian politics because of the distinctive features of each political system and the challenges of engagement in the two domains.

When probing such dynamics, we will be mindful of subgroup variations. Whether or not remote attention to US politics is compatible with political incorporation in the Canadian context, the effects of transnationalism could be most evident for Americans who have not yet put down deep roots in Canada. All immigrants lead lives that to an extent fall “in between” two countries (cf. Jones-Correa, Reference Jones-Correa1998). This would be especially true for those who do not yet identify as a member of the country of residence, continue to identify as still part of the country of origin, or have lived a relatively short while in the new country. For such immigrants, the trade-off between attention to native-country politics and political engagement in the country of residence may be most apparent (Tsuda, Reference Tsuda2012).

As immigrants become more established in the residential country and grow to see themselves as bona fide members of it, their level of political engagement in that country might tend to depend more on domestic mobilizing factors, such as patterns of outreach from political parties and interest groups and less on levels of transnational connections. In the analysis below, we also investigate the potential moderating influence of education level. If attentiveness to politics in the country of origin undercuts a willingness to become involved in the residential country because of the cognitive costs of involvement in two separate political systems, this tension could be more readily apparent for less educated migrants.

Research Design and Findings

The analysis draws from an original web survey of Americans over eighteen living in Canada (N = 686). Respondents were sampled via large opt-in panels from two different sources, Qualtrics and Asking Canadians. Each firm creates large-N panels from loyalty program websites that attract a diverse set of Canadians. In both cases, the panel from which our samples are drawn was tailored to ensure that it is representative of census data as reported by the official Statistics Canada agency. Background information in both panels included place of birth, which allowed us to target American-born respondents, including Americans who were not Canadian citizens.Footnote ³ It is worth noting with respect to the representativeness of this sample that it is similar to the 464 American-born respondents in the 2019 Canadian Election Study, a large nationally representative survey that included both Canadian citizens and non-citizens (Stephenson et al., Reference Stephenson, Harell, Rubenson and Loewen2000; see the online appendix).

This survey was conducted in August of 2020, near the time of the Democratic and Republican national conventions. A few minutes into the survey, a randomized experimental treatment was applied. Approximately half of the respondents (N = 356) were exposed to vivid images of the two major party presidential nominees, Donald Trump and Joe Biden, rallying supporters (see Figure 1).Footnote ⁴ These respondents then reported their impressions of the pictures, whether the images prompted feelings of anger, hopefulness, fear and/or pride.Footnote ⁵ Respondents assigned to the control group were not exposed to US campaign images or prompted to report any personal reactions to the presidential candidates.

Figure 1. Pictures Used for the Experimental Treatment

The goal of showing randomly selected study participants colourful imagery of political rallies in the United States was to concentrate the attention of emigrants on the presidential contest south of the border and reinforce their standing as potential participants. Relatively few rallies were held during the 2020 American elections due to COVID restrictions, which would likely have made these images particularly striking. The subsequent items on reactions to the pictures provided a rationale for exposing respondents to such images and served as a manipulation check to confirm that subjects had indeed paid attention and were personally moved in some way. Over 90 per cent of the members of the treatment group reported having an affective reaction, a very high level of responsiveness demonstrating that, at that moment, American electoral politics was indeed a salient focus of attention. This experimental treatment is admittedly exploratory. No previous research has sought to direct migrants’ attention towards elections in their country of origin to observe downstream effects on orientations towards country-of-residence politics.Footnote ⁶

What are the implications of such mobilization on engagement in Canadian politics? We focus on two outcome measures: level of willingness to take part in Canadian elections when given the chance and level of identification with one of the Canadian political parties.

Interest in Participating in Canadian Elections

Table 1 presents findings from five OLS regression models. The dependent variable in each case is a four-point item on general willingness to turn out for a hypothetical Canadian election, which was asked after the treatment. This item was worded so that all respondents, including those who were not registered to vote in Canada, could report their general attitude towards taking part: “If elections were being held in Canada and you were eligible to vote, how likely is it that you would participate—very likely [4], somewhat likely [3], not too likely [2], or very unlikely [1]?”Footnote ⁷

Table 1. Effect of Treatment on Likelihood of Canadian Vote Turnout

Note: ‘*’ = p < .10; ‘**’ = p < .05; ‘***’ = p < .01. Coefficients estimated through ordinary least squares regression. The dependent variable is a four-point scale ranging from “very unlikely to vote” (1) to “very likely to vote” (4). Results from ordered logistic regression models are substantively identical to these. N = 686.

Predictors in the first model include the treatment indicator plus a four-point scale measuring the degree of identification as a “Canadian” (4 = very strongly, 1 = not strongly at all), a similar measure asking about identification as an “American,” the number of years the respondent had lived in Canada (5 years or fewer, 6 to 10 years, 11 to 20 years, or more than 20 years), and level of formal education (high school or less, some college or trade school, college degree, or post-graduate studies). The subsequent models in this table include multiplicative interaction terms to assess potential subsample variations in treatment effects.Footnote ⁸

As shown in the first specification, exposure to images of American campaign rallies had only a very slight and statistically insignificant demobilizing effect on willingness to vote in Canadian elections.Footnote ⁹ However, the second model suggests that the experimental treatment had a greater effect on Americans who do not see themselves as fully Canadian.Footnote ¹⁰ The negative sign of the interactive effect of exposure to the treatment and level of Canadian identification imply that American migrants who were exposed to scenes of US campaign rallies and did not identify as “Canadian” were significantly demobilized in the Canadian context. This offers support to the zero-sum hypothesis for this subpopulation and is in keeping with Tsuda's (Reference Tsuda2012: 635) conjecture that the “less integrated immigrants are in the host society,” the more such zero-sum relationships would likely surface.

The chart in Figure 2 clarifies this effect. Among Americans who did not identify as Canadian, the treatment reduced enthusiasm by approximately one-third of a point, a pronounced and significant effect. The impact of the treatment diminishes to insignificance for respondents who strongly or very strongly identified as Canadian.

Figure 2. Treatment Effect of Interest in Voting in a Hypothetical Canadian Election, by Strength of Identification as “Canadian”

Note: Predictions are derived from the second regression model in Table 1. The dependent variable is a four-point scale measuring interest in voting in a Canadian election (very unlikely to vote [1]; not too likely [2]; somewhat likely [3]; very likely [4]). Members of the randomly assigned treatment group were exposed to pictures of American presidential campaign rallies. Dark grey shading indicates statistical significance at the .05-level; lighter grey shading indicates significance at the .10-level.

Turning to the third regression model in Table 1, the negative sign for the “treatment X American identification” interaction term implies that migrants who identified more strongly as Americans tended to lose interest in Canadian electoral politics following exposure to the treatment, an effect that complements the findings in the second model. This coefficient, however, does not rise to the level of statistical significance (p = .31). The fourth model provides further evidence for the zero-sum view of transnationalism for less incorporated immigrants. Here we find that exposure to US campaign rally images primarily affected Americans without deep roots in Canada. Figure 3 shows these distinctive effects. For Americans who had resided in Canada for less than five years, the negative impact of the treatment is approximately as strong as that for respondents who did not identify as “Canadian.” With respect to education level (fifth model in Table 1), we find no evidence of such moderating effects.

Figure 3. Treatment Effect on Interest in Voting in a Canadian Election, by Length of Time Living in Canada

Note: Predictions are derived from the fourth regression model in Table 1. Grey shading indicates significance at the .10-level.

Effect on Canadian Partisanship

Does this conditional “zero-sum” finding carry over from voting intention to party identification in the Canadian context? After the treatment was administered, all respondents indicated whether they identified strongly with a Canadian political party (coded 4), identified weakly (3), leaned towards a Canadian party (2), or did not identify at all with one of the parties (1). Table 2 presents findings from five regression models that are comparable to those in Table 1. As was the case with the item on turning out to vote in a Canadian election, there is only a very modest negative relationship between exposure to American campaign images and the level of Canadian partisanship.Footnote ¹¹

Table 2. Effect of Treatment on Identification with a Canadian Political Party

Note: ‘*’ = p < .10; ‘**’ = p < .05; ‘***’ = p < .01. Coefficients estimated through ordinary least squares regression. The dependent variable is a four-point scale ranging from “do not identify with a Canadian party” (1) to “strongly identify with a Canadian party” (4). Results from ordered logistic regression models are substantively identical to these. N = 686.

The second specification in this table indicates, however, that the degree of identification as a “Canadian” interacted with the treatment; Americans who did not see themselves as Canadian were somewhat more responsive (p < .10 for the interaction term). Figure 4 shows this effect. Less incorporated Americans who were reminded of the US campaigns saw a drop in Canadian partisanship (−.37) that essentially matches the drop in interest in taking part in Canadian elections. The other potential moderating variables (strength of identification as an American, time spent in Canada, and education level) do not, however, have similarly significant effects on the treatment.

Figure 4. Treatment Effect on Level of Identification with a Political Party in Canada, by Strength of Identification as “Canadian”

Note: Predictions are derived from the second regression model in Table 2. The dependent variable is a four-point scale measuring level of identification with a Canadian political party (no identification [1]; “leaning” identification [2]; weak identification [3]; strong identification [4]). Members of the randomly assigned treatment group were exposed to pictures of American presidential campaign rallies. Dark grey shading indicates statistical significance at the .05-level; none of the other effects are significant (p > .10).