When evaluating politics and policies, individuals rely on both content and cues. Particularly when a policy area is personally important, individuals might be motivated to carefully consider its content, despite the time and cognitive effort that it requires to do so (Mullinix Reference Mullinix2016; Mummolo, Peterson, and Westwood Reference Mummolo, Peterson and Westwood2021). Other times, though, individuals rely on shortcuts to help guide their opinions. One particularly effective cue is a political endorsement that might come from either a liked or a disliked source (Arceneaux and Kolodny Reference Arceneaux and Kolodny2009). In particular, partisans tend to support policies endorsed by the politicians for whom they voted (Donovan et al. Reference Donovan, Kellstedt, Key and Lebo2020) and dismiss policies promoted by the politicians they do not already support (Bolsen, Druckman, and Cook Reference Bolsen, Druckman and Cook2014). One challenge of identifying whether voters evaluate the content or the political endorsements attached to any given policy is that different political leaders tend to endorse qualitatively different proposals. In this paper, we examine public opinion toward a salient immigration policy to help disentangle the influence of the policy’s content versus the endorsement it receives from a high-profile political leader.
In existing work, scholars attempt to resolve the challenge of separating a policy’s content from its cues in a variety of ways, for example, by asking respondents to evaluate fictitious policies (e.g., Baekgaard et al. Reference Baekgaard2019), or relatively unknown policies (e.g., Druckman Reference Druckman, Schaffner and Sellers2010) or by distorting the true partisan endorsement of real-life policies (e.g., Bullock Reference Bullock2011; Lavine, Johnston, and Steenbergen Reference Lavine, Johnston and Steenbergen2012). Although these useful strategies allow for carefully controlled designs with high internal validity, they lack the mundane realism that comes with a “real-life” setting. To help understand how different motivations drive policy evaluations, we identified a unique circumstance in which two ideologically contrasting politicians proposed and enacted the same policy.
On March 20, 2020, in response to the COVID-19 pandemic, the United States and Canada entered into a bilateral agreement to halt nonessential travel across their shared border and turn away asylum-seekers at their ports of entry.Footnote 1 The border would remain closed for 16 months until Canada permitted fully vaccinated Americans to enter on August 9, 2021; it would be another three months until the US permitted fully vaccinated Canadians to enter on November 9, 2021. In implementing the border closure, America’s ideologically conservative president, Donald Trump, and Canada’s ideologically center-left prime minister, Justin Trudeau, enacted identical policies. We use this setting to test our theories as to the circumstances under which individuals rely on the content of the policy versus the endorser. To do so, we administered parallel survey experiments to more than 8,000 respondents across both countries. We first test the extent to which individuals rely on the content of the policy itself. We next identify the influence of the ideologically opposed foreign leader on domestic evaluations of the border closure. Finally, we investigate how the perceived importance of the issue moderates the influence of partisan cues on support for the policy. We find that, across both countries, respondents appear to base their policy view on the content of the policy itself, with conservatives favoring the border closure and liberals opposing it. However, when we inform respondents that their neighboring leader enacted the same policy, we find that both populations shift their views in the direction that our theory predicts: Trudeau’s endorsement increases support for the policy among Americans, whereas Trump’s endorsement makes Canadians more opposed. We finally test existing theories that individuals for whom the issue is most important should resist cue-taking, whereas those who place less importance on the issue are most likely to be persuaded by a political endorsement. We test this theory in two ways and find mixed evidence that right-leaning individuals, for whom immigration is a critical issue, remain steadfast in their views on the border closure, whereas left-leaning individuals are more likely to base their opinion of the policy on its endorsers.
This work demonstrates three significant contributions: first, that content matters. Despite the political stance of the domestic leader who enacted the policy, individuals are prone to support it if the policy aligns with their own ideological positions. But we also find evidence that ideological endorsements—even from across an international border—shift public opinion. Across both countries, we see broad shifts in line with how the endorser himself is perceived. Finally, we provide real-life confirmation for the theory that issue importance reduces ideologically-motivated reasoning. Overall, this work provides broad implications for how we understand preference formation in the real world.
President Trump and Prime Minister Trudeau in the Face of a Pandemic
Looking back over the history of coseated Canadian prime ministers and American presidents, whether Pierre Trudeau and Richard Nixon, Jean Chrétien and George W. Bush, or Stephen Harper and Barack Obama, there has arguably been no greater contrast in Canadian and American leaders than that between Prime Minister Justin Trudeau and President Donald Trump. Particularly with respect to immigration, the two leaders presented themselves as starkly different. Throughout Trump’s campaign and presidency, he described immigration with inflammatory language (Jones et al. Reference Jones, Sherman, Rojas, Hosek, Vannette, Rocha, García-Ponce, Pantoja and García-Amador2021) and vowed to cut legal immigration, to build a wall on the southern border, and to enforce “extreme vetting” of applicants for refugee admission (Pierce and Selee Reference Pierce and Selee2017). Trudeau, on the other hand, came to power “symbolizing a more open Canada as concerned immigrants and citizenship” and campaigned on a promise to admit more refugees, particularly from Syria (Abu-Laban Reference Abu-Laban2020, 18). Trudeau openly defended Canada’s relatively more generous response to refugees as compared with Donald Trump (Abu-Laban Reference Abu-Laban and Mann2017), and some argue that Trudeau’s domestic approval benefitted from playing the foil to the Trump administration on the global stage (Hillmer and Lagassé Reference Hillmer, Lagassé, Hillmer and Lagassé2018).
Despite their contrasting styles and ideological differences, both leaders ultimately faced an unanticipated event that led them to similarly rethink their policies at the border. The COVID-19 pandemic reached the United States and Canada at approximately the same time in early 2020. In polling conducted March 15–21, 2020, by SurveyMonkey (now known as Momentive), 81% of Americans expressed worry about the coronavirus outbreak, as did a nearly identical 84% of Canadians. In the same surveys, 71% Americans and 73% of Canadians expressed worry that they or a family member would be exposed to the novel coronavirus. Both Trudeau and Trump were thus sufficiently motivated by COVID-19’s escalating threat to public health and constituent concerns that they mutually agreed to turn away refugee and asylum seekers at ports of entry on the Canada–US border. In doing so, they risked violating national and international laws including the right to seek asylum, the principle of nonrefoulement (not returning refugees to an unsafe location), and the mandatory protection of individuals from torture. Given the threat of the COVID-19 pandemic, both politicians nevertheless opted to close the border.
One might expect roughly similar reactions to the policy across both countries, as Canadian and American attitudes toward immigration are arguably more similar than they are different. Lipset (Reference Lipset1986) contrasted Americans’ preference for a “melting pot” approach to immigration against Canadians’ more culturally pluralist perspective, but more recent work highlights growing ideological similarities between Canadians and Americans, specifically with respect to views on immigration. Both populations hold relatively similar levels of anti-immigrant sentiment (Harell, Soroka, and Iyengar Reference Harell, Soroka and Iyengar2017) and are more willing to support higher-skilled migrants or those with financial assets. Overall, immigration attitudes are highly salient (Carmines and Stimson Reference Carmines and Stimson1980) and stable over time (Kustov, Laaker, and Reller Reference Kustov, Laaker and Reller2021).
Among both Canadians and Americans, attitudes toward immigration are largely guided by partisanship and ideology, with more conservative or right-leaning ideological and partisan affiliation predicting greater support for restrictive policies (Gravelle Reference Gravelle2016; Reference Gravelle2018b; Hawley Reference Hawley2011; Lahav and Courtemanche Reference Lahav and Courtemanche2012). In the United States, undocumented migration and the construction of the US–Mexico border wall were rallying cries for Trump’s presidential campaigns, with his Democratic opponents offering clear contrasts with their more measured approaches to migration issues (Martin Reference Martin2017). Meanwhile in Canada, Trudeau’s pro-immigration stance, highlighted most prominently during the 2015 election at the height of the Syrian refugee crisis, offered a clear contrast to both Trump and to his main electoral competitor, the Conservative Party of Canada (Gravelle Reference Gravelle2018a). It should also be noted that, in Canada, individuals who support the Conservative Party are distinct from other partisan groups in their relative opposition to immigration (Banting and Soroka Reference Banting and Soroka2020; Gravelle Reference Gravelle2018a). The content of the border closure thus aligns with conservative or right-leaning preferences. Given the high salience of the border closure, we expect that its content should guide support among constituents, as evidenced by higher support among ideologically right-leaning individuals in both countries.
H1: In both the US and Canada, respondents with more conservative (or right-leaning) political orientations will express greater support for their own country’s border closure compared with those with liberal (or left-leaning) orientations.
The content of a policy is only one factor that determines how much support it receives from the public. Individuals tend also to be influenced by the political cues associated with policies, circumventing the need or the desire to evaluate the content of a policy itself. These political cues can act as informational short-cuts that reduce the time and cognitive effort it requires to fully vet an often-complicated piece of legislation (Petersen et al. Reference Petersen, Skov, Serritzlew and Ramsøy2013). Political endorsements can also serve as in-group cues, signaling to partisans whether a particular policy is in line with their ideological allies or adversaries (Goren, Federico, and Kittilson Reference Goren, Federico and Kittilson2009). Both processes lead to the same outcome: the tendency to support policies that are endorsed by a favored politician. This can occur when individuals evaluate nearly any political object or event, including scandals (Klar and McCoy Reference Klar and McCoy2021), economic crises (Bisgaard Reference Bisgaard2015), or the quality of political leaders (Donovan et al. Reference Donovan, Kellstedt, Key and Lebo2020). A large body of work shows that voters tend to shift their policy preferences in order to agree with ideologically similar leaders (or to disagree with ideologically dissimilar ones, e.g., Abramowitz Reference Abramowitz1978; Broockman and Butler Reference Broockman and Butler2017; Druckman, Peterson, and Slothuus Reference Druckman, Peterson and Slothuus2013; Lenz Reference Lenz2012). This poses a normatively troubling challenge to democratic accountability: if voters reward like-minded politicians regardless of what they do, and conversely penalize the opposing party no matter what, it becomes difficult—even impossible—to hold representatives accountable for their actions.
Furthermore, these endorsement cues need not come from within one’s own national borders. For example, endorsements from American presidents have been shown to shape attitudes among publics outside of the United States, with non-American respondents shaping their policy opinions around their view of the American president himself. Schatz and Levine (Reference Schatz and Levine2010) find that an endorsement from George W. Bush erodes support among adults in Kyrgyzstan and Tajikistan, whereas Dragojlovic (Reference Dragojlovic2011) finds greater support among Canadians for a US-led policy initiative when it is accompanied with an endorsement from Barack Obama. Overall, it appears that populations base their reaction to foreign cues on their views toward the foreign leaders. This would suggest that an endorsement from Trump should be a deterrent for Canadians, as only 25% of Canadians viewed Trump positively (Wike, Fetterolf, and Mordecai Reference Wike, Fetterolf and Mordecai2020) and they showed widespread disapproval of Trump throughout his presidency (Gravelle Reference Gravelle2018c). Gravelle (Reference Gravelle2018c) finds that Canadians who were reminded of Trump’s support for renegotiating the North American Free Trade Agreement, for example, increased their opposition to it. We thus expect to find that Trump’s endorsement of the border closure should reduce support for the policy among Canadians.
H2.1: When Canadians learn about U.S. President Trump’s identical policy, they will support it less.
Conversely, Trudeau has largely been warmly received by the American public and American media—for example by gracing the cover of Rolling Stone in July 2017 with the headline “Why Can’t He Be Our President?” A 2017 poll revealed that 43% of Americans approved of Trudeau’s performance (with only 16% disapproving),Footnote 2 echoing his positive brand image among publics abroad (Marland Reference Marland2018; von Hlatky Reference von Hlatky, Cull and Hawes2021). We thus predict that Trudeau’s endorsement should increase support for the policy among Americans.
H2.2: When Americans learn about Canadian Prime Minister Trudeau’s identical policy, they will support it more.
Yet, not all individuals are equally reliant on shortcuts like endorsements when forming public policy opinions. In particular, when an issue is important to an individual, endorsement cues may prove less effective at shaping their public attitudes (Mullinix Reference Mullinix2016; Mummolo, Peterson, and Westwood Reference Mummolo, Peterson and Westwood2021). For example, Ciuk and Yost (Reference Ciuk and Yost2016) show, in an experimental setting, that individuals who evaluate highly salient issues are less reliant on party cues and are instead more focused on policy content, relative to those who evaluate less salient issues. With respect to border closures specifically, existing research shows that conservative and right-leaning individuals place a greater emphasis on immigration policy—for example by naming it the most important issue facing the country at higher rates—than do those who are liberal and left-leaning (Dunaway, Branton, and Abrajano Reference Dunaway, Branton and Abrajano2010; Neiman, Johnson, and Bowler Reference Neiman, Johnson and Bowler2006). Analyses of party platforms also indicate that immigration tends to be more salient for right-wing parties like the Republican Party and the Conservative Party of Canada, and not just right-wing voters (Lehmann and Zobel Reference Lehmann and Zobel2018). Given the greater salience of immigration issues among those on the right, conservatives (or right-leaning individuals) should be less susceptible to altering their views in accordance with a political endorsement, relative to those on the left. Indeed, Lahav and Courtemanche (Reference Lahav and Courtemanche2012) find that ideological liberals shift their views on immigration in response to media frames but conservatives do not.
We thus expect that political orientations, across both Americans and Canadians, will moderate the degree to which their neighbor’s endorsement influences their own evaluation of the border closure. In the American case, we expect that self-identified liberals (as well as Democrats) should increase their support for the border closure after hearing about Trudeau’s endorsement, but conservatives (and Republicans) should not. In the Canadian case, we expect respondents who identify as left-wing, or are affiliated with the parties of the left and center-left (i.e., the New Democratic Party, the Green Party, and the Liberal Party), will decrease their support for the border closure after hearing about Trump’s endorsement, but those on the right will not. Our expectations reflect what Sniderman, Brody, and Tetlock (Reference Sniderman, Brody and Tetlock1991) describe as an interaction between a situational factor (such as the information environment provided by the survey) and an individual characteristic (such as respondents’ political orientations). Our hypotheses for Canada and the US are thus:
H3.1: In Canada, respondents with left-wing political orientations will express less support for border closures when they learn about Trump’s endorsement, but those with right-wing political orientations will not.
H3.2: In the US, respondents with liberal political orientations will express more support for border closures when they learn about Trudeau’s endorsement, but those with conservative political orientations will not.
Data and Methods
To test these theoretical expectations, we conducted parallel survey-based experiments in the US (n = 4,875) and Canada (n = 4,429). We collected our data using Momentive’s Endpage methodology between July 15 and August 3, 2020. (At the time of the survey, Momentive was known as SurveyMonkey; the corporate name changed in June 2021.) Respondents were recruited from the 2–3 million survey takers who complete one of the thousands of user-created surveys on Momentive’s platform every day. After completing a survey, samples of respondents from the US and Canada (determined from their internet protocol [IP] addresses) were presented with a survey completion web page (“Endpage”) inviting them to complete a brief, voluntary, and uncompensated survey on current events (unrelated to the study they had just completed).Footnote 3 This sampling method yields large, demographically representative samples of the American and Canadian publics that have been used previously to study political behavior and public health issues in both countries (Chen, Valliant, and Elliott Reference Chen, Valliant and Elliott2019; Clinton et al. Reference Clinton, Cohen, Lapinski and Trussler2021; Gravelle et al. Reference Gravelle, Phillips, Reifler and ScottoForthcoming).
We contend that our online samples are fit to purpose given our research objectives. Online nonprobability samples provide greater demographic representation than typical convenience samples (for example, student samples), and political survey research in both the US and Canada finds few consequential differences between online nonprobability panels and probability-based samples (Ansolabehere and Schaffner Reference Ansolabehere and Schaffner2014; Breton et al. Reference Breton, Cutler, Lachance and Mierke-Zatwarnicki2017). As it relates to experimental research designs, research finds that estimates from online nonprobability samples do not differ appreciably from those obtained by other sampling methods (Coppock, Leeper, and Mullinix Reference Coppock, Leeper and Mullinix2018; Krupnikov and Levine Reference Krupnikov and Levine2014; Mullinix et al. Reference Mullinix, Leeper, Druckman and Freese2015).
Our survey design proceeds as follows (the complete interview protocols appear in our replication materials [Williams, Gravelle, and Klar Reference Williams, Gravelle and Klar2022]). The surveys began with introductory questions gauging respondent’s opinions about whether their country (either the US or Canada) was heading in the right or wrong direction and a question asking which current issue is the most important one facing the country, with a multiple-choice response scale including foreign policy, the environment, terrorism, immigration, health care, the economy, education, and an option for “other.”
After this first section, one random half of our sample (the treatment group) was told about their neighboring country’s decision to close the border. We asked for their opinion on that policy. Canadians in the treatment group were asked, “As you may know, American President Donald Trump closed the American border to refugee claimants in March 2020 to curb the spread of COVID-19. Do you approve or disapprove of this decision?” Americans in the treatment group were asked the identical question, but about Trudeau: “As you may know, Canadian Prime Minister Justin Trudeau closed the Canadian border to refugee claimants in March 2020 to curb the spread of COVID-19. Do you approve or disapprove of this decision?” Respondents were presented with a five-category response scale ranging from “Strongly Approve” to “Strongly Disapprove.”Footnote 4
All respondents were then directed to our dependent variable: a question asking them to evaluate the policy within their own country. Specifically, all Canadians (in both the treatment and control groups) were asked, “As you may know, Canadian Prime Minister Justin Trudeau closed the Canadian border to refugee claimants in March 2020 to curb the spread of COVID-19. Do you approve or disapprove of this decision?” All Americans (in both the treatment control groups) were asked, “As you may know, American President Donald Trump closed the American border to refugee claimants in March 2020 to curb the spread of COVID-19. Do you approve or disapprove of this decision?”
All respondents then completed a brief demographic section, including sex, age, educational attainment, race, and ethnicity. In this demographic section, we also measured respondents’ self-rated ideological placement and party identification. American respondents were given a standard survey-based measure of ideology in American politics, which asks them to place themselves on a five-category scale ranging from “very liberal” to “very conservative,” reflecting the American usage of the terms “liberal” and “conservative” (Ellis and Stimson Reference Ellis and Stimson2009). Canadian respondents were asked to place themselves on a five-category scale from “left” to “right,” which is standard language in discussing ideology in Canadian politics (Cochrane Reference Cochrane2015). Although these two rating scales employ different terminology, both capture self-rated political ideology in commensurate ways (Jost, Federico, and Napier Reference Jost, Federico and Napier2009).
To facilitate the interpretation of our results, the dependent variable (the evaluation of one’s own country’s policy) is rescaled from -0.5 (strongly oppose) to +0.5 (strongly support). All independent variables and covariates are similarly rescaled; continuous independent variables are centered at their means. In our paper, we present linear models for all our analyses. In the online appendix, we include tables reporting the full model results, as well as results from an alternative ordinal logit parameterization, which yields the same substantive results (Tables A1–A2 in the online appendix report the complete results linear models; Tables A3–A4 report the complete results from the ordinal logit models.)
Results
The linear models testing our hypotheses are summarized in the coefficient plots presented in Figure 1 (United States) and Figure 2 (Canada). To explicate our results, we begin by examining how political ideology predicts support for the policy in both countries. Recall that Canada’s leader is ideologically left and the American president is ideologically right. The content of the policy, in both countries, is congruent with a more right-leaning viewpoint. As we state in our first hypothesis, the content of this highly salient policy should be an important factor in how citizens formulate their responses to it. Right-leaning respondents across both the United States and Canada should express more support for the policy than should left-leaning respondents. We test this by simply regressing support for the policy on ideology as well as demographic controls. Across all of the models we will detail as we go through our results, right-leaning ideology predicts greater support for the policy (as shown in Figure 1 and Figure 2), strongly confirming our first hypothesis. For clarity, we also display the predicted effect of ideology graphically in Figure 3.
Figure 1. Coefficient Plot (United States)
Figure 2. Coefficient Plot (Canada)
Figure 3. Support for Canada–US Border Closure and Ideology (United States and Canada)
The results from our American sample (left-side panel of Figure 3) show that ideological conservatives express more support for the border closure than liberals do. Indeed, the effect of ideology is substantively large: on the -0.5 (strongly oppose) to +0.5 (strongly support) scale, respondents identifying as very liberal score 0.05 on the dependent variable, whereas self-identified moderates score 0.20, and those who identify as very conservative score 0.35. Among Canadians (right-side panel of Figure 3), the most left-wing respondents score 0.21, those in the center score 0.30, and those on the right score 0.38. These results align with our expectations expressed in H1, as political conservatives generally hold a preference for restrictionist immigration policies. It is worth noting that we obtain substantively similar results when party identification is used as an alternative measure of left–right ideological orientations. These results are reported in the online appendix (Tables A1 through A4 and Figures A1 and A2).
We next turn to our second hypothesis which addresses the effects of our treatment: specifically, informing respondents that their neighbor to the south (for Canadians) and to the north (for Americans) enacted the exact same policy. As we state in H2, we expect that Canadians will increase their opposition for the policy after learning about Trump’s endorsement. Conversely, we expect that Trudeau’s endorsement would have the opposite effect on Americans by increasing their support for the policy.
We test Hypothesis 2 with Model 1.1 (for the American sample) and Model 2.1 (for the Canadian sample). These models test the influence of the treatment on respondents’ views of their own policy, alongside a host of relevant controls. Figure 1 and Figure 2 show each coefficient in the models. In support of our second hypothesis, we find that informing Americans of Justin Trudeau’s endorsement of the border closure increases support for the policy. The coefficient for the Trudeau endorsement (Figure 1, Model 1.1) is positive and significant, indicating that Americans’ support for the border closure increases by 0.08 on the -0.5 to +0.5 scale. This suggests that, in support of Hypothesis 2, overall the American public sees border closures as more acceptable after learning about the endorsement from the politically popular Canadian Prime Minister (Marland Reference Marland2018; von Hlatky Reference von Hlatky, Cull and Hawes2021). In Figure 4 (left-side panel), we also plot support for the policy closure among those who did receive the Trudeau endorsement (the treatment group) and those who did not (the control group).
Figure 4. Effect of a Foreign Endorsement on Support for Canada–US Border Closure (United States and Canada)
By contrast, we find the opposite result in Canada, as expected (see Figure 2, Model 2.1). Informing Canadians of the Trump administration’s role in implementing the border closures serves to erode support. The coefficient for the Trump endorsement is negative and significant, reflecting a decrease in support by 0.10. Figure 4 (right-side panel) similarly plots the support among those who received the Trump endorsement (treatment group) and those who did not (the control group). This result provides additional evidence for what Gravelle (Reference Gravelle2018c) calls Trump’s “reverse Midas touch” among publics abroad and shows strong support for our second hypothesis. This suggests that policy evaluations are not based strictly on content alone but rather that political endorsements—even from a foreign leader—can influence opinion about domestic policy.
We next engage with existing theories of when cues might outweigh content by testing our third hypothesis. We expect here that respondents who place more importance on immigration are less likely to shift their opinion after being exposed to the foreign endorsement. We begin by using ideology as a proxy for immigration importance. We do so because conservative and right-leaning individuals tend to place significantly greater importance on immigration when assessing contemporary policies (e.g., Dunaway, Branton, and Abrajano Reference Dunaway, Branton and Abrajano2010; Neiman, Johnson, and Bowler Reference Neiman, Johnson and Bowler2006). Globally, right-leaning parties spend more time and attention on immigration issues than do parties to the left (Donovan and Redlawsk Reference Donovan and Redlawsk2018; Lehmann and Zobel Reference Lehmann and Zobel2018). In the US, conservatives are 30% more likely than are liberals (Dunaway, Branton, and Abrajano Reference Dunaway, Branton and Abrajano2010) and Republicans are 2–3 times more likely than are Democrats (Neiman, Johnson, and Bowler Reference Neiman, Johnson and Bowler2006) to say immigration is the most important issue. Similarly in Canada, right-leaning parties have elevated the issue of immigration by introducing restrictive measures beginning in the late 1980s (Black and Hicks Reference Black and Hicks2008, 246). With our own survey data, we model (with a binary logit) the selection of “immigration” as the most important issue on ideology, as well as a variety of other traits: partisanship, age, gender, race, and geographic region. In Canada, right-leaning ideology is a statistically significant determinant of choosing immigration as the most important issue. In the United States, the direction is in line with what we predict, though it falls short of statistical significance. (These analyses are available in Table A5 in the online appendix.) Ideology is thus a crude proxy for immigration importance but, given consistent evidence of their association across the literature, we use it as our first test of hypothesis three.
If those who place more importance on an issue are less likely to be persuaded by political cues, then we would expect conservatives to be less responsive to the cue and liberals to place more weight on the foreign endorsement. We test this by interacting our treatment (the foreign leader’s endorsement) with ideology. The results are displayed with Model 1.2 in Figure 1 (for the American sample) and Model 2.2 in Figure 2 (for the Canadian sample). We also illustrate the effects in Figures 5 and 6. Turning first to the American sample, the ideology × Trudeau endorsement interaction is highly significant, supporting H3.1. The coefficient is negatively signed, indicating that the Trudeau endorsement serves to mute the effect of ideology by raising support among self-identified liberals while having no appreciable effect among conservatives. Put differently, as Americans become more conservative, the cue has a diminishing influence over their opinion (see Figure 5). Among the most liberal, support for the border closure increases from -0.04 (in the absence of the Trudeau endorsement) to 0.14 (when they read Trudeau’s endorsement). By contrast, the most conservative are steadfast in their support of the border closure regardless of the Trudeau endorsement, with a predicted score of 0.35, suggesting that information about neighboring-country policies fails to influence the immigration policy preferences among this segment of the American public who place greater importance on the issue.
Figure 5. Support for Border Closure, Ideology, and Trudeau’s Endorsement (United States)
Figure 6. Support for Border Closure, Ideology, and Trump’s Endorsement (Canada)
Turning to the Canadian sample, the interaction between ideology and the Trump endorsement is highly significant, supporting H3.2. The coefficient is positively signed, indicating that the endorsement amplifies (rather than mutes) the effect of ideology. In other words, among those who receive Trump’s endorsement, policy support is weakened among those on the political left and not at all among those on the right. Among left-wing Canadians, support for the border closure drops from 0.31 (in the absence of the Trump endorsement) to 0.11 (when they receive the Trump endorsement). Right-wing Canadians, on the other hand, are effectively unmoved in their support of the border closure by the Trump endorsement, with predicted scores of 0.38 (in the absence of the Trump endorsement) and 0.39 (when they receive the Trump endorsement), which are statistically indistinguishable. This suggests that the “Trump allergy” among Canadians is concentrated among the political left.
Across both samples, liberals (or those on the left) but not conservatives (or those on the right) shift their views on their country’s border policy in response to a political endorsement from their neighboring country’s leader. These effects are robust to the inclusion of various sets of covariates included in the models including sex, age, race and ethnicity, education, and region. We suggest that this is due to the greater emphasis that conservatives place on immigration issues, as indicated by their greater likelihood to name immigration as the most important issue. We next turn to our most stringent test of this theory; instead of using ideology as a blunt stand-in for issue importance, we now interact our treatment with a variable indicating whether a respondent indeed selected immigration as the single most important issue facing their country (in our American sample, 135 respondents selected immigration as the single most important issue; 193 did so in our Canadian sample). We would expect that individuals who identify “immigration” as the single most important issue are significantly less likely to be swayed by a political cue. This measure has its own drawbacks: just because someone does not choose immigration as the single most important issue does not mean they view it as unimportant. Additionally, only a small minority of respondents select immigration, dramatically reducing our sample size. Nevertheless, this additional test of hypothesis three helps to collectively demonstrate the moderating role of issue salience in how respondents react to a political cue.
We display the results for this test with Model 1.3 in Figure 1 (for Americans) and Model 2.3 in Figure 2 (for Canadians). We also plot these predicted effects in Figure 7. With our American sample (left-side panel of Figure 7), we find that the effect of the Trudeau endorsement is positive and significant among respondents who do not name immigration as the most important issue. However, we do not find the expected relationship among those who name immigration as the most important issue. For this subset of the American public, the Trudeau endorsement leads to an even greater push toward support for the border closure.
Figure 7. Support for Canada–US Border Closure, Immigration Salience, and Foreign Endorsements (United States and Canada)
Turning to our Canadian sample (right-side panel of Figure 7), we find strong evidence in support of our hypothesis. The effect of the Trump endorsement is significant and negative among respondents who do not select “immigration” as their most important issue. Individuals who do not view immigration as the most important issue become significantly more opposed to it when they learn about Trump’s endorsement. However, this negative effect of the Trump endorsement is canceled out among those for whom immigration is the most important issue (the higher-order term is positive and significant).
In Canada, we thus find that individuals who do not name immigration as the single most important issue shift their views in accordance with a political cue, but those who place the highest importance on immigration remain unmoved from their initial evaluation. The Canadian sample thus strongly supports our third hypothesis: the importance of an issue elevates its content over political cues as individuals form their preferences. We can speculate that perhaps the mixed findings among the American sample might be due to an elevated importance of immigration among American liberals, relative to left-leaning individuals in Canada, given the relative prominence of immigration as a political issue in the United States. We devote more attention to the mixed support for our third hypothesis across the American sample in the discussion section.
Robustness Checks
Our study assumes that Americans and Canadians are sufficiently familiar with each other’s leader to be potentially influenced by their position on the border closure. In our initial study, we clearly identified Justin Trudeau as the Prime Minister of Canada to our American respondents, but we did not provide any information about his ideological orientation vis-à-vis Trump. Given the oft-documented lack of knowledge about foreign policy issues among the American mass public (Almond Reference Almond1960), we conducted an additional survey that allows us to measure the degree to which Americans are familiar with Prime Minister Trudeau. Using Momentive’s Endpage methodology, we surveyed 2,118 Americans during August 25–30, 2021. We first provided one random half of the sample with five foreign leaders (whom we named, just as we did in the initial survey). We asked the respondents to place these leaders, along with former President Trump and President Biden, on an ideological spectrum ranging from left (0) to right (10). We then asked our sample to rate each leader on a 0–100 thermometer scale (from cold to warm). As an additional check on the American public’s knowledge of foreign leaders, we asked the second random half of respondents if they could name these leaders before placing them on the ideological scale and rating them.
Although Americans’ “benign ignorance” of Canada is often noted (Gravelle Reference Gravelle2014), we find that most Americans can identify Trudeau and his political orientations—even when he is not named. When asked to name the Prime Minister of Canada, roughly three in five respondents (63%) could correctly name Trudeau. More importantly, even without naming Trudeau, the American public is able to locate him politically to Trump’s left. The difference in left–right scores when Trudeau is named (mean = 3.60) and not named (mean = 3.43) is small and not significant (mean difference = 0.17, SE = 0.15, p = 0.264). As expected, Americans also locate Trudeau (mean = 3.52) some distance to the left of Trump (mean = 7.33; mean difference = 3.81, SE = 0.14, p < 0.001; see Table A7 in the online appendix). Americans also provide similar ratings of Trudeau whether he is referred to by his office alone (mean = 50.2) or by his office and name (mean = 46.5). Such a difference is of little substantive importance (mean difference = -3.70, SE = 1.77, p = 0.034; see Table A8 in the online appendix). In short, the American public shows broad knowledge of Canada’s Prime Minister, including his ideological views (even among those who are unable to recall his name).
Finally, in our models, we use liberal–conservative (or left–right) ideological placement to distinguish liberals from conservatives. An alternative approach might be to use party affiliation. Partisanship operates rather differently across the US (a two-party system) than it does in Canada (a multiparty system), with Canadians showing less loyalty to one party and more volatility in their party affiliation (Stephenson, Scotto, and Kornberg Reference Stephenson, Scotto and Kornberg2004). More recent scholarship, though, finds that partisans in both the US and Canada are increasingly sorted ideologically (Kevins and Soroka Reference Kevins and Soroka2018; Lelkes Reference Lelkes2016), implying that party attachments can serve as an alternative measure of political orientations, if only as a stringent test of the robustness of our findings. Our data support this assertion in both the two-party American case and the multiparty Canadian case: mean ideology scores for partisan groups show meaningful differences and a clear left–right ordering. Furthermore, the scoring of party platforms on a left–right continuum (by the Manifesto Project [Volkens et al. Reference Volkens, Burst, Krause, Lehmann, Matthieß, Regel, Weßels and Zehnter2021]) produces the same left–right ordering of parties observed among the American and Canadian mass publics (see Figures A3–A6 in the online appendix). In short, both parties and partisans can be meaningfully ordered from left to right. Thus to assess the robustness of our findings, we also use party identification as a measure of individual political orientations in place of ideology. These models yield the same substantive results as our models above (see Model 1.5 in Table A1 and Model 2.5 in Table A2 in the online appendix; see also Figures A1 and A2).
Finally, an alternative explanation for our American results centers on ideological differences in foreign policy orientations: American liberals may be more inclined toward cooperative internationalism, less inclined toward isolationism, and thus more favorably disposed toward policies that receive endorsements from foreign leaders. We similarly draw on data from our August 2021 survey to test this possibility. In brief, we find that ideology shapes broad foreign policy postures (cf. Gravelle, Reifler, and Scotto Reference Gravelle, Reifler and Scotto2020), yet these postures do not fully mediate the effect of ideology on perceptions of Trudeau specifically or foreign leaders generally (see Tables A9 and A10 in the online appendix). In light of these results, we contend that our focus on ideology and its relationship to support for border restrictions is appropriate.
Discussion and Conclusion
Both content and cues help guide individuals’ evaluations of cognitively complicated policies. The decision to close the Canada–US border to all travelers including refugee claimants and asylum seekers as a pandemic-related precaution is no doubt a complex policy that balances public health and humanitarian considerations. In evaluating this policy, individuals are asked to weigh their support for migration against their concern for a growing pandemic. What makes our setting so unique is that two ideologically opposed leaders enacted an identical policy in cooperation with one another. This context allows us to disentangle the influence of the policy itself from the influence of the leader himself. We find that content matters. Some might argue that right-leaning individuals in the US prefer the policy relative to their left-leaning compatriots because the incumbent President himself was right-leaning. But we are able to show that in Canada, where the incumbent Prime Minister was left-leaning, support is still higher among right-leaning citizens. The high awareness that each country has of their neighboring leader—and his ideology—is what allows us to test for the influence of political endorsements on this policy. We find, in fact, that both publics shift their views on the policy once they learn about their neighboring leader’s endorsement. Finally, ideological divisions in the importance that individuals place on the policy allow us to test the moderating role of issue importance on the reliance on political endorsements. We find strong evidence in Canada and mixed evidence in the US that individuals who place the greatest importance on the issue are indeed least susceptible to political cues. This work provides a unique real-life setting to test influential theories regarding opinion formation.
To further interrogate our proposed mechanism—that is, the perceived immigration itself—we used a particularly stringent test: comparing those who identified immigration to be the most important issue facing the country against those who did not. Here we found that the importance of immigration moderates the effect of a political endorsement but in different ways among Americans and Canadians. We speculate that the difference has to do with the distinct political contexts on either side of the border. The United States and Canada enacted an identical policy—one that ideologically aligns with conservatives and with the conservative American president but not with the center-left Canadian Prime Minister. This allows American conservatives to support the policy without having to reject their like-minded leader. But, for Canadian conservatives, supporting the policy means also supporting an out-party leader. Existing literature suggests that important issues are less susceptible to motivated reasoning. In Canada, this means that those who place less importance on the issue—and more likely to be liberal—will lower their support for a policy that they already dislike based on an endorsement from a politician whom they also dislike. Those who view the issue as important—and who are more likely to be conservative—are unlikely to support the policy even more when they learn that a disliked politician supports it.
In the United States, the political context is different. Individuals who place the least importance on immigration—and who tend to be liberal—are willing to increase their support once learning that an admired liberal politician endorses it. But, contrary to our expectation, individuals who place the greatest importance on the issue also shifted their support. In this case, though, they increased their already-high support in response to an admired politician’s endorsement. This suggests individuals who care deeply about an issue might move in response to an admired endorser but not one who is disliked. These findings complement existing research arguing that American progressives’ attitudes bear the imprint of Trudeaumania redux (von Hlatky Reference von Hlatky, Cull and Hawes2021), whereas the views of the Canadian left are shaped by an aversion to Trump (Gravelle Reference Gravelle2018c).
We believe that Canada and the United States make a well-suited match for this study. Despite cultural differences, research largely suggests that the Canadian and American populations are more similar to one another than they are different. Both populations exhibit similar ideological divides (Kevins and Soroka Reference Kevins and Soroka2018), they share similar attitudes toward immigration overall (Harell et al. Reference Harell, Soroka, Iyengar and Valentino2012), and both are experiencing rising affective polarization (Gidron, Adams, and Horne Reference Gidron, Adams and Horne2020), though this is substantially more extreme in the US (Boxell, Gentzkow, and Shapiro Reference Boxell, Gentzkow and Shapiro2020). Both populations are diverse with respect to linguistic, racial, and ethnic cleavages (and one limitation of our study is that we do not pursue differences in how these subgroups within each country respond to the cue). Yet there are also some important differences between the two countries, which raise unanswered questions for future research. First, Trump is particularly disliked in Canada and, perhaps to a lesser extent Trudeau is particularly liked among Americans. This might mean our findings present an upper bound on the potential influence of a foreign leader.
This study also takes place in the context of a global pandemic. It is worth considering how attitudes toward the pandemic may overshadow concerns relating to immigration or for the border closure more broadly. Across both countries, citizens with right-leaning or conservative orientations tend to express less concern for the public health crisis and greater reluctance to comply with precautionary measures (Clinton et al. Reference Clinton, Cohen, Lapinski and Trussler2021; Courtice et al. Reference Courtice, Quinn-Nilas, Bickram, Witoski, Hoskin and Blair2021; Druckman et al. Reference Druckman, Klar, Krupnikov, Levendusky and Ryan2021). These same groups of individuals tend to favor restrictionist border policies (Gravelle Reference Gravelle2018b) like the one that Trudeau and Trump implemented to help mitigate COVID. Left-leaning or liberal individuals faced the opposite conundrum: more concern about the pandemic coupled with greater opposition to a closed border. Individuals of all political orientations thus had to make a trade-off as they formed their policy preference. The pandemic thus complicates how respondents think about the border, though we do not believe it does so in a way that should bias our results in either direction.
We find that, overall, both content and cues matter. Conservatives and right-leaning people favored the border closure at higher rates than those who are liberal and left-leaning regardless of whether the policy was put in place by their left-leaning leader (in Canada) or their conservative president (in the US). When told that the leader of the neighboring country endorsed the same policy, though, it was conservatives, for whom immigration issues tend to be of greater importance and concern, who remained steadfast in their views. Liberals, who generally place less emphasis on issues of immigration, showed themselves to be more susceptible to a political endorsement (or antiendorsement), even from a political leader outside their own borders.
Our cross-national study demonstrates that even when policies address highly salient issues like immigration, political cues from both within and beyond one’s borders can sway opinions. Previous work demonstrates that endorsement cues may be less effective (or ineffective altogether) at shaping public attitudes for highly salient public policy issues like immigration and border security (Mullinix Reference Mullinix2016; Mummolo, Peterson, and Westwood Reference Mummolo, Peterson and Westwood2021). In line with Dunaway, Branton, and Abrajano (Reference Dunaway, Branton and Abrajano2010), we find in a large cross-national survey experiment that immigration is a more important issue for right-leaning Americans and Canadians relative to their more liberal, or left-leaning, compatriots. Therefore, we argue that those on the political right were relatively “immune” to the effects of national and neighboring-country leader cues. Our work thus makes a significant contribution to our understanding of how individuals form attitudes toward complex new policies. Although conventional policy positions on highly salient issues may overrule party cues at home, ideological cues from a well-known foreign leader appear to condition attitudes toward salient policy issues among a subset of the population.
Supplementary Materials
To view supplementary material for this article, please visit http://doi.org/10.1017/S0003055422000351.
DATA AVAILABILITY STATEMENT
Research documentation and data that support the findings of this study are openly available at the American Political Science Review Dataverse: https://doi.org/10.7910/DVN/G1BZ23.
ACKNOWLEDGMENTS
Earlier versions of this article were presented at the 2021 annual conferences of the American Association for Public Opinion Research, the Midwest Association for Public Opinion Research, the International Society for Political Psychology, and the Canadian Political Science Association, as well as at the University of Pennsylvania Department of Political Science. We thank the participants of those sessions for their constructive feedback. We similarly thank the APSR anonymous reviewers for their engagement with our work and numerous helpful suggestions. The final product is much improved for it. We also thank Jon Cohen, Jack Chen, and Austin Pettis at Momentive (then known as SurveyMonkey) for their assistance in collecting the survey data we rely upon here. All three authors contributed equally to this work.
CONFLICT OF INTEREST
The authors declare no ethical issues or conflicts of interest in this research.
ETHICAL STANDARDS
The authors declare the human subjects research in this article (specifically the survey data collection and data dissemination procedures) was reviewed and approved by legal counsel for Momentive for compliance with applicable privacy laws. The authors affirm that this article adheres to the APSA’s Principles and Guidance on Human Subject Research.
Open access
Comments
No Comments have been published for this article.