Democratic elections are traditionally viewed as society strengthening institutions. They can encourage citizen investment in the government, strengthen civil society, and lead to an increase in pluralistic values and discourse (Dahl 1973; Tolbert et al. 2003; Skocpol and Fiorina 2004). Yet while elections are an essential part of a functioning democracy, recent elections in the United States, Great Britain, Germany, and Italy have made clear that elections can be divisive even in historically strong democratic societies.

This divisiveness can occur within a range of issues; however, we focus on understanding the influence of elections on the increasingly politicized topic of immigration (Rueda 2005; Häusermann et al. 2013; Beramendi et al. 2015; Garand et al. 2017). Greater opposition to immigration not only makes it more difficult for politicians to work together to create sustainable solutions, but can also lead to outbursts of violence among disaffected voters, as seen with recent protests in Germany, Denmark, Sweden, and others. This is especially true for elections held during the European refugee crisis and for European countries where far-right parties have made significant electoral gains.

In this paper, we examine how elections influence attitudes toward immigration during European elections. Using as-if random assignment in individual interview dates for the European Social Survey between 2002 and 2015, we utilize the proximity of interviews to elections across 28 European countries to test whether increased proximity to elections influences preferences regarding immigration. The as-if random assignment of interview dates allows for a causally-identified examination of how elections might influence individual attitudes toward immigration (Eifert et al. 2010; Singh et al. 2012; Aragonès et al. 2015; Singh and Thornton 2019; Muñoz et al. 2019). Note that by design we are not testing for long-term effects of elections on attitudes toward immigration, but rather how attitudes vary with changing distance to elections over time.

We find that proximity to elections encourages individuals to be less welcoming to immigrants both before and after the election has occurred. This effect is consistent across various country and electoral contexts and model specifications. Yet proximity to elections has a heterogeneous effect depending on the prevalence of anti-immigration stances in the electoral discourse. Using new data from Dancygier and Margalit (2020) on party platforms from the Immigration in Party Manifestos (IPM), we find that the overall negative effect of proximity is driven primarily by country-elections where anti-immigration rhetoric is part of the political discourse. In country-elections with the lowest levels of anti-immigrant rhetoric, however, elections have a positive effect on attitudes toward immigration.

This country-election heterogeneity indicates that political parties and rhetoric play an essential role in mediating the relationship between elections and attitudes toward immigration. Yet there are two different avenues through which this effect might occur at the individual level. On the one hand, building off the rich literature on party influence on supporters, elections might be influential because of their effect on loyal voters (Popkin 1995; Sniderman et al. 1991; Lupia 1994). On the other hand, elections also increase awareness and salience of electoral issues among less partisan individuals, which could encourage more moderate individuals to shift their stances (Zaller et al. 1992; Wood and Vedlitz 2007; Chiang and Knight 2011). Our results suggest that neither avenue is entirely correct, with elections influencing voters across the ideological spectrum.

This work provides four primary contributions. First, it speaks to growing research on the role of political institutions in influencing attitudes toward immigration. Second, it adds to the rich literature on party influence, showing how parties influence individuals closer to elections. Third, it contributes to the literature on far right parties and issue ownership, showing that anti-immigration platforms have an effect on the public.

Finally, it provides one possible reason why anecdotal evidence would indicate that individuals are becoming increasingly hostile to immigration, yet research shows that attitudes have actually improved in the last decades (Caughey et al. 2018)—the increased focus on public opinion during electoral periods leads to overestimation of negative attitudes on the topic. Therefore, these findings have implications for how politicians can interpret anti- (or pro-) immigrant mandates from electoral periods.

Theoretical Background

While there is significant work focused on understanding the role of cultural and economic concerns on attitudes toward immigration (Hainmueller and Hopkins 2014), research on the influence of political institutions on attitudes toward immigration is more limited. In examining to what extent political institutions have an impact on attitudes towards immigration, scholars have noted that citizens in countries with more restrictive immigration policies are more likely to support immigration. In contrast, immigrants are more likely to assimilate when immigration policies are less restrictive (Hainmueller et al. 2017). Other scholars have also examined the impact of political inclusion and political power of immigrants on electoral politics (Dancygier 2010; Dancygier et al. 2015). Through analyzing the short-term effects of elections on immigration, we add to mounting evidence that political institutions can be an important third source of influence.

Elections and Public Attitudes Towards Immigration

It seems likely that elections would have an influence on attitudes toward immigration. Elections as a black box include a variety of political factors that should influence individual attitudes, including updated and more visible party platforms, priming, increased social salience, and increased information on political issues (Zaller et al. 1992; Steenbergen et al. 2007; Wood and Vedlitz 2007; Chiang and Knight 2011). Examined separately, each of these political factors have been found to influence attitudes toward immigration in a variety of contexts. For example, work on immigration has made the link between elite rhetoric on immigration and individual attitudes. Some work has argued that elite cues help shape opinions on immigration (Jones and Martin 2017), and noted that discussions of immigration are more prevalent during political campaigns (Lenz 2009).

Each of these political sources of influence, however, depend on the premise that immigration is, in fact, an issue being discussed in the political sphere. Indeed, scholars note that in Europe’s political climate, anti-globalization and anti-immigration issues have become important sources of electoral cleavage (Steenbergen et al. 2007; Odmalm 2011). This is part of a broader trend toward electoral dealignment, with a decrease in the salience of traditional cleavage issues such as class (Rueda 2005; Häusermann et al. 2013). Alongside immigration, the “new politics” that have arisen emphasize nationalism, identity politics, and welfare state reform (Beramendi et al. 2015; Garand et al. 2017).

Capitalizing on these changes, far right parties have risen to prominence in European politics. Prominent examples include the National Rally/Front in France, the Swedish Democrats, and the Golden Dawn in Greece. Within the electoral sphere, anti-immigration appeals are often seen as especially effective in mobilizing constituents and voters along identity lines. Extreme immigration positions are especially tempting for populist parties, who want to distinguish themselves from larger mainstream parties who embrace more moderate stances toward immigration (De Sio and Weber 2014; Abou-Chadi and Orlowski 2016). Scholars have also found that the presence of anti-immigration parties has a ‘contagion effect’, helping move other political parties to adopt more anti-immigration stances (Spanje 2010), especially during times of crisis (Downes and Loveless 2018).

Scholars are divided over how effective these strategies are for parties that do not identify as far-right. Some provide evidence that left-wing parties make electoral gains with anti-immigration appeals (Spoon and Klüver 2020). Others note no effect or even some backlash (Meijers and Williams 2019; Abou-Chadi and Wagner 2020). Yet despite the increasing centrality of immigration to party ideologies, the effects of elections on public attitudes towards immigration are not yet well understood.

This leads to our first hypothesis:

H1

Elections will influence individual attitudes toward immigration when immigration is a politically salient issue.

Note that our hypothesis implies the following:

H1A

In elections with positive political rhetoric surrounding immigration, we expect to see a positive effect on individual attitudes.

H1B

In elections with negative political rhetoric surrounding immigration, we expect to see a negative effect on individual attitudes.

Who Gets Influenced by Elections?

Within a given country-election, this effect might have heterogeneous effects across individuals. Past research indicates two primary potential sources of individual-level variation. On the one hand, there is a rich literature arguing that individuals with strong party affiliations will shift their attitudes to align with those of the party (Popkin 1995; Sniderman et al. 1991; Lupia 1994). This might even occur among people who are drawn to agree with that party for other, possibly economic, reasons, and then adjust their views on immigration to line up with that of the party (Sniderman et al. 2004).

On the other hand, less informed or more moderate individuals are most likely to shift their attitudes in response to new issue information that elections might provide (Zaller et al. 1992; Wood and Vedlitz 2007; Chiang and Knight 2011). This information might even occur after elections, with individuals using the issue stances of winners as proxies for those of the populace more broadly (Dekeyser 2019).

These two effects are not mutually exclusive—elections can influence moderates and hardliners simultaneously. This thus leads to two additional hypotheses:

H2

Elections will encourage more negative stances toward immigration among highly partisan individuals.

H3

Elections will encourage more negative stances toward immigration among politically moderate or less partisan individuals.

Both of these effects would be expected to influence individual attitudes both before and after elections.

Data

In order to test our hypotheses, we use data from the first seven rounds of the European Social Survey (ESS), between 2002 and 2015. These data allow us to assess the impact of over 500 elections from before and after the financial crisis of 2008. These data contain detailed individual-level information, including consistent survey items that measure perceptions of different types of immigration for 28 European countries.Footnote 1

Our outcome variable is attitudes toward immigration to one’s own country. We take the average of three questions about immigration in the ESS. These questions ask whether the survey respondent’s country should allow immigrants of either the same race or ethnic group, immigrants from a different race or ethnicity, or immigrants from outside Europe. For these questions, respondents can then state whether they want many, some, few, or none. The exact questions, their means, and their standard deviations are listed in the online appendix in Table A.1.

In order to identify the effect of exposure to elections on attitudes towards immigration, we utilized the information in the ESS about an individual’s interview year, month, and day. Though countries were surveyed by the ESS in waves, usually lasting around six months, individual interview dates were arguably as-if random within that frame. We exploit this exogenous variation to analyze the effect of the proximity of elections on individual attitudes toward immigration.Footnote 2

To create our proximity measure, we merged the ESS data with data on all elections held in Europe between 1998 and 2017, taken from the European Election database.Footnote 3 This data includes all elections—parliamentary, presidential, local, regional, and referendums. We defined proximity to elections as the number of days between a person’s interview date and the nearest election.Footnote 4

In Fig. 1, we present the distribution of our distance to elections measure. As noted in the figure, most respondents were surveyed between 100 and 500 days to elections. We present the density of distance to elections for each country-round in Fig. A.3 in Online Appendix. Countries show variation on the distance to elections both within and across rounds.

Fig. 1
figure 1

Distribution of distance to elections

Full size image

To examine Hypothesis 1, elections will influence individual attitudes toward immigration when immigration is a politically salient issue, we use data from the Immigration in Party Manifestos (IPM) (Dancygier and Margalit 2020). Specifically we use their immigration sentiment metric (Net Stance) which measures the overall sentiment of immigration rhetoric in all party platforms. This measure ‘subtracts the percentage of claims [about immigration] that are negative from those that are positive [about immigration].′

Figure A.4 shows the level of immigration rhetoric aggregated at the country level for all elections. The figure indicates that party rhetoric is most positive in countries such as Germany and Norway and most negative in countries such as the UK, Denmark, and Italy.

To examine Hypothesis 2 and Hypothesis 3, we use responses to a question that asks individuals to place themselves on a 1 to 10 left-right ideological spectrum. We classify left-leaning individuals as those who identify between 1 and 3, moderates as those who identify between 4 and 6, and right-leaning individuals as those who identify between 7 and 10. The distribution of responses to this question can be found in the online appendix (Fig. A.5).

Empirical Strategy

To estimate the effect that elections have on attitudes towards immigration, we use a panel with country fixed effects that exploits the temporal variation in individual proximity to elections.

We propose the following model:

$$\mathrm{Att}_{\mathrm{i},\mathrm{j},\mathrm{t}} = \alpha _{j} + \beta{\mathrm{elections}}_{\mathrm{i},\mathrm{j},\mathrm{t}}+ Z_{i,j,t}\cdot \Phi +\gamma _j^{\ast} \delta _r +\varepsilon _{i,j,t} $$
(1)

where the outcome variable \(Att_{i, j,t}\) refers to an individual’s i attitude towards immigration in country j on day t. The main explanatory variable \(elections_{i,j,t}\) is our individual distance to elections measure in country j on day t. We also include \(Z_{i,j,t}\) to model individual-level covariates. The model also includes \(\gamma_j,\) country fixed effects to control for common factors for each country. We also run models using the country-level covariates of GDP, inequality, and migration.

To account for the clustered nature of our data, we use clustered standard errors at the country-round level. To test country and individual-level heterogeneity, we use flexible estimation strategies that allow for non-linearity as suggested in Hainmueller et al. (2018). These methods allow us to avoid reporting relationships that are highly model-dependent. We also provide more traditional regression results on subsets of the data to demonstrate subgroup heterogeneity.

Results

We find that proximity to elections encourages more negative attitudes toward immigration. We also find that political discourse appears to play an important role in driving this result, and that elections are influential for voters across the ideological spectrum.

Impact of Elections on Attitudes Towards Immigration

Figure 2 presents the results of the basic relationship between proximity to elections and proximity to elections. We find that individuals are more likely to hold more favorable views of immigration further from elections. Specifically, people are more supportive of immigration more than 1 year from elections (2.60), and are most negative towards immigration very close to elections (0–30 days).

Fig. 2
figure 2

Elections and attitudes toward immigration. This figure shows that attitudes toward immigration increase nearly monotonically with distance from elections. The x-axis shows distance from elections, and the y-axis attitudes toward immigration, with higher values on the y-axis indicating more positive attitudes toward immigration. Vertical bars indicate 95% confidence intervals

Full size image

In Table 1, we assess the relationship between elections and attitudes towards immigration in an OLS framework. Our preferred specification includes demographic controls, country fixed effects, and clusters our standard errors at the country-round level (see Specification 1). The results in column 1 suggest that a 30-day increase from elections is associated with an increase of .001 in attitudes towards immigration. These results suggest that an increase of one standard deviation in distance to elections is comparable to the effect of an additional year of education—a substantive and statistically significant effect. Column 2 reports similar results using country-level controls such as GDP or immigration-level controls. This supports our hypothesis that proximity to elections has an overall negative impact on attitudes towards immigration.

Table 1 Impact of elections on attitudes towards immigration
Full size table

Heterogeneity by Party Platform and Political Ideology

In line with Hypothesis 1, we find evidence suggesting that party rhetoric plays a role in shifting attitudes towards immigration closer to elections.

Specifically, we examine whether proximity to elections in which parties embrace more anti-immigrant rhetoric has a greater effect on attitudes toward immigration. For ease of interpretation, we recoded our immigration sentiment measure so that higher scores reflect more anti-immigrant sentiment. Figure 3 demonstrates how the effects of electoral proximity vary over different levels of elite rhetoric.Footnote 5 The figure indicated a positive linear relationship between distance to elections proximity and anti-immigration sentiment. This implies that that the negative effects of elections are concentrated where parties compete with anti-immigration platforms.

Table 2 performs a similar analysis by separately analyzing countries with low, medium, and high anti-immigrant rhetoric. The findings are in line with the observable implications that would be expected if political discourse on immigration is the primary influence on individual attitudes. At positive levels of immigration rhetoric, elections appear to have a positive effect—precisely the expected result if political discourse plays a fundamental role. In contrast, the negative effects of elections are concentrated in places with party platforms with slightly negative or highly negative sentiment towards immigration.

Table 2 Impact of elections on attitudes towards immigration, by party rhetoric
Full size table

The overall negative effect of elections on attitudes toward immigration, however, is driven by country-elections with moderate and high levels of anti-immigrant rhetoric, which is the majority of country-elections. In these countries, proximity of individual interview dates is associated with more negative attitudes toward immigration, as in the aggregated result.

Fig. 3
figure 3

Impact of elections on attitudes towards immigration, by anti-immigration sentiment. See Table 1. Plot generated using the interflex package

Full size image

Finally, to examine to what extent different types of voters are influenced by proximity to elections, we also subset our data by self-reported political ideology. Table 3 suggests that ideology does not play a strong role moderating the overall effect of elections. Subset results are similar in magnitude for all groups, although slightly stronger for left and center-leaning voters. Thus our analysis does not find sufficient evidence to support Hypothesis 2: or Hypothesis 3: which suggest stronger effects for partisan or moderate voters. Rather, our results suggest that voters across the ideological spectrum, including those who identify as left-wing, are influenced by elections.

Table 3 Impact of elections on attitudes towards immigration, by political ideology
Full size table

Conclusions

This paper suggests that individuals hold more negative attitudes toward immigration during electoral periods. This change is most significant in elections where anti-immigration sentiment is part of the political discourse, and among individuals across the political spectrum.

These results provide a range of implications. First, social scientists have long lauded the positive role of elections, which unquestionably play a vital part in the functioning of a healthy democracy (Skocpol and Fiorina 2004; Tolbert et al. 2003). Yet our findings highlight the potentially negative externalities of elections in established democracies. Second, we add to the growing literature on the impact of political institutions on attitudes toward immigration through an examination of one of the most central of democratic institutions. Finally, we underscore the fact the political and party rhetoric can have a significant influence not only on party stalwarts, but also on moderate individuals.

Broadly, what this means for the future of elections is mixed. On the one hand, the rise of far-right parties throughout Europe indicates that elections will continue to have a negative impact on attitudes towards immigration. This is especially true for elections that were held during the European Refugee Crisis which saw a massive increase in the number of migrants/refugees coming to Europe. This crisis not only saw a rise in anti-immigration rhetoric, but also led to the collapse of governments and additional elections.

Yet, our work also notes that these results are not permanent. Our work implies that pro-immigrant rhetoric from parties in future electoral cycles could have positive effects on individual attitudes, as seen in countries with very low levels of anti-immigrant rhetoric.

Despite its temporary nature, the influence of elections on attitudes toward immigration can have longstanding effects, whether through the negative impact on social cohesion of vitriolic anti-immigrant rhetoric or perceived anti-immigrant mandates that lead to hardline policies by elected officials. As a result, the negative attitudes embraced during electoral periods can have negative externalities that endure long after the polls close.