Photo Credit: S. Hofschlaeger/pixelio.de
Germany is going to the polls in a few days. Of those allowed to cast a ballot, 12.2% have an immigrant background, i.e. were born abroad or have at least one parent who was born abroad. This group of citizens might become key to the elections results, since in 167 of the 299 single member districts, their vote choices could actually decide about who is elected (Leininger and Lagodny 2021). So, what does previous scholarly work tell us about the voting behavior of citizens of immigrant origin? They generally favor left-wing parties (Bird et al. 2010), albeit with some exceptions as suggested for instance by the finding from Germany that voters of German-Russian origin show strong support for the right-wing AfD (Goerres et al. 2018). Aiming to understand the voting behavior of this group of voters in more detail, in this blog post, I engage with the question: How does the share of district habitants with immigrant background shape the voting choices of citizens of immigrant origin?
A number of studies already indicates that context matters for vote choices: The characteristics of voters’ neighborhoods and districts affects their voting behavior (see e.g. Johnson et al. 2002, Anderson, 2000; Alvarez et al., 2000; Weng 2015). For visible minorities, I believe that such patterns are likely to occur as well, in particular as a consequence of the presence of group members in their district. For minority members, numerical strength is a key factor determining their action (Kanter 1977; Dahlerup 1988). If they form a clear minority, they tend to adapt to the behavioral patterns of the majority, because they are highly visible in this situation and try to assimilate into the majority. However, as their numerical presence increases, they become more confident to show their identity. As a result, voters of immigrant origin in districts with few habitants of immigrant origin should show a similar behavior as the majority – which might include supporting radical right parties to an equal degree as the majority. By contrast, voters of immigrant origin in districts with many habitants sharing this characteristic should display strong group-specific voting behavior. Following this rational, I expect to find that support for leftist parties by citizens of immigrant origin is more pronounced in districts with large shares of citizens of immigrant origin and less in districts with low shares of citizens of immigrant origin.
To test this proposition, I investigate voting behavior in the German federal election in 2017 using voter-level data from the Comparative Study of Electoral Systems (CSES, Module 5) combined with district-level data from Leininger and Lagodny (2021). The voter information contains the immigrant background of respondents (yes [1st and 2nd generation] or no), their vote choice (all six parliamentary parties) and electoral district number. Additional background information will serve as control variables for the analyses such as year of birth, sex, level of education (7-point ISCED scale), and an 11-point left-right self-placement. The district information includes the population share with an immigrant origin, as well as information on the vote share for all parliamentary parties, population density, and two variables capturing whether there are MPs of immigrant origin from these districts elected via lists or in districts (yes or no). In the analysis, I focus on studying vote choice for the second electoral tier (party-list), since the first-past-the-post races in the first tier of the mixed member proportional electoral system might lead to different and more complex dynamics. In the 2017 election, six parties were elected to parliament and three of them belong to the left side of the ideological spectrum (SPD, Left party, Green party) while the remaining three belong to the right side of the ideological spectrum (CDU/CSU, AfD, FDP).
The data set contains information for 1495 respondents – 342 of them have an immigrant background. Table 1 shows the distribution of party support of respondents with and without immigrant background and reveals that there are no substantial differences between the two groups at the aggregate level within the sample. The gap in electoral support does not exceed 1.5%-points in any case. Notably, the table also points towards potential problems in the CSES sample: Voters supporting the AfD are strongly underreported in the survey data.
Table 1: Distribution of party votes in the CSES sample and overall electorate.
A correlational analysis confirms this pattern and shows that there is no clear bivariate link between an immigrant background and vote choice in the data (see correlation coefficients displayed in the first column of Table 2). The district composition, however, appears to be more influential. The share of habitants with an immigrant background displays a positive correlation with Green party votes and a negative correlation with CDU/CSU, AfD and Left party votes (see Column 2 in Table 2). Table 2: Correlation coefficients of party choice with immigrant origin and share of habitants of immigrant origin in the district.
Table 2: Correlation coefficients of party choice with immigrant origin and share of habitants of immigrant origin in the district.
How do these patterns play out in a multivariate analysis that also takes demographic weights into account? I estimate separate logistic regression models for the likelihood to vote for each of the six parties. The main explanatory variables in the models are the immigrant origin of respondents and the share of habitants of immigrant origin in the district (Model 1) as well as their interaction (Model 2). Full models are displayed in Table 3 and 4 below this blog post, while I highlight some key findings using Figure 1 and 2.
The first round of models for each party indicates the effect of immigrant origin on the likelihood to vote for the respective party independent of the district context. The results again suggest that respondents’ immigrant status has no effect on the likelihood to vote for the CDU/CSU as well as the three leftist parties (SPD; Green party, Left party). By contrast, respondents of immigrant origin indicate a lower likelihood to vote for the liberal FDP and – contrary to the original assumption – a higher likelihood to support the right-wing populist AfD. The composition of the district does not unfold a stand-alone effect on vote choices according to these models.
The second round of models introduces the interaction between the immigrant origin of respondents and the district composition. This effect is displayed visually in Figure 1 for the parties on the right side of the ideological spectrum and in Figure 2 for the parties on the left side of the ideological spectrum. For the CDU, the results again indicate no substantial effect of any kind, meaning that, even when taking the share of habitants of immigrant origin in the districts into account, voters with an immigrant background are neither particularly repelled nor attracted by Chancellor Merkel’s party.
For the AfD, Figure 1 clarifies that the citizens of immigrant origin supporting the extreme right usually live in districts with a very low population of immigrant origin. Within this group, the likelihood to vote for the AfD appears to be slightly more pronounced compared to citizens without immigrant background in districts with low diversity. This effect ceases as the presence of citizens of immigrant origin increases. In this sample, voters of immigrant origin do hence not only show no difference when living in districts in which they constitute a clearly minority, but actually display above-average support for the party most hostile to foreigners and citizens with foreign roots. However, remembering that the share of AfD voters in the sample is not representative of the population, uncertainty about the generalizability of this finding persist.
A similar pattern occurs for the FDP, albeit at an overall lower level of party support within the group of citizens with an immigrant background: If the share of population of immigrant origin is rather low, voters of immigrant origin display neither a clear preference nor clear refusal for the liberal party. However, as the population becomes more diverse, the likelihood of citizens with an immigrant background to vote for this party turns negative, meaning that they are less likely to make such a vote choice compared to citizens without an immigrant background in a comparable district environment.
Figure 1: Marginal effect of immigrant origin on likelihood to vote for CDU/CSU, AfD and FDP depending on share of district population of immigrant origin (with 90% confidence intervals).
Figure 2: Marginal effect of immigrant origin on likelihood to vote for the SPD, Green party and Left party depending on share of district population of immigrant origin (with 90% confidence intervals).
The opposite pattern emerges for two of the leftist parties: The higher the share of population with an immigrant background, the stronger the positive effect of voter’s own immigrant experience on the likelihood to vote for the Social democrats and the Greens. The gap in support for these two parties between citizens with and without an immigrant background hence becomes more pronounced as voters live in a highly diverse environment.
While the Left party belongs to the same ideological spectrum as the SPD and the Green party, no comparable pattern emerges for this case. The null-findings for the Left party and the CDU/CSU indicate that, beyond party ideology, party specific factors might matter. In case of the Left party, for instance, the debates surrounding the party stances on asylum policy might have hampered its reputation within the group of citizens of immigrant origin.
The evidence presented in this blog post indicates that the environment in which individuals live shapes their voting behavior substantially. The differences in the vote choices of citizens with and without immigrant background become less pronounced for parties on the right side of the ideological spectrum as the environment becomes more diverse, while the gap for parties on the left side of the ideological spectrum expands. My hint is that this is a consequence of group dynamics, but potentially, individuals might also self-select into districts with more or less habitants of immigrant origin depending on their political preferences. An interesting avenue to take this research further might be to run survey experiments with voters with and without an immigrant background from different types of districts to uncover which factors trigger such processes and how they think about politics more broadly and their identity as citizens of immigrant origin more specifically.
By Corinna Kroeber in September 2021
Table 3: Logistic regression of party choice in the 2017 election on immigrant origin of respondents and share of immigrant origin population the district for the CDU/CSU, AfD and FDP.
Table 4: Logistic regression of party choice in the 2017 election on immigrant origin of respondents and share of immigrant origin population the district for the SPD, Green party and Left party.
 The data actually refers to the 2021 electoral districts, but district modifications were moderate.
Alvarez, R.M., Nagler, J. and Willette, J.R. (2000). Measuring the relative impact of issues and the economy in democratic elections. Electoral Studies 19(2): 237-253.
Anderson, C.J. (2000). Economic voting and political context: a comparative perspective. Electoral Studies 19(2): 151-170.
Bird, K., Saalfeld, T. and Wüst, A.M. (2010). The political representation of immigrants and minorities: Voters, parties and parliaments in liberal democracies. Milton Park: Routledge.
Dahlerup, D. (1988) From a small to a large minority: women in Scandinavian politics. Scandinavian Political Studies 11(4): 275-298.
Goerres, A., Spies, D.C. and Mayer, S.J. (2018). Deutsche mit Migrationshintergrund bei der Bundestagswahl 2017: Erste Auswertungen der Immigrant German Election Study zu Deutschtürken und Russlanddeutschen. Available at: https://www.researchgate.net/publication/323534980_Deutsche_mit_Migrationshintergrund_bei_der_Bundestagswahl_2017_Erste_Auswertungen_der_Immigrant_German_Election_Study_zu_Deutschturken_und_Russlanddeutschen.
Johnson, M., Shively, W.P. and Stein, R.M. (2002). Contextual data and the study of elections and voting behavior: connecting individuals to environments. Electoral Studies 21(2): 219-233.
Kanter, R.M. (1977). Some effects of proportions on group life: Skewed sex ratios and responses to token women. American Journal of Sociology 82(5): 965-990.
Leininger, A. & Ladogny, J. (2021). Wähler*innen mit Migrationshintergrund als wahlentscheidender Faktor. Ihr potentieller Einfluss auf die Bundestagswahl 2021. Available at: https://vielfaltentscheidet.de/waehlerinnen-mit-migrationshintergrund-als-wahlentscheidender-faktor/.
Lu Chung D. W. (2017). Can economic profit influence public opinion? Observing generational change on cross-strait relations in Taiwan. Asian Journal of Comparative Politics 2(4): 362-384.
The Comparative Study of Electoral Systems (2021). CSES MODULE 5 THIRD ADVANCE RELEASE [dataset and documentation]. JULY 20, 2021 version. doi:10.7804/cses.module5.