‘Bravo to our AfD allies for this historic showing! It is a new sign that the people of Europe are waking up…’ ∼Marine Le Pen, following [the] 2017 German elections
Introduction
In 2017, following a disappointing performance in the 2013 elections where the party barely missed gaining seats in the Bundestag, the Alternative for Germany (AfD) intentionally set out to make itself more palatable to radical right voters in Germany. Notably, the party was considered relatively moderate by radical right standards in 2013, focusing predominantly on its anti-EU positions during the Greek debt crisis and fiscal conservatism (Arzheimer Reference Arzheimer2015; Berbuir et al. Reference Berbuir, Lewandowsky and Siri2015; Dilling Reference Dilling2018; Grimm Reference Grimm2015; Schmitt-Beck Reference Schmitt-Beck2018). However, by 2017, the party had become much more representative of the radical right family in the rest of Europe, championing more xenophobic and nativist cultural positions on immigration and adopting the now standard populist rhetorical style (Arzheimer and Berning Reference Arzheimer and Berning2019). This push to be viewed as a true champion for radical right values culminated in a rally in early 2017 in which the AfD leader, Frauke Petry, hosted prominent radical right leaders from across Europe such as Marine Le Pen of the National Rally (RN), Geert Wilders of the Dutch Party for Freedom (PVV), and Matteo Salvini of Italy’s Northern League (LN). Petry’s intent was to emphasize the party’s shift to the right and signal its allegiance to figures who are associated with the radical right family, highlighting the AfD’s arrival as a true radical right party.Footnote 1 The goal of this article is to evaluate how this set of strategic choices to embrace the party family label, illustrated by the AfD from 2013 to 2017, influences how parties are perceived by voters.Footnote 2
Research has shown that parties are responsive to politics that transpire abroad (Böhmelt et al. Reference Böhmelt, Ezrow, Lehrer and Ward2016; Senninger et al. Reference Senninger, Bischof and Ezrow2022), yet the question remains whether voters are also picking up on the growing convergence of policy positions within party families or whether transnational politics is only a party-level phenomenon. Recent work has shown that voters are responsive to elite cues and endorsements from abroad (Williams et al. Reference Williams, Gravelle and Klar2022) and will adjust their voting behaviour based on electoral returns from foreign countries (Turnbull-Dugarte and Rama Reference Turnbull-Dugarte and Rama2022), yet the literature has not tackled whether voters are paying attention to a party’s relationship with its transnational allies when evaluating the party. Work on the effect of elite interactions and who a party is associated with more generally demonstrates that voters are responsive to a party’s relationships with other parties and elites (Adams et al. Reference Adams, Weschle and Wlezien2021; Fortunato and Adams Reference Fortunato and Adams2015; Lee et al. Reference Lee, Santoso and Stevenson2018; Williams and Whitten Reference Williams and Whitten2015), which suggests that a party’s decision to associate itself further with its party family label can have a meaningful impact on how it is perceived by voters.
In this article, I argue that parties can align or distance themselves from the rest of their party family as a tool to strengthen their brand and communicate their positions to voters. Parties that embrace the rest of their party family signal commitment to ideological goals most commonly associated with the party family, while parties that push back and distance themselves from their sisters abroad signal that they may not be credible champions for core issues. Therefore, how a party positions itself relative to the rest of its family should inform how voters evaluate where political parties fall on the ideological spectrum.
To test this argument, I rely on the Comparative Study of Electoral Systems (CSES, Modules 1–4), which provides valuable individual-level survey data alongside country-election-level data from multiple elections and countries in Europe. Combining this survey data with data on party positions from the Comparative Manifestos Project (CMP), I find that the further a party positions itself from its transnational party family, the greater the gap between a voter’s perceptions of the party’s left-right ideological position and its actual position. I then assess this relationship using panel data from the German Longitudinal Election Study (GLES) to look within individuals, finding additional support for my expectations. These results speak to the increasingly transnational nature of party politics and how voters can evaluate their own party relative to their expectations about the party’s broader transnational party family. I conclude by discussing the implications of this work and the avenues for future research.
How Voters Evaluate Party Positions
There is a large literature on whether voters update their perceptions of a party’s position in response to party behaviour. According to Fernandez-Vazquez (Reference Fernandez-Vazquez2019), voters can tell the difference between cheap talk and strategic position-taking, and do not update their evaluation of a party when they deem the change to be a performative, strategic shift. Furthermore, Adams et al. (Reference Adams, Ezrow and Somer-Topcu2011) find no evidence that voters respond to the policy shifts of political parties (see also Adams et al. Reference Adams, Ezrow and Somer-Topcu2014). Similarly, Fernandez-Vazquez and Somer-Topcu (Reference Fernandez-Vazquez and Somer-Topcu2019) find that voters do not update their perceptions of a party’s position unless there is a change in leadership that prompts more attention to the party’s policy changes.
Conversely, several studies challenge the conclusion that voters do not pay attention to a party’s strategic position-taking. Fernandez-Vazquez (Reference Fernandez-Vazquez2014) finds that voters do update their perceptions of a party’s position in response to policy changes, contradicting the evidence presented by Adams et al. (Reference Adams, Ezrow and Somer-Topcu2011). Along these lines, studies show that voters are responsive to press releases (Somer-Topcu et al. Reference Somer-Topcu, Tavits and Baumann2020), leadership changes (Fernandez-Vazquez and Somer-Topcu Reference Fernandez-Vazquez and Somer-Topcu2019; Söderlund and Grönlund Reference Söderlund and Grönlund2024; Somer-Topcu Reference Somer-Topcu2017), electoral debates (van der Meer et al. Reference van der Meer, Walter and Aelst2016), and policy outputs (Adams et al. Reference Adams, Bernardi and Wlezien2020). This strand of literature generally argues that voters do pay attention to parties and update their evaluations when they are presented with new information about a party’s position. Subsequent work on the importance of information availability further supports this view, with several studies finding that the more parties are covered in the overall information environment, the more voters will accurately update their perceptions of a party’s positions (Adams et al. Reference Adams, Ezrow and Somer-Topcu2014; Banducci et al. Reference Banducci, Giebler and Kritzinger2017; Fernandez-Vazquez and Somer-Topcu Reference Fernandez-Vazquez and Somer-Topcu2019).
Importantly, work on voter perceptions has established that voters pay attention to not only what the party is doing itself but who it is associated with (Williams and Whitten Reference Williams and Whitten2015). Parties can leverage this to their advantage through strategic interactions that voters can use to evaluate the party’s positions. For instance, Adams et al. (Reference Adams, Weschle and Wlezien2021) find that the extent of coverage of cooperative interactions between political elites near elections results in voters evaluating those parties to be closer to each other on the left-right ideological spectrum. Similarly, Lee et al. (Reference Lee, Santoso and Stevenson2018) find experimental evidence that voters evaluate the positions of political parties based on the type of interactions those parties have with other parties more than they use the party’s manifestos. Cooperative interactions lead voters to place parties closer together than they may actually be on the ideological spectrum (Lee et al. Reference Lee, Santoso and Stevenson2018). Work on governing coalitions, the most public and binding form of political cooperation, demonstrates further that voters infer party positions based on who the party chooses to associate itself with. Political parties that govern together are often perceived to be ideologically similar (Falcó-Gimeno and Fernandez-Vazquez Reference Falcó-Gimeno and Fernandez-Vazquez2020; Fortunato and Adams Reference Fortunato and Adams2015; Fortunato and Stevenson Reference Fortunato and Stevenson2013; Klüver and Spoon Reference Klüver and Spoon2020; Spoon and Klüver Reference Spoon and Klüver2017).
As this work demonstrates, while there is some debate over whether voters respond to the policy positions of parties, there is general agreement that who a party associates itself with shapes how voters view the party. However, this work focuses, importantly, on the associated relationships voters have with the party within a domestic polity. To date, the literature on whether voters update their perceptions of a party has not examined the impact of party behaviour and its transnational associations on a voter’s perceptions of the party.
Embracing the Transnational Brand
To address this gap, I argue that parties can leverage their membership in a broader party family to strengthen their brand at home by attempting to present themselves as a dutiful ally for party family goals or as a distant cousin to party family allies that may be stigmatized. This, in turn, has consequences for how voters evaluate the positions of the political party based on the positions and traits they associate with the party family label. To build this argument, it is first important to discuss how party families create transnational political brands that transcend borders. The family label must be able to communicate information on the policy positions of parties associated with the party family’s brand. Lastly, voters must be responsive to that information and update their perceptions.
Transnational Family Brands
Parties belong to party families, which emerged as an academic tool to classify parties based on similar emergence patterns and the shared ideological goals of member parties (Lipset and Rokkan Reference Lipset, Rokkan, Lipset and Rokkan1967; Mair and Mudde Reference Mair and Mudde1998). However, we know that these labels play an important role in structuring European politics. On the party side, the transnationalization of the European Parliament led to the creation of European Party Groups (EPGs) that organize interests at the European level based on similar ideologies, which have come to closely mirror party family classifications (Hix et al. Reference Hix, Noury and Roland2007).
For voters, I argue that party families have their own political brands that carry a set of associations about what it means to be a member of that family. Often studied at the party level, research has shown that political brands are the set of associations voters have towards political parties that include their policy positions, leaders, and rhetoric but can also include colours, images, and the party’s name (Avina, Reference Avina2024; French and Smith Reference French and Smith2010; Grynaviski Reference Grynaviski2010). In turn, these political brands are useful information shortcuts for voters when determining which parties to support, and parties with weak brands, or a weaker set of identifiable associations for voters, are less likely to receive strong electoral support (Lupu Reference Lupu2013, Reference Lupu2016; Nielsen and Larsen Reference Nielsen and Larsen2014).
Importantly, party families have similar political brands that can help shape how voters evaluate their own political parties. For example, radical right parties are often associated with extreme positions on immigration and can often be stigmatized and considered taboo. As Bolin et al. (Reference Bolin, Dahlberg and Blombäck2022) demonstrate, policies associated with the radical right party family are often perceived to be more extreme, even if the same policy is proposed by a mainstream party. Furthermore, parties associated with the stigmatized positions of radical right parties often perform poorly at the polls, while parties that can distance themselves from the stigmas surrounding the radical right are better able to establish themselves in the party system (Mendes and Dennison Reference Mendes and Dennison2021; van Spanje and Azrout Reference van Spanje and Azrout2019; van Spanje and Van Der Brug Reference van Spanje and Van Der Brug2007).
Beyond the radical right, other party families in Europe have their own transnational political brands. Green parties, for example, have all embraced the sunflower logo and the colour green on many of their posters, leaflets, and images.Footnote 3 Case studies into party brands find that voters from different countries hold similar views of green parties as the core champions of environmental issues, and are often associated with the outdoors, nature, and climate change (Grimmer and Grube Reference Grimmer and Grube2019; Rutter et al. Reference Rutter, Hanretty and Lettice2018). These studies also identify similarities for other party families as well, such as the social democratic, liberal, and conservative families (French and Smith Reference French and Smith2010; Grimmer and Grube Reference Grimmer and Grube2019; Rutter et al. Reference Rutter, Hanretty and Lettice2018). These shared associations across countries demonstrate that voters have a common understanding of what it means to belong to a certain party family. As Fortunato (Reference Fortunato2021, 53–54) observes, when asked to rank-order hypothetical parties based solely on the names of the party family it belongs to, voters in the UK and the Netherlands can accurately discern where these parties fall in the ideological spectrum. As this work highlights, party families do have a set of associations akin to a political brand that transcends national borders such that voters in different contexts will have a set of expectations for what it means for a party to be associated with a given family.
Party Families and Policy Diffusion
For a party to be able to leverage its association with a party family label to help it remind voters of its positions and values, one piece of information that must be associated with the label is a general sense of the positions and values of the larger party family. Importantly, as research on party policy diffusion observes, party families have become ideologically cohesive over time and have converged on similar policy profiles across borders. Bélanger and Wunsch (Reference Bélanger and Wunsch2021), for instance, find, when examining Euroscepticism among the radical right, that these parties have become more similar in how they talk about the EU. Camia and Caramani (Reference Camia and Caramani2012) find that all party families have converged on similar ideological positions among both parties and voters of the various families. Caramani (Reference Caramani2015) presents robust evidence that European party systems have ‘Europeanized’ as parties of the same family have converged such that party systems in European countries have increasingly begun to look similar to one another.
This convergence is due to transnational party policy diffusion in which policies adopted by one party spread across borders to other parties (Böhmelt et al. Reference Böhmelt, Ezrow, Lehrer and Ward2016; Gilardi Reference Gilardi2010, Reference Gilardi, Carlsnaes, Risse and Simmons2013; Gilardi and Wasserfallen Reference Gilardi and Wasserfallen2019). Evidence shows that parties learn what works and what policies do not provide much upside based on what foreign parties do abroad (Gilardi Reference Gilardi2010). For example, Böhmelt et al. (Reference Böhmelt, Ezrow, Lehrer, Schleiter and Ward2017) show that the greater the vote share of the incumbent party, the more parties from other countries will emulate its policy positions since dominant incumbents send strong signals about what policy positions constitute a successful electoral platform (see also Juhl and Williams Reference Juhl and Williams2022; Schleiter et al.,Reference Schleiter, Böhmelt, Ezrow and Lehrer2021). Importantly, Ezrow et al. (Reference Ezrow, Böhmelt, Lehrer and Ward2021) show that this emulation from successful foreign leaders is, in fact, an electorally successful strategy. The positions of a successful foreign leader signal where the median voter in the electorate is in their polity, which can signal where the median voter may fall in a party’s home electorate. Work on the success of the radical right has shown that the widespread electoral relevance of this insurgent party family is a function of radical right parties learning from each other and emulating successful strategies (Kallis Reference Kallis2013; Rydgren Reference Rydgren2005; Van Hauwaert Reference Van Hauwaert2019).
Additionally, the multi-level politics of the European Union helps national parties adopt similar policies as their allies abroad (Senninger and Bischof Reference Senninger and Bischof2018; Wolkenstein et al. Reference Wolkenstein, Senninger and Bischof2020). The EU is a powerful conduit for party policy diffusion due to the presence of European Party Groups (EPGs), which coordinate parties of similar ideological interests across national boundaries to simplify legislative politics in the EU. Recent work has shown that these EPGs facilitate the transnational diffusion of policies as parties of the same EPG are more likely to adopt policies from abroad than parties of different EPGs (Senninger et al. Reference Senninger, Bischof and Ezrow2022). Importantly, work has found that these networks extend beyond institutionalized channels, with parties of similar party families using shared networks of ideological actors and activists abroad to facilitate the transnational diffusion of policies (Schleiter et al. Reference Schleiter, Böhmelt, Ezrow and Lehrer2021).
Consequences of Embracing the Transnational Brand
Building on this literature of the increasingly cohesive nature of party families around common policy positions, I argue that the party family a party is associated with can be a useful heuristic for voters when evaluating the party’s ideological position. As demonstrated by the extensive work on how voters evaluate party positions, who a party is associated with in the domestic arena plays a large role in shaping how voters perceive the party (Adams et al. Reference Adams, Weschle and Wlezien2021; Spoon and Klüver Reference Spoon and Klüver2017; Williams and Whitten Reference Williams and Whitten2015). Voters will make assumptions about parties based on who they interact with, often assuming parties that frequently cooperate with each other are ideologically similar (Adams et al. Reference Adams, Weschle and Wlezien2021). For voters, it is not always what the party says it is going to do but who it chooses to do it with. The literature on perceptions of governing coalitions highlights this dynamic, observing that junior partners are often perceived to be more ideologically similar to the Prime Minister (PM) they govern with, despite adopting agendas and platforms that are distinct from the PM party’s platform (Fortunato and Adams Reference Fortunato and Adams2015; Fortunato and Stevenson Reference Fortunato and Stevenson2013). In turn, these patterns in voter behaviour demonstrate that who a party is linked to, such as their party family, will help inform how voters perceive the party just by being connected to the party family label. Bolin et al. (Reference Bolin, Dahlberg and Blombäck2022) demonstrate that voters will view the same policy as more extreme if it is tied to a radical-right party than a mainstream party.
This is especially the case since party families have converged on similar policy profiles within party families, making the party family label an association that carries increasingly relevant and visible policy-related information. When a voter hears that a party belongs to the Green or Social Democratic or Liberal Party family, they can infer that that means the party advocates for a certain set of policy proposals. Just as coalitions signal what policies the party is associated with through its governing record, the party family label communicates to voters a set of policy positions they can expect the party to adopt based on what voters associate with the family. Therefore, how a party chooses to align or distance itself from the party family brand, and the associations that it may carry, will inform how voters evaluate the party’s positions.
A party may choose to embrace the transnational brand by adopting policies associated with its allies abroad, appearing alongside prominent figures from them, and leaning into the label to signal its commitment to broader ideological fights. Alternatively, a party may want to distance itself from the party family to avoid being mis-associated with characteristics that could hurt its brand, such as being perceived as too extreme or as an ineffective governing option (Adams et al. Reference Adams, Böhmelt, Ezrow and Schleiter2022; van Spanje and Azrout Reference van Spanje and Azrout2019). Importantly, the AfD’s attendance at the radical right rally mentioned at the outset captures these tensions for a party and whether it should embrace its party family. Many within the AfD feared that by appearing alongside Le Pen in 2017, their party would lose a part of its own identity by aligning itself with what was perceived to be a more radical party in the National Rally and Party for Freedom (PVV).Footnote 4 On the other hand, others in the party argued that by appearing to be a member of a united radical-right front in Europe following the Brexit referendum and the election of Donald Trump in the US could bolster the AfD’s credibility as a radical-right champion and highlight for voters its embrace of radical right principals.
Importantly, choosing to break from the party family brand creates uncertainty about where the party may stand. As discussed, party family brands communicate some degree of information about where a party may stand on the issues. By distancing itself from the party family brand, a voter is forced to weigh the party’s own rhetoric and actions against what it knows about the party family brand. In turn, this creates ambiguity about the party’s positions since voters are given competing pieces of information. If a party is associated with the social democratic party family, a family that is often associated with left-leaning economic and socio-cultural positions, but the party itself advocates more centre-right positions on these issues, voters are receiving two signals about the party’s positions. The first is the party’s own rhetoric, which would signal centre-right positions; the second is the party family it is associated with, which signals centre-left positions. As we know from work on voter perceptions, when voters have competing signals of a party’s position, the accuracy of their perceptions decreases as they attempt to make sense of the different messages they receive about the party’s positions (Falcó-Gimeno and Fernandez-Vazquez Reference Falcó-Gimeno and Fernandez-Vazquez2020; Fernandez-Vazquez and Somer-Topcu Reference Fernandez-Vazquez and Somer-Topcu2019). For instance, as Falcó-Gimeno and Fernandez-Vazquez, (Reference Falcó-Gimeno and Fernandez-Vazquez2020) argue, coalition participation only changes voter attitudes about the party when the coalition a party joins is at odds with what voters expect from the party, creating uncertainty about the party’s goals by deviating from expectations. Meanwhile, coalitions that reinforce voters’ views of the party do not change perceptions.
Inversely, by aligning with the party family label, the party sends reinforcing signals about the party’s positions to voters from its own programme and the party family brand. A social democratic party that advocates centre-left positions in its platform sends complementary signals, which communicate to voters that the party is an advocate for centre-left policies. Importantly, parties have strong incentives, given the need for differentiation, to send as many signals as possible that reinforce each other about the party’s positions. For example, junior partners need to overcome the competing signals of their participation in a government that is not aligned perfectly with the party’s platform by taking additional steps to signal their true positions (for example Fortunato Reference Fortunato2021; Nonnemacher and Spoon Reference Nonnemacher and Spoon2023; Sagarzazu and Klüver Reference Sagarzazu and Klüver2017). Similarly, work on political brands establishes that strong political brands are ones where the associations voters hold about the party are self-reinforcing, meaning one piece of information does not contradict another to make the brand uniquely identifiable (Gunderson Reference Gunderson2024; Nielsen and Larsen Reference Nielsen and Larsen2014).
As a result, when the party family label and the party’s positions are aligned, voters receive information that reinforces each other, helping the voter accurately place the party. By contrast, if the party family label and the party’s position are in contrast with each other, this creates ambiguity about the parties’ true positions, forcing voters to make more assumptions about the party’s positions, which can breed uncertainty as voters must weigh the importance of each signal. Thus, the central hypothesis of this paper is as follows:
H1: As the divergence between a party and its party family increases (decreases) from one election to the next, the accuracy of a voter’s perceived ideological position decreases (increases).
Data and Methods
To test these expectations, I rely on data from the CSES, which provides individual-level survey data on a cross-national sample of individuals from across the world and through multiple elections. The value of this survey for the purposes of this study is that it asks voters to place parties on the left-right scale based on where they think the party falls, allowing for cross-national comparisons of how voters perceive the positions of their parties. In this study, I focus my efforts on Western Europe since the left-right dimension is similar cross-nationally within the region, alleviating concerns that the left-right scale could have different meanings in different contexts (Bakker et al. Reference Bakker, Jolly, Polk and Poole2014). Additionally, De La Cerda and Gunderson (Reference De La Cerda and Gunderson2024) observe that the applicability of the traditional family labels I focus on performs stronger at matching parties in the West than in the East. As a result, I have data on elections in fourteen countries from 1996 to 2015.Footnote 5
In addition to the CSES, I also rely on data from the CMP, which provides comprehensive cross-national data on the positions of parties via content analysis of their manifestos and party programmes to capture a party’s ideological position. Specifically, for the purposes of this study, I look at the calculated left-right (RILE) position of the party in each election as this corresponds with the scale used to gauge how individuals perceive the ideological positions of their parties in the CSES. The RILE score from the CMP data is calculated using the percentage of the total quasi-sentences in a manifesto coded under the CMP codebook as left or right, with the difference between the two percentages being the overall RILE score. Ranging from −100, which represents the most left-leaning position, to 100, which represents the most right-leaning position, these values were then rescaled to a 0–10 scale to ease interpretation.Footnote 6
Perception Accuracy
To evaluate the accuracy of a voter’s perceptions of a party’s left-right positions, I combine the CMP data on the RILE position of parties with responses to a series of questions in the CSES that ask individuals to place the political parties on the RILE scale from 0 to 10. Each voter-party combination is a unique observation, meaning individuals are in the dataset up to nine times, based on the number of parties in their party system included in the CSES (see Spoon and Klüver Reference Spoon and Klüver2017 for a similar construction). Then, I calculate the absolute value of the difference between those perceptions with the actual position of each party as measured by the CMP (rescaled to a 0 to 10 scale), which gives me an indicator for the distance between a voter’s perception and the actual position of each party. Taking the absolute value of the difference gives me the accuracy gap between a voter’s perception of a party and its actual position. Higher values represent a greater accuracy gap, or lower accuracy, while lower values equal greater perception accuracy. This creates a continuous measure, which ranges from 0.003 to 8.53 with a mean score of 2.07. This variable is coded at the voter-party level. Figure 1A displays the distribution of this placement accuracy variable.Footnote 7
Divergence from the Party Family
The key independent variable to test my hypotheses in this study is the ideological divergence of the party’s RILE position from the average position of the rest of its party family.Footnote 8 To capture the ideological position of party families, I developed a novel measure of the party family position, which used the RILE position of every member of the party family and the classifications of parties into one of several party families. In this article, I focus on the eight most prominent party families. In order from left to right, these are the radical left, greens, social democrats, centre, liberals, conservatives, Christian democrats, and the radical right.Footnote 9
I then collect the ideological positions on the left-right ideological spectrum for all parties that belong to the same party family in other West European countries to calculate the average RILE position of the rest of the transnational party family. For example, to calculate the average position of the rest of the party family for the Swedish Social Democratic Party (SAP) before the 2014 elections, I computed the average RILE position of all social democratic parties in West European countries, except the SAP, from the most proximate election in the respective countries before 2014. This process was repeated until I had the average position of the rest of the party family for each domestic party in the dataset for each election. While I do not expect voters to have perfect knowledge of the transnational party family’s exact ideological position, I do assume that they have a general sense of where these families are likely to fall on the ideological spectrum.Footnote 10 Therefore, taking the average position of the party family serves as a useful proxy for the associated position of the party family in the minds of voters.Footnote 11
Importantly, in many countries, party families can be represented by more than one party. For example, in Norway, multiple parties belong to the ‘radical left’ party family (the Socialist Left Party and the Red Party). To compare these parties’ positions to the rest of the transnational party family when calculating the average, I exclude all domestic parties of the same party family if there is more than one in each country. Therefore, the average position for the “radical left” transnational party family in Norway, for example, does not include the Socialist Left Party or the Red Party when calculating the average. This ensures that my measure is only recording the average position of the party family using the position of foreign parties and is independent of where domestic parties are located.
Figure 1 plots the distribution of domestic party positions and the average position of the rest of the party family. As this figure shows, there are clear ideological clusters of like-minded parties. As we would expect, there is clear clustering among green (green), liberal (yellow), and radical right (purple) parties. Social democratic parties (red), conservatives (blue), and Christian democrats (black) are more dispersed but still demonstrate clear groupings, which makes sense given these parties tend to adopt a more broad-appeal strategy centred around the middle (Somer-Topcu Reference Somer-Topcu2015). The centre parties of Scandinavia (orange) and the radical left (pink) are the least clustered of the party families.Footnote 12
Distribution of Domestic Party and Party Family Ideological Positions.
Notes: On the x-axis is the average position of the rest of the party family while the y-axis is the ideological position of each domestic party. The scale for both runs from 0 to 10, where lower values indicate more leftward positions while higher values represent rightward positions. Each dot represents one domestic party in one election. The black line through the graph represents where there is no difference between the two measures. The closer a dot is to this line, the more aligned that party is with the rest of its party family.
Source: Comparative Manifesto Project.
This clustering provides confidence in the operationalization of the transnational party family position as all clusters appear in the ideological position that we would expect for each party family. Statistically, there is very little deviation among the party’s ideological positions on the RILE scale (y-axis) for many of the party families. Standard deviations, for example, are as follows in order from left to right: radical left (0.884), green (0.435), social democrats (0.667), centre (0.56), liberal (0.78), conservative (0.772), Christian Democrats (0.725), and radical right (0.821). Given these standard deviations are relatively low, it alleviates some concern about weighting the measures by the size of the party or other characteristics that might make one party’s position more important than another since their positions are relatively similar so it would not radically change the averages.
Importantly, these clusters demonstrate that there is variation within party families as not all parties are located directly on the position of the rest of its party family (see also De La Cerda and Gunderson Reference De La Cerda and Gunderson2024). The further a party is from the black line that runs through Figure 1, which represents zero divergence, the further it is from the ideological position of the rest of its party family. The radical left Greek Communist Party serves as the most extreme example of where the party has an ideological position of 8.529, which is the most right-leaning party in the dataset, yet the average position of the rest of the radical left in Western Europe at the time was 3.441, a much more left-leaning position. This divergence within party families is the key variation I use to test my hypotheses.
This measure serves as a useful proxy measure to capture a party’s relationship with its transnational party family. I assume that parties that are ideologically further away from the average position of their transnational allies are unlikely to celebrate those party’s successes abroad, attend party family conferences, and identify themselves as reliable members of the party family. For instance, the decision by Petry and the AfD to appear alongside Le Pen, Wilders, and Salvini in 2017 coincided with a systemic shift in the party’s goals for 2017 to become a more prototypical radical right party, shifting further to the right and embracing a more right-leaning base (Arzheimer and Berning Reference Arzheimer and Berning2019).Footnote 13 The meeting would not have happened if the party had not also shifted its programme to better align with the leaders it was hosting.
To test H1, I first calculate the divergence from the party family by taking the absolute value of the difference between a party’s ideological position and the rest of its party family. Lower values on the divergence scale resemble less ideological divergence, with zero representing perfect alignment between the party and the party family.Footnote 14 I then take this measure of divergence in the current election (t0) and find the difference from the divergence score the party received in the prior election (t-1) to capture how a party has changed its position from one election to the next.Footnote 15 Positive values indicate that the party has moved further away from the ideological position of its party family, while negative values indicate that the party has moved closer to the ideological position of its party family. Values close to and at zero indicate little to no movement. This measure ranges from −1.822 to 3.777 with a mean value of −0.037.
Controls
I include several controls to account for extant explanations for how voters perceive parties. First, I control for conventional demographic variables including age, education, gender, and the voter’s left-right ideological position to account for individual-level attributes that can shape how accurately a voter can evaluate parties since perceiving parties is inherently an individual-level question. Additionally, to account for a respondent’s political knowledge, I use the political information battery included in the CSES, which asks fact-based knowledge questions. Responses are coded as either correct or incorrect. Since only three questions are asked in all waves, I take the count of correct responses to measure political knowledge. Higher values reflect higher political knowledge and should be associated with greater perception accuracy since politically informed voters are likely to be following politics and know where parties stand.
At the voter-party level, to account for how a voter’s relationship with the party shapes how it perceives the party, I first control for whether the individual is a partisan, coded as whether the voter reported the party in each voter-party dyad as the party they are closest to. I expect them to be more likely to misperceive the party if they are a partisan since they may assign their own views to the party since it is how they identify themselves, which may be at odds with where the party stands (Somer-Topcu Reference Somer-Topcu2015).Footnote 16 Additionally, I account for the distance between the voter and the party. I expect voters who are further away from a party to be more likely to inaccurately perceive the position of the party (Bakker et al. Reference Bakker, Jolly and Polk2018; Downs Reference Downs1957).
Then, I include four party-election-level controls. The first is the size of the party measured by the vote share of the party, and I expect the accuracy of a voter’s perception to be lower for larger parties (Somer-Topcu Reference Somer-Topcu2015). The second is the magnitude of the shift in the party’s RILE position from the prior election (t-1) to the current election (t0). This is calculated by finding the difference between a party’s prior position in the last election and their position in the current election and then taking the absolute value of this difference. This captures how much a party shifted between elections, which I expect to be associated with a greater accuracy gap as voters may not have picked up on the shift or the extent of it (Dahlberg Reference Dahlberg2009). I also control for the left-right position of the party, weighted by the party’s size to account for any outsized influence a party’s position may have on how voters perceive the position of the party family averages. I weigh this control by size so that smaller parties exert less influence than larger parties. Then, I account for whether the party is an extreme party by using the distance between a party’s position and the mean position of the party system. More radical parties should be better perceived by voters than mainstream parties that have converged in the middle (Spoon and Klüver Reference Spoon and Klüver2019). Finally, I include the age of the party as measured by the number of years since the party’s founding. Older parties should be more accurately perceived by voters as they have been around longer to cultivate a unique brand. Next, I include fixed effects for party families to account for other traits associated with the various party families that may impact how voters perceive parties in those families (Schleiter et al. Reference Schleiter, Böhmelt, Ezrow and Lehrer2021). Finally, I include an additional control for how the party family has shifted over time to make sure the change I am capturing is due to the party’s change and not the slight change in party family positions from one election to the next.
At the country election level, I control for the effective number of parties in each election. Fewer effective numbers of parties should be associated with greater accuracy in the perceived positions as there are clearer dividing lines between parties.
Since the observations are nested within individuals, parties, elections, and countries, it is necessary to account for this hierarchical structure of the data through a multi-level model. Failure to do so may overrate the significance of the estimated effects by deflating standard errors and inflating the type I error rate (Steenbergen and Jones Reference Steenbergen and Jones2002, 219–220). As the outcome variable is continuous, I thus estimate a multi-level linear regression with random effects at the party level and fixed effects at the country and year level. It is necessary to use fixed effects at the country and year level as I do not have enough countries or years to provide unbiased estimates in a hierarchical model (Stegmueller Reference Stegmueller2013). Tables A1 and A2 in the appendix report summary statistics for all variables for the whole sample and the estimated sample, respectively.
Results
Table 1 presents the results of multi-level models estimating the accuracy of a voter’s perceptions of a party’s positions. There is strong support for H1. As a party moves further away from its party family, the accuracy gap between its actual positions and voters’ perceptions of its positions increases.Footnote 17 Inversely, as a party embraces the positions of its party family, the accuracy gap shrinks.Footnote 18 Figure 2 highlights the positive relationship between a party’s shift to or away from the party family’s position and the size of a voter’s accuracy of their perceptions of the parties’ perceptions. On the x-axis is the change in divergence from the party family and the y-axis is the accuracy gap about the positions of the party. The distribution of change in party family divergence scores is presented in the rug plot along the x-axis. The solid line represents the estimated accuracy gap as divergence from the party family increases, while the dashed lines represent 95% confidence intervals. As the graph shows, there is a clear positive relationship such that parties that shifted closer to the party family were more accurately placed than parties that shifted further away from the party family.Footnote 19 When the party moves two points closer to the position of its party family from the last election, the average accuracy gap is 1.82 points. When there is no change in the party’s position relative to its transnational party family, the average gap is 2.13 points. Alternatively, a two-point shift away from the position of the party family results in a 2.44-point accuracy gap between the party’s positions.Footnote 20 This captures a 0.62 point difference between a large shift towards the associated positions of the party family and a large shift away from the party family image.Footnote 21
Estimated Effect of Party Family Divergence (CSES)
Estimated Perception Accuracy Gap by Change in Party Family Divergence.
Regarding the controls, interesting findings emerge across all models. Women are slightly more likely to misperceive the positions of parties than men; however, the substantive effect is quite small at around 0.09 ideological points. As one gets older, their misperceptions of parties increase. Considering political sophistication, both education level and political information are statistically significant. The more educated the individual, the more accurate their perceptions. Similarly, the more politically informed are more accurate in their perceptions.Footnote 22 Next, more right-leaning individuals hold more accurate perceptions. At the individual-party level, the further a voter is from the party, the further off her perceptions. Lastly, partisans hold less accurate perceptions of their parties than nonpartisans. These two findings in tandem suggest that voters may hold higher misperceptions about parties they are ideologically close to due to partisan attachments that cloud their perceptions of their own party, making the voters think the party is closer to them than it may actually be. At the party-election level, the accuracy gap increases as the size of the party increases, which follows from research that larger parties will often adopt broad appeal strategies that blur their positions to attract a larger voting bloc (Somer-Topcu Reference Somer-Topcu2015). Next, larger shifts in a party’s position from the last election increase the accuracy gap, as expected, suggesting that large changes in a party’s position hurt a party’s ability to hold distinguishable positions (Dahlberg Reference Dahlberg2009). Lastly, the older the party, the greater the accuracy gap.
Evidence from Germany
The results presented above provide compelling evidence that the positions associated with the party family matter. The further a party moves from the party family’s positions, the more inaccurate voters become about its positions. However, while the CSES provides a good snapshot across countries and time, the inability to look within individuals limits its ability to establish a clear causal connection between the divergence of the party’s position and that of its party family and how a voter subsequently perceives the party. To evaluate this causal link, I turn to panel data from the German Longitudinal Election Study (GLES) from 2009 to 2021, which allows me to evaluate how shifts in German party’s positions relative to the rest of their party family over time influence a voter’s perceptions of those parties. In total, 47.58 per cent of respondents participated in more than one of the four waves, allowing for analysis of the evolution of their perceptions of party positions over at least two elections.
The GLES is particularly useful for further testing the causal link between party family divergence and where a voter perceives her parties because they ask respondents to place, on a scale from zero to ten, where they believe German parties fall in each wave of the survey just as the CSES. Respondents are asked to place the Christian Democratic Union (CDU/CSU), Social Democratic Union (SPD), Free Democratic Party (FDP), Alliance 90/Greens, Die Linke and, beginning in 2017, the Alternative for Germany (AfD). Following my analysis using the CSES, I calculate the accuracy of a voter’s perceptions by finding the absolute value of the difference between a voter’s placement and the actual position of the party as identified by the CMP data for my dependent variable. This measure ranges from 0.013 to 7.096 with a mean accuracy score of 1.85.
For my independent variable, I calculate the change in the party family’s position from the previous election (t -1 ) to the focal election (t 0 ) to evaluate if the party is embracing the party family’s positions or distancing itself from the party family. Positive values signal that the divergence between party and party family has grown, while negative values indicate that the divergence has shrunk, symbolizing that the party is moving closer to the position of the party family. This measure ranges from −2.576 (Left from 2005 to 2009) to 0.519 (SPD from 2009 to 2013) with a mean of −0.139, representing a general trend towards party family alignment among German parties.
As controls, at the individual level, I account for standard demographic controls including each respondent’s age, gender, and ideological position, as above, in the CSES analysis. I also include the distance between their position and the party’s, expecting voters further away from the party to be less accurate about its positions. Lastly, I include if they reported the intention to support the party in the upcoming election to account for partisan preferences. At the party level, I account for the party’s size in the election, which should be negatively associated with voter accuracy (Somer-Topcu Reference Somer-Topcu2015). I also control for the size of the change in the party’s positions from one election to the next, which should also be negatively associated with accuracy. At the party level, I control for the governing status of the party which should be negatively associated with accuracy (Klüver and Spoon Reference Klüver and Spoon2020). I also control for the shift in the party family’s position to account for changes in divergence not a result of the party.Footnote 23 Most importantly, to account for the temporal dependencies in both the independent variable and dependent variable, I include both a lagged DV and a lagged IV. Doing so allows me to evaluate how perceptions of the party’s positions are changing from one election to the next, independent of how accurate voters were in the past or how far apart the party was from the party family in the previous electoral cycle.Footnote 24 Summary statistics are available in Table A11. Since individuals are nested within parties and elections, I run a hierarchical model with both election and party fixed effects given the small number of groups, three and six respectively, for each and random effects at the individual level to account for heterogeneity between individuals.
Results
Table 2 reports the results of my analyses of voter perceptions in the GLES panel.Footnote 25 According to Table 2, the results from the GLES panel provide additional confidence in the general trends observed cross-nationally from the CSES.Footnote 26 For H1, there is a statistically significant positive effect of the shift in the divergence from the rest of the party family. A one-unit increase in the distance between a party and its party family is associated with an increase of 0.6 points in the accuracy gap of voters’ perceptions. Inversely, if a party moves closer to its party family’s associated positions between elections by one point, they are perceived more accurately by 0.6 points. When a party moves one point closer (a value of −1) to the associated positions of the party family, voters are roughly 1.17 points off from the party’s true positions on average; however, when a party moves one point further away (a value of 1), voters are inaccurate in their assessments of the party’s positions by about 2.36 points on average.Footnote 27
Estimated Effect of Party Family Divergence in Germany (GLES)
The FDP serves as an illustrative example of these trends. Prior to the 2013 federal elections, the FDP was the junior partner alongside the CDU/CSU, greatly hurting its ability to distinguish itself again and contributing to a disastrous cycle in the 2013 elections when the party lost nearly ten per cent of its votes from 2009 and lost all of its seats in the Bundestag (Klüver and Spoon Reference Klüver and Spoon2020). Between 2009 and 2013, the FDP moved away from the associated positions of the liberal family by roughly 0.26 points and, by 2013, the accuracy gap of its positions had increased by 0.26 points from 1.29 in 2009 to 1.49 in 2013. Desperate to recover its electoral standing and distinguish itself, the party embraced the liberal family in the lead-up to the 2017 elections, moving closer to the associated image of a liberal party by 0.43 points. Fortunately for the FDP, its image recovered and, by 2017, voters had become 0.23 points more accurate in their evaluation of the party, inaccurately placing the party on average by 1.26 points.
Turning briefly to the controls, interesting patterns emerge. Both lagged variables are significant and predict greater inaccuracy. The higher the voter’s inaccuracy in the last election, the greater it will be in the subsequent election. Likewise, the greater the distance between the party and the party family in the last election, the greater the inaccuracy of its positions in the subsequent election. Additionally, larger parties tend to be perceived less accurately. Voters also hold more inaccurate perceptions of governing parties. More right-wing individuals tend to be more inaccurate than left-wing respondents. The further away a party is from a voter’s ideological position, the more inaccurate they are about its positions. Lastly, respondents are more accurate about the positions of the parties they intend to vote for.
These results from the GLES panel provide further confidence in the cross-national findings from the CSES. By looking within-individual using the GLES panel data, these results demonstrate a significant causal relationship between a party’s decision to align itself with its party family and the accuracy of voter perceptions about its positions. Not only is there a strong cross-national relationship between voter accuracy and a party’s position relative to its family, but these findings from Germany demonstrate a strong causal connection as well.Footnote 28
Conclusion
In summary, voters hold more inaccurate views of a party’s left-right position when it becomes ideologically divergent from the rest of its party family. Should a party choose to align itself with its family and adopt positions that converge with those held by the rest of the family, the more accurate voters become about the party’s positions. These conclusions are supported by a cross-national analysis from the CSES from 1999 to 2015 and with panel data from the GLES from 2009 to 2021. Through both analyses, these findings establish a strong association between the party’s divergence from its party family and how accurately it is perceived by voters, as well as evidence of a causal connection. These findings highlight the strategic calculus for parties as embracing the party family by adopting its ideological profile helps clarify the positions of the party for voters. These results demonstrate the utility of the transnational family brand as an informational cue for voters that they can use when evaluating the positions of parties and as a strategy for parties to strengthen their brand.
While the question certainly remains over how voters learn about foreign parties, this study provides evidence that voters use the information they do have about parties abroad, namely a party’s party family label, to make judgements about their domestic electoral arena. With voters increasingly exposed to parties from other countries due to the spread of transnational information networks and supranational institutions that force parties to group together by ideological similarity, these findings are a start at unpacking the role this additional source of political information plays in a voter’s decision-making. If voters are not paying attention to the information they get from the party family linked to a party and the associations voters hold about that family label, whether that information reinforces the party’s positions or contradicts them would not matter for the accuracy of voter perceptions. By finding that parties send reinforcing signals about their positions by aligning themselves with brands voters know about, such as the party family, this article establishes a clear pattern that voters are paying attention on some level to transnational party families and what these cross-national labels mean for parties tied to them.Footnote 29 The use of informal conceptual categorizations as opposed to formal groupings such as the EPGs gives further confidence in these results. The informality of these groups presents a harder test as voters must make the connections between parties and transnational families themselves to use the information provided by the transnational brand to evaluate parties.
For parties, these results suggest that it can be a viable strategic decision to align itself with their transnational party family. Parties have a strategic incentive to distinguish themselves from their competition as this better allows them to appeal to distinct portions of the electorate (Downs Reference Downs1957). Parties that fail to distinguish themselves risk losing their ability to attract voters (Spoon and Klüver Reference Spoon and Klüver2019). Additionally, how a party is perceived matters and can have meaningful consequences for its electoral fate (Adams et al. Reference Adams, Bernardi, Ezrow and Somer-Topcu2023). The findings from this study expand the toolkit available to parties to distinguish themselves from their alternatives by leaning on their party family connections and credibly committing themselves to championing party family goals (Fernandez-Vazquez and Somer-Topcu Reference Fernandez-Vazquez and Somer-Topcu2019; Nonnemacher and Spoon Reference Nonnemacher and Spoon2023; Sagarzazu and Klüver Reference Sagarzazu and Klüver2017; Somer-Topcu et al. Reference Somer-Topcu, Tavits and Baumann2020). The associations that voters hold about a party family brand serve as a useful information heuristic for voters and can strengthen a party’s political brand by being seen as a credible champion of broader ideological fights. This finding provides more evidence that associations matter, and the alliances a party chooses to emphasize, foreign or domestic, shape how it is perceived (Adams et al. Reference Adams, Weschle and Wlezien2021; Williams and Whitten Reference Williams and Whitten2015).
While this article examines whether embracing the party family label is a viable strategy for parties to distinguish themselves, it is important to note that the utility likely varies by party family.Footnote 30 Importantly, this article establishes that it is generally always a good strategy to embrace the party family a party is associated with. However, this may not be the case for all party families, especially pariah party families such as the radical right and left, which carry meaningful negative associations in the minds of voters that could damage a party’s ability to appeal to voters (Mendes and Dennison Reference Mendes and Dennison2021; van Spanje and Van Der Brug Reference van Spanje and Van Der Brug2007). Similarly, centrist party families like the liberal party family and the centre party family, which are left-leaning on some issues, typically socio-cultural, and right-leaning on others, typically economic, and rely more on valence attributes to get elected than other party families (Zur Reference Zur2017, Reference Zur2021), may place less emphasis on their ideological alignment with the family than on common reputational attributes. Finally, green parties may only wish to highlight their alignment with other party families on the green issue given the more policy-oriented focus of green elites and voters (Adams et al. Reference Adams, Clark, Ezrow and Glasgow2006). Future work should aim to unpack what kinds of factors may explain when a party chooses to embrace a party family label as opposed to creating distance from the family and how this varies by party family.
This work is an important first step into how the relationship between parties and their party families influences voter behaviour and how voters engage with transnational political environments, but future work is needed. Importantly, this project examines the general strategy to embrace the family through how it adjusts its positions but there are many different approaches parties can take to do so, including changing its rhetorical style (Rydgren Reference Rydgren2005), changing its colours and logos, embracing new leaders (Söderlund and Grönlund Reference Söderlund and Grönlund2024), and attending rallies with prominent international actors. Future work must be done to evaluate the effectiveness of each of these strategies to better understand the optimal way for parties to leverage their transnational brands. Experiments that manipulate exposure to information about a party engaging in one of the aforementioned strategies or other quasi-experimental methods can more directly estimate the effect of these more precise instances of a party embracing its allies. Furthermore, more research is needed to see how the party family label shapes evaluations of other aspects of a party’s brand, such as a party’s reputation, legitimacy, and overall image. Understanding how voters utilize the party family brand to evaluate their own party provides us with an important foundation for future inquiries into an increasingly transnational dimension of party politics, and helps us unpack how voters, parties, and transnational brands interact with each other.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S000712342400070X.
Data availability statement
Replication data for this paper can be found at https://doi.org/10.7910/DVN/1FRQWC.
Acknowledgements
I would like to thank Jae-Jae Spoon, Steve Finkel, Jude Hays, and Pablo Fernandez-Vazquez for their help and feedback throughout this project. I would also like to thank Sheri Berman, Endre Borbáth, Ruth Dassonville, Lawrence Ezrow, Patrick Fournier, Noam Gidron, Valentina González-Rostani, Jacob Gunderson, Mareike Kleine, Jelle Koedam, Nina Obermeier, Lukas Obholzer, Jonathan Polk, Zeynep Somer-Topcu, António Valentim, the anonymous reviewers, and the editor for their helpful feedback. All mistakes remain my own.
Financial support
None.
Competing interests
None.
Open access
