The Group Appeal Strategy: Beyond the Policy Perspective on Party Electoral Success

Abstract

Political parties use policy appeals to change their policy images and increase electoral support. Building on the idea that parties’ group images also matter to voters, this article shows that group appeals can benefit parties as well. Combining manifesto data on policy and group appeals covering 50 years, we revisit the shift from class to catch-all politics in Britain and present three findings. First, a vote share analysis shows that the Labour Party benefited from using not just policy but also group appeals to downplay its class image. Second, consistent with a catch-all strategy, survey evidence shows that this boost in Labour support was most prominent outside the traditional base. Third, while group appeals had independent effects, we show how group and policy appeals also interact and reinforce each other. This suggests that non-policy strategies matter to parties’ electoral success on their own but also in combination with policy strategies.

Which party strategies are successful in electoral politics? This question occupies political scientists, media experts, and party elites alike. Comparative political research has often addressed it from a policy positioning perspective (Downs, 1957), assuming that election strategies are about supplying policies that match the policy demands of the electorate (Abou-Chadi and Wagner, 2019; Adams and Somer-Topcu, 2009; Karreth et al., 2013; Kitschelt, 1994; Meguid, 2008; Tavits, 2007). According to this view, the winning strategy for vote-seeking parties is to adjust their policy images in ways that fit what voters want.

However, while there is no reason to doubt that parties use policy appeals to target voters, there is good reason to doubt it is all they do. Since voters are driven by more than just policy preferences, parties have incentives to do more than state policies. Hence, a growing body of research is moving beyond the policy perspective. For example, Somer-Topcu (2015) shows how the choice of party leader affects vote shares, while Jung and Tavits (2021) consider the electoral benefits of statements about parties’ competence and integrity. This article focuses on another non-policy strategy that has recently emerged on the agenda of party research: the use of group appeals (Dolinsky, 2023; Horn et al., 2021; Howe et al., 2022; Huber, 2022; Stückelberger and Tresch, 2022; Thau, 2019).

The group appeal perspective builds on the extensive literature about opinion formation and voting showing that social group belonging is central to political behavior (e.g. Achen and Bartels, 2016; Bornschier et al., 2021; Butler and Stokes, 1974; Campbell et al., 1960; Dalton, 2014; Kinder and Kam, 2009). Voters judge parties on which policies they are thought to stand for, but they also judge parties on their group images (i.e. which groups they are thought to represent), motivating parties to pursue strategies that are explicitly group centric (Dickson and Scheve, 2006). However, while recent studies have considered either the factors shaping parties’ group appeals (Huber, 2022; Stückelberger and Tresch, 2022; Thau, 2019) or the effects of group appeals on individuals’ vote choice (Evans and Tilley, 2017; Robison et al., 2021; Thau, 2021), our understanding of how the group appeal strategy helps parties overall, and among key target groups, is still lacking.

This article provides an example of how group appeals can shape party electoral success using the transition from class to catch-all politics in Britain as its case. The British case is interesting for two reasons. First, the class voting literature, particularly Evans and Tilley (2017), has already shown how class-related group appeals affect class divides in voting in Britain (see also Thau, 2021). Although it has not been directly tested whether Labour profited from changing class image, these findings certainly indicate that group appeals have a potential to increase vote shares and support among particular groups. Second, the literature on party strategies routinely uses the British case as the prime example of how parties can benefit from changing their policy image (Adams and Somer-Topcu, 2009; Karreth et al., 2013; Mair et al., 2004). Faced with a declining working class and trade union member constituency, the Labour Party moderated its economic policies to pursue a catch-all strategy that proved successful, particularly during the 1990s under Tony Blair.

Combining a recent dataset on group appeals with the widely used Manifesto Project Dataset on policy appeals over a 50-year period (1964–2015), we show three ways that group appeals mattered to party electoral success in the British case. First, an analysis of vote shares obtained in elections suggests that Labour gains votes not just from centrist policies but also from downplaying its symbolic working-class ties. Second, consistent with the Labour Party’s notorious catch-all strategy, survey data on voters’ party preferences over the same period show that this change in group image by Labour is particularly effective outside its traditional base (i.e. people who are neither working class, nor trade union members). Finally, exploring the catch-all strategy further, the survey evidence shows that Labour’s efforts to change class image and attract non-base voters are most successful when group and policy appeals are combined in a compatible way (i.e. when symbolic class ties are downplayed and centrist economic policy positions are adopted). Considering such interactions is a novel perspective in a literature that has otherwise tried to isolate their respective effects.

This article thus contributes to a more nuanced understanding of the transition from class to catch-all politics in Britain. While prior research has shown how Labour’s changing class image shaped class voting (Evans and Tilley, 2017), we illustrate that this group appeal strategy also mattered to Labour’s overall success. This speaks to ongoing debates about which strategies social democratic parties in Western Europe might adopt to stay electorally relevant (e.g. Abou-Chadi and Wagner, 2019), while also adding, in broader terms, to the emerging research agenda that considers party strategies beyond policy statements, especially recent studies on the causes and effects of group appeals (Huber, 2022; Robison et al., 2021).

Party Electoral Strategy: Policy Image and Group Image

Comparative politics scholars have long studied the strategies that parties employ to generate support, with the policy positioning (Adams and Somer-Topcu, 2009; Downs, 1957) and issue emphasis (Budge and Farlie, 1983; Petrocik, 1996) perspectives being the two main approaches. While both perspectives provide key insights about party strategy, our point of departure in this article is the policy positioning perspective because studies here have been particularly occupied with the electoral effects of strategies, whereas studies within the issue emphasis approach have paid more attention to party competition and agenda-setting (e.g. Green-Pedersen, 2007). ¹ The dominant assumption in the policy positioning perspective is, as Kitschelt (2011: 620) puts it, that policy statements are the “currency of competition”: Political parties supply policies to generate support; voters support parties based on the positions they take. Research analyzing vote gains and losses from policy shifts illustrates this logic well (e.g. Adams and Somer-Topcu, 2009; Karreth et al., 2013; Tavits, 2007).

There is good reason to look beyond policy statements, however. We know that people consider other things than policy cues when voting. Thus, assuming that electoral strategies are tailored to “fit” voter decision-making, parties should employ a range of tools to connect with voters. This article focuses on group appeals as one important alternative to policy appeals (Dolinsky, 2023; Huber, 2022; Thau, 2019). The group appeal perspective builds on the widely held notion that voters rely on group memberships, collective identities, and social categorization when making political decisions (Bartolini and Mair, 1990; Campbell et al., 1960; Goldberg, 2020; Heath, 2015; Kinder and Kam, 2009; Lipset and Rokkan, 1967; Vivyan et al., 2020). In politics, group-centric reasoning remains “an inescapable fact of life” (Popkin, 1994: 218) because “people are naturally group-oriented” (Achen and Bartels, 2016: 215).

This group orientation in politics is perhaps best expressed in how voters think about political parties. As noted by Campbell et al. (1960), voters often distinguish between political parties based on which social groups they are thought to represent. When asked, people generally say that leftist parties represent groups like workers or racial minorities while rightist parties do not (Ahler and Sood, 2018; Nicholson and Segura, 2012), and along with policy stances, such group ties are among the primary reasons voters themselves give for party choice (Dalton, 2014: 31). Election-oriented parties exploit this and try to shape their images in voters’ eyes. One strategy is to change policy images, that is, the public perceptions about the policies a party offers. But another is to change group images, that is, the public perceptions about which groups a party represents.

Policy Appeals, Group Appeals, and Electoral Consequences

Comparative politics scholars generally agree on how parties try to change policy images: They state policy achievements or promises in election manifestos and elsewhere (e.g. Abou-Chadi and Wagner, 2019; Adams and Somer-Topcu, 2009; Meguid, 2008; Tavits, 2007). Although the literature rarely discusses how best to understand such policy appeals at the statement level, Dolezal et al. (2014: 64) usefully define them as a party stating that it (or another party) is for or against some specific policy such as tax cuts, environmental regulation, or immigration reform. This understanding not only squares well with how party policy is already measured in the widely used Manifesto Project Dataset (Volkens et al., 2021) but has also informed recent work on group appeals, which argues that parties use group appeals to manage group images in much the same way as they manage policy images using policy appeals (e.g. Huber, 2022; Stückelberger and Tresch, 2022; Thau, 2019).

Following Thau (2019: 65), group appeals can be defined as a party associating or dissociating itself (or another party) with a specific group category such as workers, young people, or women. The difference between group and policy appeals then lies in whether their content concerns groups or policies. Although this may seem like a marginal difference, research consistently finds that group cues about social class, race, nationality, religion, and gender can affect voter decision-making (see Huddy, 2013).

The literature is not clear on how precisely group and policy appeals work, especially not in relation to each other. This is partly because the use of group appeals has often been seen as the dependent variable (Dolinsky, 2023; Huber, 2022; Stückelberger and Tresch, 2022; Thau, 2019) and partly because the goal of studies on effects has mainly been to isolate the impact of group appeals from that of policy appeals (Thau, 2021). That said, there is an acknowledgment in the literature that group and policy appeals can interact. For example, for the British case discussed in the following section, Evans and Tilley (2017: 13) argue that it was the combined change “in policy and class image terms” that shaped individual-level class differences in voting. Likewise, focusing on Austria, Howe et al. (2022: 835) conclude that the success of nationalist parties during the early twentieth century was predicated on a combined use of “both group identity appeals and policy offers.” This interplay of group and policy appeals remains undertheorized, however.

We argue that parties make appeals along a symbolic group dimension and a substantive policy dimension to strategically target the matching distribution of group sympathies and policy preferences in the electorate. ² Since voters and groups want both (substantive) policy and (symbolic) social representation in politics (Heath, 2015; Huddy, 2013), it matters whether parties signal their commitment to a given voter group via policy appeals, group appeals, or both. The latter sends the strongest signal of commitment, we argue. Take the working class as an example: A social democratic party that both adopts a left-wing position on redistributive policy and emphasizes its symbolic ties to workers targets the “hard” policy interests as well as “soft” identity concerns of the working class and thus spurs higher support among this group than if commitment is only signaled on one of the dimensions. This assumes that group and policy appeals speak to distinct aspects in voters’ party perceptions: the former to a party’s group image and the latter to its policy image. Although there could obviously be a degree of overlap—that is, voters may infer some mix of policy stance and group representation from each type of appeal—Robison et al. (2021) have indeed found that symbolic group appeals primarily influence group images (rather than policy images).

While sending a strong signal is an effective way to change sometimes “sticky” party images (Adams, 2012), it is not necessarily wise to change group and policy images in a compatible way. As political parties often need to build support across multiple voter groups, especially mainstream ones seeking to vote-maximize (Mair et al., 2004), there are many situations where the electoral incentives drive parties to diversify their appeals. Huber (2022) has shown how this coalition-building logic guides the way that parties pick the targets in their group appeals and thus the relative emphasis put on different group categories (e.g. unemployed, pensioners, women). Here, we emphasize that the same logic also guides the combination group and policy appeals. Dickson and Scheve (2006) make a similar argument in their game-theoretic model, suggesting that political elites can change policies to appeal to new voters while maintaining support from their base through their group appeals. This casts the use of group appeals as a defensive strategy that keeps the core “on board” (Dickson and Scheve, 2006: 7), whereas policy appeals are used to chase new voter segments. However, we see no a priori reason why parties could not do the opposite as well (i.e. stick with their policies while attempting to change their group image). The main point is that parties can send either a strong signal with group and policy appeals targeting the same voter segment or a mixed signal with group appeals targeting one segment and policy appeals targeting another. In the latter case, the electoral effect is expectedly weaker within each voter group, but since support accumulates across groups and adds to a party’s overall vote, this mixed signal strategy is sometimes preferable.

The choice between the narrow but stronger signal and the broad but weaker signal depends on multiple factors (Dickson and Scheve, 2006). Perhaps the most important one is the size of various voter groups, as this ultimately decides their electoral value (Best, 2011; Mair et al., 2004). Indeed, as discussed in the following sections, in the case of class politics, the primary driver of party strategy has undoubtedly been the shrinking working class and the growing middle class (Evans and Tilley, 2017), which motivated social democratic parties in particular to attempt to expand their catchment areas.

This discussion leads us to explore three ideas. First, we examine whether group appeals can benefit parties independently of policy appeals, as one would expect if they target parties’ group images and attract voters based on their symbolic group concerns rather than policy preferences. Second, we examine whether such group appeals are especially effective among the target groups that a party is particularly interested in. This seems to be a reasonable expectation if group appeals are indeed one of the tools that parties use to selectively target various segments of the electorate when building its support base. Third, we explore how group and policy appeals interact. Given our class politics case—where social democratic parties had unambiguous incentives to expand catchment areas beyond the working class—we test if the strong signal strategy, with group and policy appeals used to target a particular voter segment, has the reinforcing effects that one would expect.

Revisiting the Move to Catch-All Strategies in Britain

Britain is a useful place to examine the electoral benefits of group appeals for two reasons. First, compared to proportional multi-party systems, its majority two-and-a-half party system simplifies the party competitive setting. While this raises a generalizability question that we return to in the conclusion, it also allows for a more straightforward empirical analysis. Second, and more importantly, Britain stands as the single best case in support of the policy positioning perspective, which dominates the literature. While comparative evidence that policy shifts can bring vote gains has been inconsistent (Adams, 2012), the Labour Party in Britain experienced substantial vote gains by adopting more moderate images in terms of economic policies and left-right ideological positions (Adams and Somer-Topcu, 2009; Karreth et al., 2013; Kitschelt, 1994; Mair et al., 2004). As the number of workers and trade union members declined during the second half of the twentieth century, Labour pursued a “catch-all policy moderation strategy” targeting the middle classes, which proved successful—at least in the short run (Karreth et al., 2013: 803).

Yet, as argued, adjusting policy images is only one way that parties appeal to voters. Indeed, as Evans and Tilley (2017) demonstrate, since the 1960s, Labour has also worked to downplay its more symbolic working-class ties whether measured in terms of its MPs class background, media coverage, party leaders’ speeches, or election manifestos: The catch-all strategy pursued by Labour clearly entailed a change in both policy and group image. However, while several studies have focused on how parties’ class images can shape class voting, that is, the difference in voting between class groups (Evans and Tilley 2017; Heath, 2015; Thau, 2021), their consequences for parties’ overall success have not been directly studied. We expect that group appeals were key in helping the Labour Party increase its vote share by expanding its traditional catchment area.

What about the other major competitor in Britain? There are contradictory arguments as to whether the Conservative Party would benefit from attempting to adjust its class image. On the one hand, comparative scholars often argue that leftist and rightist parties dealt with individualization and modernization processes in parallel ways. According to Dalton (2014: 164), for example, there has been a “general convergence” of the parties, where “socialist parties vied for the votes of the new middle class, and conservative parties sought votes from the working class.” Green (2007) highlights this convergence for the British case. On the other hand, in a class perspective, the conservative base—the middle class, self-employed, and other upscale groups—has not contracted but expanded, providing little incentive to make long-term changes to either the policy or group image. In any case, to the extent that the Conservative Party did use group appeals to build a more inclusive class image, we can hardly expect any electoral effects to match those for Labour, whose catch-all strategy proved unusually successful in comparative perspective (Mair et al., 2004).

The analysis that follows first shows the aggregate developments of class-related group appeals in Britain and uses election results data to explore whether group appeals are associated with increases in vote shares for the two main parties. Then, tracing this electoral effect in more detail, individual-level survey data are used to examine whether group appeals are particularly effective for the Labour Party in increasing support outside its traditional base. Finally, we analyze whether Labour’s popularity among outgroup voters is most pronounced in situations where policy and group appeals are combined to signal a particularly clear change in class image.

Data and Measures

The empirical analysis is based on combined data on group appeals, policy appeals, aggregate-level vote shares, and individual-level voting over a 50-year period in Britain. The data on parties’ electoral appeals come from existing content coding of all the Labour and Conservative election manifestos between 1964 and 2015. Our analysis thus focuses on the two major parties, as they alone have competed for office during the long-term period studied. Furthermore, the focus on election manifestos means that our results speak directly to party research on policy positions using these same manifesto data.

While data on policy appeals can be taken from the Manifesto Project Dataset (Volkens et al., 2021), the data on group appeals are taken from the dataset introduced in the study by Thau (2019). Parallel to the policy appeals data, this group appeals data come from sentence-by-sentence coding of election manifestos and measure (at the party-year level) how much Labour and the Conservatives emphasized positive or negative relations to 97 group categories between 1964 and 2015. These group emphasis variables are based on around 10,000 individual group appeals, which were coded according to subject (party), object (group), and subject-object relation (whether party and group were associated or dissociated). ³ Since cases where parties either dissociate themselves from groups or talk about other parties’ group ties are extremely rare in manifestos (and varies across group categories), we decided to base the indicators used in this article on cases where parties associate themselves with given groups. Our analysis of class politics focuses on class emphasis using three particular group categories: trade union members, workers, and businesses.

Group Emphasis Measures: Class Emphasis

In the British case, group appeals were used by Labour, and possibly the Conservatives, to moderate class-stereotypical group images and broaden electoral catchment areas (Evans and Tilley, 2017). Trade union members, workers, and business owners were important group categories in this process because they were at the core of British parties’ class images (Butler and Stokes, 1974). Labour was historically seen as representing the working class and the trade union movement—often in direct contrast to business owners and capitalists. The group image of the Conservatives has always been fuzzier, as they needed to build coalitions across many groups to reach high vote shares (Denver et al., 2012). In the eyes of the public, the Conservative Party was first and foremost seen as not representing workers and trade union members. Among the groups that they were seen to represent, business owners, self-employed, and capitalist were prevalent (Butler and Stokes, 1974; Dalton, 2014; Evans and Tilley, 2017). ⁴

We capture attempts to moderate class images with two variables that pit either trade union member or worker emphasis against business emphasis. This offers a positional measure tapping how close a party ties itself to opposing groups. For union-business emphasis, we thus subtract the business emphasis percentage in a given party-year from the trade union emphasis percentage in that party-year. If union and business ties are emphasized equally, the value is zero. If union emphasis outweighs business emphasis, the value is negative, and if emphasis is vice versa, the value is positive. The same coding is used for worker-business emphasis. We thus get two bipolar measures that resemble the bipolar and positional way through which policy positions are typically measured (e.g. left-right redistributive position). ⁵

We interpret the class emphasis variables differently for Labour and the Conservatives because a catch-all strategy that downplays traditional class images requires opposite behaviors from the two parties. Labour needed to downplay its positive ties to the working-class, trade union member base and negative ties to businesses, in order not to alienate the middle class (Przeworski and Sprague, 1986). Hence, an increase in the class emphasis variables indicates a more moderate group image. For the Conservative Party, however, the same catch-all strategy would require changing its anti-working class, pro-business image. A more moderate group image is therefore indicated by a decrease in the class emphasis variables.

Policy Position Measures: Redistributive Policies and Left-Right Positions

Previous studies have shown that the catch-all strategies pursued most evidently by Labour, but also the Conservatives, involved the use of policy appeals to shed left-wing or right-wing policy images in favor of centrist positions (Denver et al., 2012; Evans and Tilley, 2017; Green, 2007; Karreth et al., 2013). In line with Evans and Tilley (2017), we capture class-related policy appeals in Britain using the Manifesto Project Dataset, which codes policy statements into a range of policy categories and offers aggregate-level variables that measure the emphasis put on each policy. We use two different indicators to tap class-related policy positions. First and foremost, we use an indicator of redistributive policy suggested by Bakker and Hobolt (2013), which subtracts eight capitalist, pro-market categories from 12 socialist, pro-state categories to estimate policy positions. Second, we also use general left-right positions (RiLe), which result from subtracting 13 so-called left categories from 13 right ones (Laver and Budge, 1992). While redistributive policy seems the best operationalization of “catch-all policy moderation” strategies in the class politics context, we include RiLe because (1) it taps ideological party positions more generally, (2) its inclusion offers the hardest test, and (3) it has been used in prior studies that we want to speak to (Adams and Somer-Topcu, 2009; Evans and Tilley, 2012; Karreth et al., 2013). The more positive the value on redistributive policy and left-right position, the more right-wing the parties are in terms of class-related policies.

As with class emphasis, we interpret changes in the policy variables differently for the two parties. For Labour, a more right-wing position (increase) indicates a more moderate policy image because it breaks with stereotypical class ties, while for the Conservatives, the same policy moderation is indicated by a more left-wing position (decrease).

Trends and Associations: Class Emphasis in the British Case

Before considering any electoral consequences, we show how class emphasis was actually used in the British case. While other studies have considered class appeals in Britain by looking at the percentage of group appeals targeting workers, for example, the bipolar and positional perspective on class emphasis introduced here has not been covered (e.g. appealing to businesses at the expense of workers).

Figure 1 shows the development in class emphasis for both parties between 1964 and 2015. In the left-side plots, we see that Labour has clearly attempted to change its class image over the years, particularly during the early 1990s. This supports the popular view that party leader Tony Blair championed Labour’s catch-all strategy in 1997 but also shows that these efforts started already under Neil Kinnock. In fact, for group appeals, the 1992 Kinnock campaign marks the most pronounced shift. Furthermore, the February 1974 election, where Harold Wilson ran in the context of a miners’ strike, stands out as especially class-centric for Labour.

Figure 1.

Trends in Class Emphasis for Labour and the Conservatives.

The right-side plots show that the Conservative Party has generally been less extreme in class emphasis than Labour. The most notable shift occurred under party leader Margaret Thatcher in 1979, where business ties were clearly emphasized over union member and worker ties in a clearer way than previous years. This shift seems to mark a persistent reorientation in the class image that the Conservatives choose to present. Unlike Labour, in recent decades, the Conservatives appear more willing to embrace their traditional group image as the representative of upscale classes (e.g. self-employed, business owners, high-level professionals and managers). This asymmetry contradicts generalized statements about how mainstream parties across the left-right spectrum have moderated their class images (Dalton, 2014) and, rather, fits how economic restructuring has reshaped the electoral incentives of leftist parties in particular (Best, 2011).

These trends notwithstanding, Figure 1 also shows a good deal of short-term variation in class emphasis for both parties. From a party strategic perspective, this is important because it suggests that class emphasis is used in relation to individual elections and their particular circumstances. The election-to-election adjustments are most notable for Labour. Unsurprisingly, the class emphasis variables are strongly related to each other, but they are also related to the two policy variables (see Tables A1–A3 of the Supplementary Information for correlations; Tables A1–A17 and Figures A1 and A2 are found in the Supplementary Information in the article’s online appendix). Thus, in the case of British class politics, the strategies of two main parties have generally been to try to change group and policy images in tandem, although the two appeal types are not perfectly related. ⁶ For the analysis, this means that British politics provide a hard test of group appeal effects as much of their variation is already accounted for by the policy appeals.

The Electoral Benefits: How Vote Gains Relate to Class Emphasis

Did the two major British parties benefit from a catch-all strategy contingent on class-related group appeals? That is, did it pay off to moderate group as well as policy images? We explore this at the aggregate level using election results data on Britain from 1964 to 2015 (available from the Manifesto Project Dataset). Although the limited number of observations in the time-series (2 parties × 14 elections = 28 party-year observations) means that an aggregate analysis will be suggestive rather than conclusive, we believe it is informative to see if parties’ actual vote shares are associated with their use of group appeals. Like most studies on the electoral consequences of party strategies, we measure the dependent variable as changes in parties’ vote shares from election t−1 to election t (Adams and Somer-Topcu, 2009; Tavits, 2007). In other words, the outcome is whether parties do better or worse than the last election. Positive values on vote change indicate vote gains, while negative values indicate vote losses. Beyond being theoretically interesting, this measure has the methodological advantage of not trending over time, thereby minimizing a key source of potential bias (see Table A1 of the Supplementary Information for correlations with time).

Using an Ordinary Least Squares (OLS) specification with election-clustered standard errors (Hsiao, 2003), our baseline models include either union-business emphasis or worker-business emphasis, a party dummy (conservative), and the interactions between class emphasis and party. We then extend these two models in gradual steps to maximize transparency and accommodate any small-N robustness concerns. First, three potential confounders highlighted in the literature are added (e.g., Abou-Chadi and Wagner, 2019): governing status (in government), gross domestic product per capita growth (GDP growth), and a lagged dependent variable (vote change (t−1)). Second, excluding these confounders, we add interactions between the party dummy and both the policy variables to estimate the class emphasis—vote change association net of policy effects. Finally, the full models are specified, including the three confounders as well as the party-wise policy interactions. The full set of results suggests that the following conclusions are robust across the varying model specifications (Table A4 of the Supplementary Information). ⁷ In Table 1, we focus on the baseline and fully specified models. Models 1 and 2 show naive effects without any controls, while models 3 and 4 control for policy effects and the confounders. Note that since the party dummy (conservative party) is included with Labour coded as 0, the estimates for union-business and worker-business emphasis, respectively, indicate vote changes for the Labour Party.

Table 1.

Regression Models Estimating the Effect of Class Emphasis and Policy Position on Vote Change by Party.

	(1)	(2)	(3)	(4)
Union-business emphasis	0.534*(0.290)		0.656** (0.303)
Worker-business emphasis		0.410 (0.363)		0.374 (0.442)
Redistributive policy			0.844*** (0.220)	0.804** (0.271)
Left-right position			–0.297** (0.126)	–0.205 (0.124)
Conservative party	–1.821 (2.482)	–2.113 (2.516)	–12.93** (4.309)	–13.82** (5.433)
Conservative party × union-business emphasis	0.0197 (0.974)		0.592 (0.847)
Conservative party × worker-business emphasis		0.205 (1.116)		1.117 (1.012)
Conservative party × redistributive policy			–0.577* (0.281)	–0.524* (0.288)
Conservative party × left-right position			0.0691 (0.276)	–0.0606 (0.271)
Vote change (t-1) LDV			0.284* (0.139)	0.224 (0.151)
In government			–7.515*** (1.715)	–7.389*** (1.904)
GDP growth			–0.0480 (0.378)	–0.104 (0.373)
Constant	–0.537 (1.366)	–0.505 (1.526)	16.98*** (4.369)	17.57*** (5.316)
Observations	28	28	28	28
R 2	0.151	0.082	0.633	0.556

Unstandardized OLS estimates with election-clustered standard errors in parentheses. Since 0 = Labour on the party dummy variable, the class emphasis estimates show electoral effects for the Labour Party.

LDV: Lagged Dependent Variable; GDP: Gross Domestic Product; OLS: Ordinary Least Squares

*

p < 0.1, **p < 0.05, ***p < 0.01.

Vote Share Results: Overall Benefits

Overall, it appears that parties can indeed benefit electorally from changing their group image, at least in some circumstances. In model 1, the positive and significant estimate for union-business emphasis shows that Labour does better in elections where ties to union members are downplayed, while business ties are highlighted. In model 2, however, there is no equivalent effect of worker-business emphasis (although the positive sign is as expected). In terms of the size of electoral effects, Labour seems to benefit a good deal from changing its union-business emphasis. Specifically, with the naive model specification, we would predict Labour’s vote share to increase by 0.53 pp, as union-business emphasis changes 1 pp in favor of businesses. However, to explore whether this electoral effect works beyond policy appeals, models 3 and 4 controls for redistributive policy, left-right position, and the potential confounders. In model 3, the positive and significant estimate for union-business emphasis shows that the electoral benefits that Labour reaped from reorienting its class image from trade union members to business owners are actually slightly higher, when controlling for policy positions. For a 1-pp change in union-business emphasis, we would predict a 0.66-pp vote gain for Labour, which amounts to a 2.4-pp vote gain for a 1-SD change (σ = 3.7%). Considering that the mean (absolute) vote change between two consecutive elections during the study period is about 0.92 pp, this effect is substantial. As the insignificant estimate for worker-business emphasis in model 4 shows, however, the effect is still confined to union-business emphasis.

One can see what the picture looks like for the Conservative Party by adding the estimated interactions between class emphasis and party to the effects found for Labour. Yet, to facilitate this, the marginal effects by party are plotted in Figure A2 of the Supplementary Information. The results suggest that the Conservative Party did not benefit from a more inclusive group image. The class emphasis effects are insignificant (p > 0.1) and, if anything, indicate that the Conservatives were best off reinforcing its traditional business ties. Thus, as prior research suggests, leftist parties, like Labour in Britain, seem to be the ones that consistently benefited from moderating class-stereotypical images (Adams and Somer-Topcu, 2009; Evans and Tilley, 2017; Karreth et al., 2013; Kitschelt, 1994). However, our analysis indicates that Labour’s success was tied to a change not just in policy image but also in group image.

The Catch-All Strategy: Labour Support Among Non-Base Voters

Our theoretical argument has observable implications at the individual level, too. Research approaching class politics from the voter side shows that group appeals affect the class gap in electoral behavior (Evans and Tilley, 2017: 16; Thau, 2021). Yet, the party side perspective adopted here is more interested in effectiveness within particular target groups than the between-group differences. Specifically, we expect that subgroup effects should be strongest and most positive among voters outside the traditional base. Raising support among “outgroup” segments is indeed the main purpose of catch-all strategies (Kitschelt, 1994; Mair et al., 2004; Rohrschneider, 2002).

We examine this at the individual level using survey data from the British Election Study (BES; election years 1964–2015) and the British Social Attitudes surveys (BSA; all years 1983–2015) on around 68,000 respondents. ⁸ Given the aggregate-level findings, we focus on the Labour Party, who was actually successful in raising overall support by moderating its class emphasis. To use all the available survey data, our dependent variable is a general indicator of Labour support. The BSA measures this by asking if people see themselves “as a supporter of any one political party” before following up on which party. The BES instead asks if people think of themselves as “Conservative, Labour, Liberal Democrat or what?.” Labour support is coded 1 for Labour and 0 for other parties. We thereby leave out abstention to focus on which party voters prefer, as this is the individual-level process that directly translates into parties’ aggregate vote shares. Party support might not fully capture voting (e.g. party support is likely more stable than vote choice), but a strong correlation in election years between being a Labour supporter and recalled Labour vote suggests that, at least in our data, these two measures tap the same underlying party preference (r = 0.85). ⁹ The sample distribution is 45% Labour supporters and 55% supporters of another party (mainly the Conservatives).

To see whether the group appeals were particularly effective among voters not belonging to the traditional Labour base, voters are divided into three groups based on class position and trade union membership. Considering Labour’s historic ties to the working class and the trade union movement (Evans and Tilley, 2017), we operationalize the base as voters, who are both working class and trade union members. Another group, referred to as the leaners, consists of voters, who are either workers or union members but not both. Finally, the outgroup refers to voters, who are neither working class nor trade union members. Looking across all years, the sample distribution confirms that these three voter groups generally differ in their Labour support: The base has 71% Labour supporters, the leaners group has 55%, and the outgroup has just 31%. ¹⁰

We run four logit models that estimate the probability of supporting the Labour Party. Since our data consist of pooled cross-sections with individuals nested in surveys, a random effects (REs) specification is used (specifically, random intercepts are included in the main models). This helps to deal with the potential autocorrelation originating from the hierarchical data structure, while allowing us to estimate the higher-level class emphasis variables (Wooldridge, 2010). ¹¹ Models 1 and 2 focus on emphasis of trade union members relative to businesses; models 3 and 4 focus on emphasis of workers relative to businesses. The models that focus on the same class emphasis variable differ in whether they control for redistributive policy or left-right position. For example, in addition to union-business emphasis, model 1 includes voters’ class-union background; an interaction between this background variable and class emphasis; controls for GDP growth and governing party status, as in the aggregate-level analysis; and an interaction between class-union background and redistributive policy (including constitutive terms). Model 2 then switches out redistributive policy for left-right position. Models 3 and 4 do the equivalent with a worker-business emphasis focus. Figure 2 shows the average marginal effects of interest, while the full results appear in Table A7 of the Supplementary Information. ¹²

Figure 2.

The Average Marginal Effect of Labour’s Class Emphasis Among the Base, Leaners, and Outgroup.

Subgroup Results: Inside and Outside the Traditional Base

Has Labour’s class emphasis been useful in expanding its catchment area? Figure 2 shows the average marginal effect of the class emphasis variables among voters that are both working class and trade union members (the base), voters that are either working class or trade union members (the leaners), and voters that are neither working class nor trade union members (the outgroup). Controlling for redistributive policy in panel A, and left-right ideological positions in panel B, the estimates show changes in Labour support as class emphasis shifts by 1 pp in favor of businesses. In panel A, we see noticeable subgroup differences. Labour’s de-emphasis of trade union and worker ties was clearly most influential outside the traditional base, that is, among voters that were not trade union members with working-class backgrounds.

Specifically, while we would predict a small and insignificant reaction within the traditional base, Labour support increases 0.5 pp among leaning voters and 0.9 pp among outgroup voters, as union-business emphasis shifts 1 pp toward businesses. The difference in group reactions is equally striking for worker-business emphasis. As Labour shifts emphasis 1 pp away from workers and toward businesses, support increases by 1.1 pp among outgroup voters, and 0.7 pp among leaners, but is small and insignificant within the base. These differences in subgroup reactions are all significant at the 0.01 level in our main models, while more conservative mixed-effects specifications in Table A11 of the Supplementary Information suggest that it is really the outgroup that clearly differs from the base (i.e. RE models including random slopes at the individual level; Heisig and Schaeffer, 2019).

In panel B, which controls for left-right ideological positions, the differences in subgroup reactions are washed out. Thus, Labour was able to use class-related group appeals to expand its catchment to include non-base voters, independent of its redistributive policies, but not independent of its general left-right position in the party political space. Somewhat surprisingly, we see no consistent evidence of a backlash among the traditional base. The base was seemingly not pushed away by the de-emphasis of old class ties, at least not in the short run. ¹³

The Clarity of Signals: When Group and Policy Appeals Are Combined

One reason why the subgroup effects of class emphasis disappear when controlling for left-right ideological positions could be that such effects are conditional on a party’s policy position. Indeed, that is consistent with the idea about how group and policy signals could have amplified each other in Labour’s catch-all strategy.

Using the same survey data as above, we parse out the interaction between class emphasis and policy position by voter group using three-way interaction models. Specifically, we specify four models that are similar to those in Figure 2 except for the additional higher-order interactions. Our main interest is to see if the effect of Labour’s class emphasis among outgroup voters (i.e. non-WC, non-TU members) (i.e. non-working class, non-trade union members) might be particularly strong in cases where Labour adopts a compatible position in terms of redistributive policy or left-right ideology. Figure 3 thus shows how the class emphasis effect varies with the redistributive policy and left-right positions for outgroup voters (and among base voters as a reference, see Table A14 of the Supplementary Information full results). ¹⁴

Figure 3.

The Average Marginal Effect of Labour’s Class Emphasis Across Redistributive Policy and Left-Right Positions Among Outgroup and Base Voters.

Three-Way Interaction Results: Reinforcing Effects Among Outgroup Voters

Overall, Labour’s efforts to moderate its class image and build support with outgroup voters do seem most effective when group and policy appeals are combined and point in the same direction. In Figure 3’s panel A, for example, we see that the marginal effect of union-business emphasis changes noticeably over the observed range on redistributive policy among outgroup voters. While at the most leftist position, a 1-pp shift in emphasis away from union members toward business has no real effect on support, the same shift in class emphasis leads to a 1.6-pp increase in support at Labour’s right-most position on redistributive policies. In panel B, we see a similar outgroup reaction, when Labour emphasizes businesses instead of workers. With left-wing policies, the marginal effect of worker-business emphasis is practically zero, while it shifts to a 1.9-pp increase in Labour support at the most rightist end. In both cases, the interaction of group and policy signals is less pronounced within the traditional base (the right-side subplots within each panel): The line is more flat across the redistributive policy range, and the class emphasis effect never becomes quite significant at the right-wing extreme.

If we turn to left-right positions, panel C predicts for outgroup voters that the marginal effect of union-business emphasis changes from a 0.3-pp decrease in Labour support (insignificant), at the most leftist position, to a 1.1-pp increase in support, at the most rightist position. In panel D, concerning worker-business emphasis, we see that the marginal effect is a 0.2-pp decrease in support (insignificant) at the most leftist position, while this shifts to a 1.3-pp increase at the right-wing end. Again, the equivalent interplay is not found within the base (right-side plots within the panels): The lines are flatter, and the class emphasis effect remains insignificant across the whole range of left-right positions.

In summary, this suggests that class emphasis was indeed most effective among outgroup voters when Labour took a more centrist policy position than its traditional left-wing stance. Thus, while group appeals can be effective on their own, it is also the case that a combined change in group and policy image might pay off electorally—especially among outgroup segments in need of convincing. Once more, we note that the traditional base does not appear to punish Labour, not even when the policy and group appeals are combined to send a strong signal of commitment to the non-base outgroup. This is likely a key explanation for the overall success of Labour’s catch-all strategy.

Conclusion

Scholars have made significant progress in understanding the nuts and bolts of party electoral strategy by analyzing how parties use policy statements to position themselves vis-à-vis the electorate or set the issue agenda. Few would deny, however, that parties appeal to voters in a number of ways. Building on three research fields—group-centric reasoning, class voting, and group appeals—this article has illustrated empirically how group appeals can help parties electorally.

Revisiting the influential British case, we found that the success of Labour’s renowned catch-all strategy was not only predicated on centrist economic policies. Labour also used more symbolic group appeals to downplay ties to its declining base and increase electoral support overall, but especially among voters, who were not of the working class or trade union members. We also showed how the Labour Party was particularly effective in raising support among this outgroup segment, when group and policy appeals were combined to send a strong signal about a change in class image.

Contributions

Our findings provide important nuance to the picture of catch-all politics in Britain by suggesting that Labour’s electoral success was predicated on both a change in policy image and group image. This is consistent with recent studies on class voting in Britain (Evans and Tilley, 2017; Heath, 2015; Thau, 2021) but shows that group appeals not only shaped class differences at the individual level but also Labour’s overall electoral success—an outcome which has otherwise been tied to a “catch-all policy moderation strategy” (Karreth et al., 2013). The results from the British case also contribute to the comparative literature on social democratic parties (e.g. Best, 2011; Kitschelt, 1994; Mair et al., 2004; Przeworski and Sprague, 1986), which has long argued that these parties would need to shed their one-dimensional working class image and build support across various subgroups of the electorate. We add to recent studies on how policy-centric strategies are used for this (Abou-Chadi and Wagner, 2019) by suggesting that old class parties can also renew themselves via group-centric strategies and, not least, the combination of the two.

Furthermore, this article speaks to research on party strategies more generally. Focusing on policy shifts, Adams (2012: 403) has concluded that there is “only weak and inconsistent evidence” on the electoral effects of parties’ policy strategies. This has motivated some studies to explore the conditions under which policy strategies actually work—including accounting for time lags and considering heterogeneity across party family, incumbent status, issue domains, and subgroups in the electorate (Abou-Chadi and Wagner, 2019; Adams and Somer-Topcu, 2009; Bawn and Somer-Topcu, 2012). However, other studies have looked beyond policy to explore alternative strategies (e.g. Jung and Tavits, 2021; Somer-Topcu, 2015). We agree with a recent strand of research that group appeals constitute one such non-policy strategy (Dolinsky, 2023; Horn et al., 2021; Huber, 2022; Thau, 2019) and extend this work by empirically assessing whether group appeals benefit parties electorally, as is assumed.

We also contribute theoretically by laying out how group and policy appeals interact. Although prior work has argued that both policy and group appeals are important (e.g. Howe et al., 2022), their interplay has largely been neglected. Our point that the two appeals target different aspects of the voting calculus, thus presenting parties with a strategic choice between sending a strong but narrow signal or a mixed but broader signal, provides a framework that scholars can use to understand how parties shape cleavage politics (Thau, 2021), how they navigate electoral change (Mair et al., 2004), or how they attempt to build winning majorities in ways less risky than offering conflicting policies or taking positions that are vague and ambiguous (Somer-Topcu, 2015; Tomz and Van Houweling, 2009).

Limitations and Future Directions

We want to discuss four important limitations to our study that future research could usefully address. First, while both the aggregate and individual-level analyses generally supported that group appeals are an effective party strategy, the evidence still comes from one case and a limited number of party-year observations. Future research should therefore focus on additional cases to test if these initial findings generalize. While evidence on voters’ sensitivity to group cues is accumulating (e.g. Bornschier et al., 2021; Robison et al., 2021; Vivyan et al., 2020), the question about whether parties benefit from group appeals has largely been overlooked. Our hypothesis would be that the group appeal strategy works in other contexts, but which role factors such as the electoral system, party types, or cleavage structure might play need to be considered in future studies.

Second, our reliance on existing data on group appeals naturally sets some limitations regarding the scope of the analysis. Beyond testing our findings in other country contexts, as discussed earlier, new group appeals data would help us analyze and understand political developments more recent than 2015 (the end point in our data). For example, many of the social democratic parties that followed the “New Labour” recipe and downplayed their working class images in the late 1990s appear to have reembraced their working class ties. Social democratic parties in Britain, Germany, and Scandinavia are widely understood to have “moved to the left” in recent years, and part of that story is likely a deliberate change in group appeals (e.g. Jeremy Corbyn’s claim “to give working class people a voice in politics” in the Labour Party’s 2019 manifesto). As the available data on group appeals are based on manual coding, studies taking an automated approach would be particularly useful in assessing reliability and scaling up data-collection efforts (see Licht and Sczepanski, 2023).

Third, while we advanced a novel perspective on how group emphasis works in relation to policy positions, our analysis is not ideally suited to test whether a strong or mixed signal strategy is generally most beneficial to parties. Our case selection allowed demonstrating that the strong signal strategy can be effective but not that the mixed signal strategy is generally ineffective. To pursue that question, we need richer data on the incentives that underly parties’ strategic decision-making (Dickson and Scheve, 2006; Huber, 2022), most importantly variations in the size of voter groups, and on other types of voter groups than class-based ones, for example, those rooted in other well-known cleavages like religion, place, gender, or ethnic background. Other aspects of parties’ strategic choices not covered by our analyses include the timing of changes in group and policy appeals (i.e. whether it is more effective to adjust group or policy images first) and parties’ efforts to affect other parties’ group images (e.g. Labour saying that the Conservatives are the party of the rich).

Finally, our focus on group and policy appeals have left out other ways that parties appeal to voters. Particularly, how parties’ issue emphasis relates to their use of group and policy appeals is worth pursuing in future research. For example, the behavior and success of social democratic parties in recent decades have not only been about changing redistributive policies and adjusting class images but also about the attention devoted to welfare issues such as childcare, education, and health, as well as the issue of immigration (e.g. Seeberg, 2023). Thus, examining the full repertoire of what Rohrschneider (2002: 367) has called parties’ “targeting strategies,” including how various non-policy strategies play together, is an important step toward understanding the consequences of party electoral strategies. Once non-policy strategies are considered, it could well turn out that parties have more control over their own electoral fortunes than the evidence provided by the policy perspective has sometimes suggested (Adams, 2012).

Supplemental Material

sj-pdf-1-psx-10.1177_00323217231220127 – Supplemental material for The Group Appeal Strategy: Beyond the Policy Perspective on Party Electoral Success

Supplemental material, sj-pdf-1-psx-10.1177_00323217231220127 for The Group Appeal Strategy: Beyond the Policy Perspective on Party Electoral Success by Mads Thau in Political Studies