Associative issue ownership in a highly fragmented multiparty context: The Netherlands (2021)

Abstract

Associative issue ownership (AIO) has proven its value in describing issue competition and explaining voting behavior. Yet, it is unclear whether and to what extent AIO also differentiates parties and influences vote choice in highly fragmented, multiparty systems. In such a context, parties must differentiate from many electoral competitors, which makes AIO worth pursuing. At the same time, obtaining unequivocal ownership may be a very difficult endeavor in the face of so many rivals. This paper aims to assess these questions empirically by employing the Dutch Parliamentary Election Study 2021 on a system with 17 elected parties (ENPP = 8). At the aggregate level, we find unequivocal issue ownership for 4 of the 14 issues under study. AIO of most other issues is contested, either by parties with very similar policy positions (within-block competition) or by parties with opposing positions (between-block competition). A final set of issues remain unclaimed. At the individual level, perceptions of issue ownership explain the composition of voters’ party consideration sets (pre-elections) and their actual vote choice (post-elections). These impacts are stronger when voters associate the party with an issue they find important. We conclude that AIO perceptions are an important factor to consider when studying party dynamics and voting behavior in a context of highly fragmented multipartyism.

Supplementary Information

The online version of this article (10.1057/s41269-022-00274-3) contains supplementary material, which is available to authorized users.

Introduction

Over the past decades, national elections have become increasingly volatile across Western Europe (Mair 2008). For most electorates, socioeconomic orientations have lost much of their predictive power, simultaneously weakening and realigning the connections between parties and citizens (Kriesi et al. 2006). As voters began to choose, party politics has become more competitive and more fragmented (Borman and Golder 2013). Traditional mass and niche parties can no longer rely on a steady base of supporters. Volatility and fragmentation made it more important for parties to differentiate themselves from the pack, but these very same trends simultaneously made it more difficult for them to do so.

Stimulated by its highly proportional electoral system, the Netherlands has been a forerunner in these trends. Since the turbulent elections of 2002, high levels of electoral volatility have remained a permanent feature of Dutch politics. In 2017, more than a quarter of all Lower House seats switched from one party to another. In 2021, the Dutch electorate voted a post-war record of 17 parties (including four new ones) into the 150 seats parliament.1 The effective number of parties remained approximately 8 in both election years.

In a highly competitive environment, political parties have different strategies to appeal to voters. One of these strategies builds on the idea of issue ownership (Van der Brug 2017; Walgrave et al. 2012, 2015) to which political parties have resorted in response to the declining relevance of socioeconomic determinants and the rising importance of issue competition (Van der Brug 2004). At its core, issue ownership theories state that voters link some themes more closely to one party (the issue owner) than to other parties (Petrocik 1996). This might be particularly useful in multiparty systems, where parties compete with ideologically similar parties: “In order to be competitive they [parties] would want to emphasize those issues that help to distinguish them from their direct competitors” (Van der Brug 2017, p. 531). The literature distinguishes between two types of issue ownership: competence and associative issue ownership (AIO; Walgrave et al. 2012). The former emphasizes performative judgements, centered on the question which party has the best reputation in managing an issue according to voters. The latter emphasizes salience, centered on the question which party is most closely tied to an issue, for instance because that theme is most important to them.

AIO has been developed by “comparativists who mainly study multiparty systems” (Van der Brug 2017, pp. 530–531), if only because competence issue ownership (CIO) is a more contested concept in systems that encompass multiple parties that never had government responsibility (the majority of parties in the Netherlands). In these circumstances, competence-based issue ownership approaches would conflate government experience with issue ownership more strongly.

Yet, while AIO theory has proven its value in empirical research, it has mainly been tested in less fragmented multiparty systems. Van der Brug (2004) developed the concept in the Netherlands for six parties across five issues. Lachat (2014) applied the concept in Switzerland for seven parties and six issues. And Walgrave and Lefevere (2017) tested the concept in Belgium for 13 parties (albeit split in 2 separate party systems) and 5 issues. It is unclear whether and to what extent AIO differentiates parties and issues at the macro-level in highly fragmented, crowded party systems. At the micro-level, it is not clear to what extent AIO helps to explain party consideration and vote choice in this specific context.

This paper aims to assess empirically the presence and predictive power of AIO in a highly fragmented multiparty system. Hence, our study covers 14 political parties and 14 political issues in the 2021 parliamentary elections in the Netherlands. We investigate the concept of AIO with three foci: as a trait of issues, as a trait of political parties, and as a determinant of voting behavior. Thereby this paper aims to contribute to the literature in three distinct ways. First, we assess the presence and strength of AIO in a multiparty context across a wide range of issues, in comparison to a large number of electoral competitors. To this end, we propose a more fine-grained measurement, based on the distance between ‘first owning’ and ‘second owning’ parties and on the extent to which ownership perceptions are present across different segments of the electorate. Second, we test the impact of individual issue ownership perceptions in explaining both stages of the two-stage process of voting behavior in multiparty systems: (i) determination of the party consideration set (pre-elections), and (ii) the actual vote choice within these party consideration sets. Earlier studies mainly tested the effects of ownership on vote choice (second stage). Third, we integrate AIO theory with voter perceptions of the most important problems facing society. This informs a tentative understanding of campaign dynamics during the 2021 elections and provides a more specific test of issue ownership theory at the micro-level. Our fixed effects models control for rivaling voter–party connections (ideological proximity; leadership sympathy) and assess the unconditional and conditional effects of AIO on voting behavior.

Theory: issue ownership against the backdrop of volatility and fragmentation

Electoral volatility in the Netherlands

Until the 1960s, during the time of pillarization, Dutch voters exhibited a remarkable level of electoral stability, spurred by their segregation into internally coherent societal systems with their own political party. As these pillars eroded from the 1970s onwards, socioeconomic and demographic traits became less decisive factors predicting vote choice. This culminated in the twenty first century, which consistently saw high levels of electoral volatility. While floating voters have been on the rise across Western Europe (Drummond 2006), the Netherlands stands out from a comparative perspective (Mair 2008). An average of more than 20% of the parliamentary seats changed ownership in each subsequent parliamentary election since 2000, with peaks in 2002 (> 30%) and 2017 (> 25%).

Evidently, context matters. Proportional and multiparty systems facilitate electoral volatility because voters are offered more options (Tavits 2005). This is in line with recent studies that demonstrate how vote switching often takes place between parties that are ideologically close (Van der Meer et al. 2012). The presence of multiple parties with similar worldviews that must compete for electoral support makes vote switching more likely than in two-party systems in which the election campaign is between parties that propagate ideologically opposing programs.

On an individual level, research has pointed to the weakening attachment between parties and voters, due to the declining impact of social cleavages (e.g., Aarts and Thomassen 2008). Increasing volatility can be considered the result of an emancipation process in which voters are no longer bounded by traditional loyalties but instead make independent, informed political choices. Hence, to understand the volatility of party preferences, it is important to bring issues, parties, and candidates into the equation (Dalton 2010).

Issue ownership

Traditionally, voting models have focused on factors such as ideological proximity (e.g., Downs 1957) and party and candidate evaluations (e.g., Hopmann et al. 2010) to explain voting behavior. Against the backdrop of increased electoral volatility, issues attracted scholarly attention as an additional factor influencing vote choice. One central component of this strand of research is the idea of issue ownership, which refers to the fact that specific political parties are, in voters’ minds, identified with specific policy issues (Van der Brug 2004; Walgrave et al. 2015). The party most strongly linked to an issue is said to ‘own’ it and may greatly benefit from the salience of this issue during an election campaign, as perceived issue ownership allows it to stand out in comparison to potential rivals.

Following priming theory (Iyengar and Kinder 1987), highly salient issues are expected to be most central in voters’ considerations when evaluating political actors. When citizens link an issue to a specific party and perceive this party as the legitimate owner, they are more likely to vote for this party at the ballot box (Geers and Bos 2017). Issue ownership becomes a strategic asset for parties, particularly when their owned issues become salient in the run-up to elections (Budge 2015; Damstra et al. 2021). Parties have a strong incentive to focus their campaign efforts on their own ‘owned’ issues, to force their competitors to speak out on these themes, and to sidestep the issues that may benefit their competitors.

The literature distinguishes between two dominant conceptualizations of issue ownership (Walgrave et al. 2012; Van der Brug 2017). The first refers to competence: people link a given issue to a party based on the perceived competence of that party to manage the issue (e.g., Green and Hobolt 2008; Petitpas and Sciarini 2022; Petrocik 1996). In this conceptualization, the party that is considered to be the most capable of handling or resolving a problem of concern by voters is perceived owner of that issue. The second conceptualization is more freely associative. It refers to the spontaneous connection of parties with issues in the minds of voters, regardless of whether voters consider the party most competent (Walgrave et al. 2012, p. 772). Competence and AIO may overlap. Yet, they represent different dimensions of ownership that are independent determinants of voting behavior, not only conceptually but also empirically (Walgrave et al. 2012; Lachat 2014).

AIO is a more appropriate concept in multiparty systems (Van der Brug 2017). Especially in the case of high fragmentation, the opportunities for political parties to gain a reputation of competence on specific policy issues are scarce and skewed. In the Netherlands in 2021, only 6 of the 17 parties elected into parliament have ever been part of a national government coalition. The other eleven parties never had this sort of responsibility, depriving them from the possibility to profile on specific policy dossiers in terms of (proven) competence. In addition, CIO comes with challenges of endogeneity. Due to its motivational component, CIO is as much the result of party preference and vote choice as it is a driver of it (e.g., Stubager 2018; Walgrave et al. 2016). Perceptions of AIO lack this motivational component and are less likely to suffer from endogeneity issues in explanatory models of voting behavior (cf. Lachat 2014). Even without the motivational component, AIO is a strategic benefit to parties, as it enables them to stand out from their rivals. This is even more imperative in an electoral system in which many parties must compete along multiple political dimensions.

Yet, fragmentation makes it hard to achieve undisputed issue ownership. First, while ownership tends to be rather stable over time (e.g., Christensen et al. 2015; Seeberg 2017), especially the associative dimension (Tresch et al. 2015), new parties can emphasize and claim new issues that arise (Walgrave and De Swert 2007). Particularly in multiparty systems, some issues remain de-emphasized, whereas the ownership of other issues is challenged by multiple parties (Green-Pedersen and Mortensen 2010, 2015). Second, as the electoral size of political parties shrinks and as the effective number of parties increases, the classification of parties as issue owners becomes problematic: “In a fragmented party system with six parties, it is very tough for a party to pass the 50-percent threshold” (Walgrave and De Swert 2007). Therefore, we propose a set of criteria to assess the strength of ownership in a fragmentized, multiparty context:

• (i)the absolute percentage of citizens who associate the issue with a specific party;
• (ii)the difference between that percentage of associations and the percentage of citizens associating the issue with the runner-up; and
• (iii)the degree to which different subgroups of voters share the same issue-party association.2

Based on these criteria, we are able to identify issues that are convincingly owned by a single party, issues for which ownership is still undecided as multiple parties compete for it, and issues that are not owned at all. Hence, our macro-level analyses on AIO will focus on these three criteria.

The impact of issue ownership on voting behavior

Voting behavior in multiparty systems can be characterized as a two-stage process (Oscarsson and Rosema 2019). Voters do not place the same level of consideration on all relevant political parties on election day; rather, they first tend to reduce the number of parties they consider as true alternative options. This subset of parties that voters take into consideration is called a consideration set. The actual vote choice tends to take place within this consideration set. Voting as a two-stage process offers an explanation for the paradox that voters’ political positions can be relatively consistent at the aggregate level over time, even when the actual vote distributions are not (e.g., Van Holsteyn and Den Ridder 2018): most voters switch within blocks of parties that are ideologically close (Van der Meer et al. 2012), i.e., within ideological relatively consistent consideration sets. Variation in issue ownership offers a tentative explanation for the composition of and the choice that is made within consideration sets.

Empirically, voter perceptions of issue ownership influence voting behavior, next to factors such as ideological proximity and general party and candidate sympathy (e.g., Van der Brug 2004; Bélanger and Meguid 2008; Green and Hobolt 2008; Walgrave et al. 2012; Lachat 2014). Any form of issue ownership may help to give profile to political parties, making it more likely not only that they will be elected (Karlsen and Aardal 2016; Petitpas and Sciarini 2022) but also that voters include them in their choice sets. However, these effects are unlikely to be unconditional. Voters will be particularly swayed by ownership of issues that they consider salient (Bélanger and Meguid 2008; Walgrave et al. 2012; Lachat 2014).

Based on these considerations, we expect that the influence of perceived AIO on the probability of a party being included in voters’ consideration sets and on the chance of receiving the vote increases when issue salience is high. In sum, the above considerations can be formalized in the following set of hypotheses:

H_1a

Perceptions of AIO increase the probability of being included in voters’ consideration set of political parties.

H_1b

This probability is higher when the owned issue is salient to voters.

H_2a

Perceptions of AIO increase the probability of voting for a party, given inclusion in the consideration set of political parties.

H_2b

This probability is higher when the owned issue is salient to voters.

Data and method

To answer our research questions and test our hypotheses, we rely on panel survey data of the Dutch Parliamentary Election Studies (DPES) 2021 (Sipma et al. 2021). Data were collected during two waves, the first in the months preceding the parliamentary elections of March 17, 2021, and the second in the weeks after the elections. Fieldwork was carried out by I&O Research and CentERdata.

Independent variables

We measure AIO in the pre-electoral wave of the DPES by means of the following questions: ‘What party comes to mind in the first place when you think about [issue]?’ and ‘What party comes to mind in the second place when you think about [issue]?’ The list of 14 parties covers those that were already elected into parliament before 2021 complemented with new party JA21 (for which polls predicted that it would win at least one parliamentary seat in the 2021 elections). In addition, respondents could fill in another party (not from the list), by choosing the open-ended answer category ‘other party.’ The selection of 14 issues is based on an analysis of the most important or prominent policy dossiers in the years preceding the 2021 election, including political as well as valence issues.

The literature does not offer a standardized measure of AIO (Walgrave et al. 2015). Our operationalization, as the spontaneous association of a party to a given issue, is very similar to earlier operationalizations (Walgrave et al. 2012; Lachat 2014).

Moderator

The saliency of issues was measured by means of a question in the post-election wave: ‘What do you think are the most important national problems in our country?’ Respondents could give multiple answers to this open-ended question (which led to a maximum of nine problems per respondent). We recoded the open-ended answers into 24 broad categories (see first column Table 3, “Appendix”). Some but not all of these categories could be connected to the 14 issues included in the issue ownership battery. Models that include issue salience therefore only cover the issues (from the issue ownership question battery) and the categories (of national problems) that can be linked to each other.3

Table 3.

Connection between issues related to salience and ownership

Most important national problem	Issue: ownership party	No match
V061–9	N46–73
1 Economy/financial situation	Economy 1
2 Social security	Poverty 1
3 Politics		X
4 Crime	Security
5 Defense		X
6 Healthcare	Healthcare
7 Education	Education
8 Income/price levels/taxes	Economy 2
9 Employment	Employment
10 Traffic/mobility		X
11 Housing	Housing
12 Environment	Climate change
13 Population		X
14 Minorities	Immigration 1
15 Norms and values	Norms and values
16 Media		X
17 European integration		X
18 Inequality/poverty	Poverty 2
19 Intolerance/discrimination	Discrimination
20 Foreign policy/international security		X
21 Regulation/big government		X
22 Polarization/dividedness		X
23 Immigration	Immigration 2
24 Corona		X

Dependent variables

In line with the two-stage model of voting behavior in multiparty systems, we employ two dependent variables: party consideration set inclusion (pre-election) and vote choice (post-election).

To establish each voter’s party consideration set, we rely on the reported propensity to vote (PTV) for a party (Van der Brug 2004). In line with the model of the process (Oscarsson and Rosema 2019), this question was posed in the pre-election wave and reads: “Would you please indicate on a scale from 1 to 10 how probable it is that you will ever vote for [party]? On this scale, ‘1’ means that you will never vote for this party and ‘10’ means that you will certainly vote for this party sometime.” This question has been raised for the same 14 parties that are included in the issue ownership question battery. PTV measures have various advantages over the categorical measure of vote choice, providing detailed information on utility (Van der Eijk et al. 2006). We use these PTV scores to operationalize the composition of the consideration set along the two common principles behind establishing consideration sets (Oscarsson et al. 1997; Oscarsson and Rosema 2019): the relative gap (the distance between the PTVs of parties in the consideration set should not be too large) and the absolute preference (the non-negativity of PTVs of parties in the consideration set). The procedure proposed by Van Holsteyn and Den Ridder (2018) balances both.4 To be included in respondents’ consideration sets, the party needs to have been assigned the highest PTV by the respondent or one or two steps lower. Inclusion in the consideration set is thus a dichotomous variable for each party. Respondents who report little chance to vote for any of the 14 parties (PTV < 5)5 and respondents who do not differentiate between parties (0 or more than 10 parties in their consideration set) are treated as missing.

The second dependent variable, vote choice, was measured as the reported party choice at the 2021 Lower House election. This question is asked in a straightforward way: “Which party did you vote for in the 2021 parliamentary elections?” We excluded non-voters and voters for parties other than the main 14 on the list from the explanatory analyses.

Control variables

The multivariate models control for rivaling explanations of consideration set inclusion and vote choice. The first is perceived ideological distance between the respondent and the party (cf. Van der Brug 2004; Lachat 2014). We operationalize this ideological distance as the absolute difference between self-placement and the perceived party position on an 11-point left–right scale. The second is party leader sympathy (cf. Van der Brug 2017, p. 534), measured on a scale from 0 (no sympathy) to 10 (complete sympathy). Both questions (perceived positions and leader sympathy) have been asked in the pre-election wave for the same 14 parties mentioned before.

Method

To assess the impact of AIO on consideration set composition and vote choice at the individual level, we rely on fixed effects models. We transposed the data from the wide format in DPES2021 (unique variables for each party) to the long format. In other words, we nest respondent-party dyads (as level 1 observations) in respondents (level 2). By estimating fixed effects models in Stata 14 (employing the xtlogit, fe command), we eliminate the impact of respondent traits and focus on within-respondent variation. This has the advantage of mitigating the statistical noise that occurs in between-person comparisons.

We build up the analyses sequentially. First, we analyze the general impact of perceived ownership unconditionally, i.e., regardless of the salience that respondents attach to these issues. In these models, we include all 14 issues. Next, we specify that model to assess the relevance of issue salience, testing whether perceived ownership of salient issues has an additional explanatory effect. These analyses are blind to the substance of the issues, in order to focus on the overall effects of issue ownership.6

The models of consideration set composition encompass all 14 parties in the standard data set. These 14 parties collected 93.7% of the votes. The models of vote choice do not encompass all 14 parties; rather, for each respondent, we only include the parties that were mentioned as part of the consideration set in the pre-election wave.7 Missing values on any of the variables led to list-wise deletion from the analysis.

Descriptive analyses I: issues and salience

To explore the extent to which issue ownership can be observed in a highly fragmented, multiparty context, we start by examining the essential part of associative ownership: the connection between issues and parties that is made in the minds of voters. In the weeks before the 2021 parliamentary elections, we presented Dutch voters a list of policy issues and asked them to indicate, for each issue, whether they could spontaneously name a political party that they associated with it. Figure 1 displays the degree of ownership per issue. The bars indicate the percentage of voters not having any party association. Issues with many party associations are on the left side of the figure; issues with few party associations at the right.

Fig. 1.

Percentage of respondents not having any party association per issue

Associative ownership varies greatly across issues. Several issues are frequently linked to a political party: the economy, climate change, and discrimination stand out: less than 15% of the respondents does not connect these issues to a party. On the other end of the scale, we observe the issues for which fewer people have a spontaneous party association. Safety, euthanasia, and housing are still relatively open to party ownership.

However, for ownership to have electoral consequences, the issue at hand must have a certain level of societal relevance. We asked respondents at the time of the 2021 elections what they considered the most important national problems. We recoded this open-ended question into a nominal variable consisting of 24 problem categories (for the complete list, see Table 3 in “Appendix”). Given the focus of the current study, we only use the eleven categories that can be linked to one of the issues examined in the context of party associations. Figure 2 plots the share of respondents mentioning each of these issues, arranged from the issues most often mentioned (left-hand side) to the ones least often mentioned (right-hand side). Climate change stands out in terms of perceived societal relevance, followed by housing, healthcare, and immigration. The high salience of housing is interesting considering that this issue scored ‘lowest’ on party association.

Fig. 2.

Percentage respondents mentioning each issue as important national problem

Descriptive analyses II: patterns of AIO

To assess the strength of issue-party associations in a fragmented, multiparty context, we look at three distinct features: (i) the absolute association, (ii) the extent to which there is a clear gap between the first issue owner and a challenger, and (iii) the similarity of ownership perceptions among different segments of the electorate. AIO is strongest when many voters associate the party to the issue, when the gap with potential challengers is large, and when ownership perceptions are shared by partisans and non-partisans.

The aggregate data set reports the percentage of respondents associating each of 14 parties with each of 14 issues. We employ exploratory factor analyses (PCA, Varimax rotation) to identify overarching structures behind party–issue associations (i.e., clusters of issues with similar patterns of issue ownership).8 We find four such structures. Figure 3 displays the issues that make up each cluster, together with the parties that score highest on ownership per issue. Remarkably, these clusters align well with the clusters identified in a different context by Seeberg (2017, p. 479).

Fig. 3.

Percentage of respondents naming party as issue owner, by issue (four clusters)

The first cluster consists of five issues that are mostly associated with parties at the right-hand side of the political spectrum: the economy, the European Union, safety, democracy, and norms and values. As associative owner of these issues, voters most often mention the liberal conservatives (VVD), followed by the Christian-democrats (CDA) and the progressive liberals (D66). In the second cluster, we observe four issues that stem from the socioeconomic sphere: poverty, health care, employment, and housing. Voters connect these issues primarily to economically left-wing parties Labour (PvdA) and the Socialists (SP), followed by their main traditional antagonist on these issues, the liberal–conservatives (VVD). Only two issues make up the third cluster: immigration and discrimination. Voters associate the radical-right populist Freedom Party (PVV) with these themes, followed by the extreme-right Forum for Democracy (FVD). Finally, the fourth cluster consists of three issues that voters mainly link to progressive parties: climate change, euthanasia, and education. Progressive parties most frequently associated with these topics are the Greens (GroenLinks) and the progressive liberals (D66). For euthanasia, however, there is a group of voters that assigns associative ownership to the orthodox Christian parties: the Reformed (SGP) and the ChristianUnion (CU).

To assess the strength of AIO in a fragmented, multiparty context, absolute numbers are not a sufficient indicator. A cut-off point of 50% may be an unlikely threshold (cf. Walgrave and De Swert 2017, who argued this for 6 rather than 14 parties). Hence, we propose an alternative measurement by adopting two additional criteria: the gap in issue associations between the modal party and the runner-up, and the agreement of ownership perceptions across different segments of the electorate.

In the first issue cluster, two issues are owned unequivocally. First, we observe strong ownership by the liberal conservatives (VVD) on the economy. No fewer than 70% of the respondents spontaneously associates the issue with this party, compared to 35% for the runner-up.9 This association is made across groups of voters: even among those who do not consider voting for the VVD, a majority (66%) spontaneously associates the economy with this party. In a similar vein, though less clear-cut, we observe associative ownership by the liberal conservatives on the safety issue: 48% of the respondents spontaneously connects this theme to this party, followed by 26% for the runner-up. However, party preference plays a larger role: among respondents who do not consider voting for the VVD perceived ownership is much lower (38%).

In the second cluster, we find very small differences in party-issue associations. This suggests that there may be latent conflict between parties on the ownership of these issues. Indeed, across these issues, perceived ownership is (much) higher among the actual voters of these parties than among the potential voters and the non-voters.

In the third cluster, we find another clear case of AIO. The immigration issue is strongly owned by the radical-right Freedom Party (PVV). The distance to the runner-up is substantial (55% versus 31%), and the spontaneous association is made by different segments of the electorate, even by respondents who do not consider voting for the PVV (55.4%). By contrast, the issue of discrimination does not work at all as a measure for issue ownership. Response patterns show that different logics are at play.10 Some associate discrimination with a party because it puts the issue on the political agenda as a societal problem; others associate discrimination with a party because of perceived discriminatory rhetoric or policies by that party. Because of this inherent duality, we refrain from using this issue in further analyses.

Finally, in the fourth cluster, we observe strong ownership on climate change. The Greens clearly dominate this issue, the distance to the second owning party is big (68% versus 30%), and ownership perceptions are broadly shared, even by voters who would never consider the Greens (66%).

All in all, AIO exists even in a highly fragmented multiparty context. We only find strong party ownership for 4 of the 13 issues under study (excluding discrimination): the economy and safety (owned by the liberal conservative VVD), immigration (owned by the radical-rightwing PVV), and climate change (owned by the Greens). On the other issues, ownership is not that clear-cut. Finally, ownership of some issues is being contested by electoral rivals. This is most notably for the issues of poverty and health care, where particularly Labour (PvdA) and the Socialists (SP) are in fierce competition.

Explanatory analyses: AIO and voting as a two-stage process

Next, we test to what extent AIO is an electoral boon. Table 1 presents the effects of AIO on the inclusion of 14 parties in voters’ consideration sets. Model 1 shows the linear effect of generic issue ownership: The more issues a party owns in the perception of respondents, the more likely it will be included in these respondents’ consideration sets (b = 0.43). Next to issue ownership, we find a significant negative effect of ideological distance and a positive effect of leadership sympathy. Model 2 replicates model 1, but now only using the subset of ten issues that we could connect reliably to the most important problems facing the Netherlands according to the respondents. The results confirm that AIO has an effect beyond ideological distance and leadership sympathy. This supports Hypothesis 1a. Model 3 adds ownership of salient issues as a determinant. We find that, in addition to the general effect of AIO, being the owner of at least one issue that respondents find salient provides an additional boost to the likelihood of being included in these respondents’ consideration sets (b = 0.43). Model 4 adds the number of salient issues of which parties are the owner. That addition proves to be non-significant (b =  − 0.03, ns), suggesting that owning at least one salient issue is all that matters. All in all, we find support for Hypothesis 1b.

Table 1.

Explaining inclusion of parties in voters’ consideration sets, logistic fixed effect models (Stata: xtlogit, fe)

	Model 1			Model 2			Model 3			Model 4
	b	s.e.	Significance	b	s.e.	Significance	b	s.e.	Significance	b	s.e.	Significance
Issues owned (full set)	0.43	0.01	***
Issues owned (subset)				0.56	0.01	***	0.51	0.02	***	0.51	0.02	***
Issue owner of salient problem (dichotomous)							0.43	0.05	***	0.46	0.11	***
Issue owner of salient problems (number)										− 0.03	0.09
Left–right distance	− 0.42	0.01	***	− 0.43	0.01	***	− 0.43	0.01	***	− 0.43	0.01	***
Party leader sympathy	0.46	0.01	***	0.48	0.01	***	0.49	0.01	***	0.49	0.01	***
n(observations)	30,713			30,610			23,196			23,196
n(respondents)	2688			2677			1996			1996

*p < .05; **p < .01; ***p < .001

Next, we turn to the second step of the voting process in fragmented multiparty systems (see Table 2): vote choice, given parties’ inclusion in respondents’ consideration sets. Again, the control variables have consistently large and significant effects. In line with the literature, leadership sympathy has a particularly strong effect in this second stage of the voting process (b = 0.82 in model 5). It also fits the interpretation of the Dutch elections according to political commentators, who emphasized that sitting Prime Minister (and VVD party leader) Mark Rutte primarily campaigned on the theme of leadership and was challenged on precisely that theme by runner-up D66 party leader (and sitting Minister) Sigrid Kaag.

Table 2.

Explaining party choice given consideration set, logistic fixed effect models (Stata: xtlogit, fe)

	Model 5			Model 6			Model 7			Model 8
	b	s.e.	Significance	b	s.e.	Significance	b	s.e.	Significance	b	s.e.	Significance
Issues owned (full set)	0.17	0.02	***
Issues owned (subset)				0.18	0.02	***	0.5	0.03	***	0.14	0.03	***
Issue owner of salient problem (dichotomous)							0.24	0.09	**	0.19	0.16
Issue owner of salient problems (number)										0.05	0.11
Left–right distance	− 0.23	0.03	***	− 0.24	0.03	***	− 0.25	0.03	***	− 0.25	0.03	***
Party leader sympathy	0.82	0.04	***	0.85	0.04	***	0.85	0.04	***	0.85	0.04	***
n(observations)	6403			6379			4919			4919
n(respondents)	1794			1787			1384			1384

*p < .05; **p < .01; ***p < .001

The conclusions are similar to the ones related to Table 1. We find significant effects of issue ownership, regardless of the salience of these issues. Based on the full set of issues, AIO has a significant, positive effect in model 5 (b = 0.17); and so does issue ownership of the subset of ten issues in model 6 (b = 0.18). This supports Hypothesis 2a. Owning at least one salient issue has an additional positive effect on vote choice, according to model 7 (b = 0.24). Yet, again, this effect is not linear: The addition of a linear effect in model 8 is not significant (b = 0.05, ns) and does not improve the model fit. All in all, we find support for Hypothesis 2b.11

Discussion

This paper aimed to assess empirically to what extent AIO (Walgrave et al 2012; Van der Brug 2017) is a useful concept in a highly fragmented, multiparty context. On the one hand, obtaining or maintaining issue ownership may be a way for parties to differentiate themselves from their rivals. On the other hand, unequivocal ownership may be nigh impossible in the face of so many competitors. To this end, we focused on the 2021 parliamentary elections in the Netherlands, which resulted in a severely scattered parliament of 17 parties (including four new ones). The rather extreme Dutch case functions as a Petri dish. Precisely because of its high proportionality and large (effective) number of political parties, it allows us to isolate mechanisms behind party consideration and voting behavior. Particularly, it shows the use of approaching voting behavior as a two-stage process, and the value of AIO when there are many parties that compete along many issues on many dimensions.

First, we explored the presence and strength of AIO on the aggregate level by looking into patterns of issue–party associations made by citizens. These analyses confirm that AIO exists, even in the face of severe fragmentation. We identified four issues on which a single party has established undisputed ownership: economy, safety, immigration, and climate change. Partisanship plays a very modest role on issue ownership perceptions on these issues. On other issues, AIO is not (yet) decided, as multiple parties are still in fierce competition and ownership perceptions are more strongly related to partisanship, although in different ways. In some cases, parties that compete for ownership represent opposing views and opposing electoral blocks on the issue at hand (such as progressive versus conservative positions on ethical issues). Ownership effects may hold, though primarily within party blocks and corresponding subsets of the electorate. In other cases, the competing parties come from ideologically congruent party blocks (such as the socialists and the social democrats), that likely share the same consideration sets of voters. Finally, we observe a set of issues that are simply not owned yet. Housing is a particularly intriguing one: although many citizens consider this issue highly relevant, it has not yet been claimed.

Second, we examined the applicability of AIO by its impact on individual vote choice. To account for the fragmentized, multiparty context, we adopted the idea of voting as a two-stage process (cf. Oscarsson and Rosema 2019): the determination of a consideration set, followed by actual vote choice. The analyses indicate that AIO matters. Above and beyond the impact of ideological distance and leadership sympathy, perceived issue ownership has a positive effect on voting behavior (particularly when the issue is salient). These effects are not limited to the second stage (the final vote, given the consideration set composition) but also apply to the first stage (composition of the consideration set). This finding challenges a silent assumption in models of voting behavior. All in all, we conclude that even in a highly fragmented, multiparty system, the associative dimension of issue ownership is a relevant factor to consider when studying party dynamics and voting behavior.

There are some implications to employing AIO in a fragmented multiparty setting. First, absolute percentages do not suffice to assess the strength of AIO, even though we found similar clusters of issue associations as in comparative studies (cf. Seeberg 2017). We therefore proposed a measurement that also looks at (i) the discrepancy between the parties that are most associated with an issue, and the runner-up and (ii) the breadth of the electoral support base. This allowed us to obtain a more fine-grained understanding of patterns of issue ownership. We distinguish between issues with undisputed party ownership, with contested but not crystallized issue ownership, and with no ownership. Second, we show that AIO has relevant explanatory power at both stages of the voting process. In multiparty systems, it is important to distinguish between (i) inclusion into a consideration set, and (ii) actual vote choice (Oscarsson and Rosema 2019). AIO explains both elements, thereby contributing to competition within as well as between blocks of parties.

Unfortunately, our analyses could not capture time dynamics beyond a single election to assess the stability of composition sets (Rekker and Rosema 2019) and perceived issue ownership (Christensen et al. 2015; Seeberg 2017). Although AIO perceptions tend to be rather stable over time (Tresch et al. 2015), they are not set in stone, especially not when issues are contested. Because issue salience has a moderating impact and can vary greatly over time, a long-term perspective would have allowed us to isolate the underlying mechanisms driving the effects of AIO. This is particularly relevant for three reasons. First, a longitudinal perspective enables a better understanding of the dynamics between parties’ campaign strategies, media attention, and perceived issue ownership, as all face stronger constraints in a fragmented multiparty setting. Second, a long-term panel perspective will offer more clarity on the contested causal relationship between issue ownership and voting behavior. Third, our data stem from the first months of 2021, amidst the severe COVID-19 pandemic. The pandemic affected the Dutch election campaign, not least because Prime Minister (and VVD party leader) Mark Rutte, repeatedly sought to reduce televised debates to discussions about leadership and crisis management. Consequently, our findings that AIO matters in fragmented multiparty systems may even have been a conservative estimation of the dynamics of issue competition under more regular circumstances.

Supplementary Information

Below is the link to the electronic supplementary material.

Biographies

Tom van der Meer

is a Professor of Political Science, in particular Legitimacy, Inequality, and Citizenship, at the University of Amsterdam. In 2017 and 2021, he was Co-director of the Dutch Parliamentary Election Study. His main research interests are political trust, electoral behavior, and social capital.

Alyt Damstra

holds a research position as a Member of the Scientific Staff of the Netherlands Scientific Council for Government Policy (WRR). She is also a Postdoctoral Researcher at the Department of Political Science of the University of Amsterdam. Her research focused on the triangular relationship between the economy, news government, and public opinion.

Appendix

See Table 3.

There are two ways in which we might want to measure the number of respondents that mention the economy as the most important national problem. First a narrow operationalization, covering only those respondents who gave an answer that was grouped under the heading ‘economy/financial situation’ (variable: mip_economy1). Second, a broader operationalization, covering those respondents whose answers were categorized as ‘economy/financial situation’ and/or ‘income/price levels/taxes’ (variable: mip_economy2). Similarly, mip_poverty1 refers to those respondents whose answers were grouped under ‘inequality/poverty’ (more narrow operationalization) and mip_poverty2 refers to those respondents whose answers were grouped under ‘inequality/poverty’ and/or ‘social security’ (broader operationalization).

Declarations

Conflict of interest

The authors hereby state that there is no conflict of interest—social, financial, or otherwise—with regards to this manuscript.