Sleeping With the Enemy: Partners’ Heterogamy by Political Preferences and Union Dissolution. Evidence From the United Kingdom Open Access

Abstract

We examine whether union dissolution is associated with partners’ (mis)match on political preferences, defined as self-reported closeness, intention to vote, or reported vote for a specific party. Previous studies have shown that partners’ heterogamy by ethnicity, education, and other dimensions increases the risk of union dissolution because of differences between partners in lifestyles, attitudes, and beliefs or because of disapproval from family and community members. We posit that similar arguments can apply to political heterogamy and test this hypothesis using UK data from the British Household Panel Study and the UK Household Longitudinal Study. The data offer a unique opportunity to assess the role of heterogamy by political preferences while controlling for heterogamy in other domains and for other partners’ characteristics over a long period (1991–2019). The data also facilitate a more specific analysis of the referendum on the United Kingdom's permanence in the European Union (known as the Brexit referendum). We find a positive association between political heterogamy and union dissolution, which is as strong as some other forms of heterogamy. The role of diverging opinions on the Brexit referendum in union dissolutions appears to be even more important than the role of partners’ differing party preferences.

Introduction

Social scientists have identified various factors that influence union survival (for reviews, see Lyngstad and Jalovaara 2010; Mortelmans 2020), including partners’ (mis)match on several dimensions, such as socioeconomic status (Musick et al. 2020; Qian 2017; Schwartz and Mare 2012; Theunis et al. 2018), social origin (Henz and Mills 2018), religion (Wright et al. 2017), race and ethnicity (Feng et al. 2012; Smith et al. 2012; Wong 2016; Zhang and Van Hook 2009), age (England et al. 2016), health (Torvik et al. 2015), and personality traits (Arpino et al. 2022). These studies have typically found that couples in which the partners have different characteristics (heterogamous) have a higher risk of breaking up than homogamous partnerships (Schwartz 2013).

Our study expands the literature on union stability and dissolution by integrating political heterogamy into an area of research traditionally dominated by a focus on sociodemographic characteristics, such as age, education, and ethnicity. Unlike previous research that has largely focused on these dimensions, we examine how political preferences interact with union stability and dissolution. The inclusion in the analysis of union stability of political similarity as a form of homogamy reflects current societal trends of the increasingly personal and polarizing role of politics, suggesting that political alignment among partners could be as critical to partnership stability as other well-established factors.

Our study aligns with the emerging field of political demography, which explores the intersection of demographic processes and political behaviors (Teitelbaum 2005). Notable examples of political demography research include studies on how demographic changes influence electoral outcomes (Teixeira 2018), the impact of age structure on political stability (Goldstone et al. 2012), the role of migration in shaping political landscapes (Hampshire 2013), and the influence of political ideology on fertility intentions (Arpino and Mogi 2024).

Our research has implications for the emerging field of political demography by highlighting the interplay between individual-level partnership dynamics and larger political structures. By examining dissolution risks among politically heterogamous couples, our study not only underscores the importance of political alignment for individual relationships but also hints at the macro-level implications, such as how political homogamy might contribute to societal polarization. In the context of the United Kingdom, where political preferences are sharply divided, our work sheds light on the potential demographic consequences of these divisions, providing insights into how political affiliations can influence broader patterns of social cohesion. Studies on union dissolution have focused relatively little on discord among partners in values and attitudes. Although our focus is on political heterogamy, our study can stimulate further research on partners’ discord on other value dimensions.

In increasingly divided times (e.g., see Duffy et al. 2019; Hobolt et al. 2021; Layman et al. 2006; Pew Research Center 2014, 2019), it is important to understand how politics and union dissolution intersect. Several recent studies have considered partners’ mating in terms of political ideology or preferences (e.g., Alford et al. 2011; Hersh and Ghitza 2018; Horwitz and Keller 2022; Klofstad, Sokhey, and McClurg 2013). These studies have mostly examined the extent to which agreeing on politics positively affects a couple's relationship quality, but they have not directly addressed whether it influences the likelihood of staying together. Our study examines whether having different political views plays a role in couples’ breaking up, which is important to understand because politics is a salient dimension in many individuals’ lives and can influence relationships. Politics, in fact, is often a topic of conversation or a joint activity for partners (Daenekindt et al. 2020). Even individuals who are not directly interested in politics are often confronted with the need or opportunity to adopt political positions. Even individuals who do not actively search for political news are exposed to it, for example, via (social) media (Fletcher and Nielsen 2018).

Individuals might feel passionate about political issues, and conversations about politics often become heated. Opposing political views can lead to difficult interactions. Although political arguments can be avoided at work or with friends, constantly avoiding hot political topics at home is much more difficult and exhausting. Research that examined partners’ matching on political attitudes and preferences seems to support the idea that individuals try to avoid forming a union with partners who have different political views (Alford et al. 2011; Hersh and Ghitza 2018; Klofstad, McDermott, and Hatemi 2013), which might indicate that they foresee risks of conflicts from political heterogamy. Although homogamy on other traits (e.g., ethnicity) is stronger, couples also tend to show a considerable degree of political homogamy (Huber and Malhotra 2017). In the United States, for example, 70% of married couples match on political affiliation (Hersh and Ghitza 2018). In the United Kingdom, up to three quarters of couples are composed of partners who identify with the same party (e.g., Bélanger and Eagles 2007; Lampard 1997). In other systems with more political parties, these rates might be lower owing to the greater diversity of political options available.

One could contend that any association between political heterogamy and union dissolution might simply reflect the influence of other partners’ dissimilarities (e.g., by socioeconomic status or religion). Instead, we argue that political heterogamy might have an independent role in union dissolution.

We focus on the United Kingdom, an interesting case study for the high level of political polarization that characterizes its political system, which is dominated by a few big parties (Conservatives, Labour, Liberal Democrats). Following prior research (Tilley 2015), we operationalize political homogamy as a match in party preference or vote and define political heterogamy as a mismatch. In addition, we expand the representation of partners’ political convergence by using an emerging dimension that is increasingly more salient than party positions (Hobolt et al. 2021) in the political debate in the United Kingdom: Brexit. Research has found that Brexit identities are prevalent, are personally important, and cut across traditional party lines (Hobolt et al. 2021). For partners with different opinions, the referendum might have triggered stress and conflict, affecting partnership stability.

Specifically, using data from two population-representative British surveys—the British Household Panel Study (BHPS) and the UK Household Longitudinal Study (UKHLS)—we consider three measures of political heterogamy. First, we categorize couples as politically homogamous or heterogamous by whether partners support the same or different political parties. In this first set of analyses, we group all homogamous couples together and all heterogamous couples together, regardless of the specific party supported (party heterogamy). Second, we use the same party-based definition of homogamy/heterogamy but differentiate couples within each group according to the specific party preferences (party heterogamy–refined). Third, we define homogamy/heterogamy using a different dimension: partners’ positions on the 2016 Brexit referendum (Brexit heterogamy), which might not align with party-based classifications.

Background

Partnerships and Political Views

This study examines the consequences of homogamy and heterogamy that have not been studied in connection to union dissolution: partners’ political preferences. To understand the possible implications of heterogamy by political preferences, it is worth first discussing the reasons partners tend to share similar characteristics.

The processes that produce assortative mating can be divided into two broad categories: preferences and constraints (Huber and Malhotra 2017; Kalmijn 1998, 2005; Lichter and Qian 2019; McPherson et al. 2001; Schwartz 2013). First, individuals select partners on the basis of shared interests or characteristics, a process often driven by personal preferences (Kalmijn 1998; Lichter and Qian 2019). A preference for politically similar partners might reflect a general tendency to prefer similarity for any given personal characteristic (Kalmijn 1998; Luo 2017; Schwartz 2013). In this view, political homogamy might mirror a preference for similar others, just as with other social identities for which sorting is widespread, such as ethnicity (Fu and Heaton 2008; Wimmer and Lewis 2010) or education (Skopek et al. 2011).

Second, individuals choose their partners from a pool of similar candidates because of preexisting homogeneity in their social environment, which is influenced by the constraints of the marriage market (Kalmijn 1998). For instance, individuals might form unions with like-minded partners because of politically segregated social networks (Liben-Nowell et al. 2005). Also, individuals might form unions with partners with whom they share other characteristics, such as religion or ethnicity, which are correlated with political views (Anderson et al. 2014). In this respect, shared political beliefs could be a by-product of other commonalities (Huber and Malhotra 2017).

Finally, partners might influence each other's views while dating and living together (convergence) because political attitudes are malleable over time (Arránz Becker and Lois 2010). In romantic relationships, partners influence each other's political beliefs because of similar interests (e.g., sharing a socioeconomic condition orients support for specific political instances), common environments (e.g., living in the same area leads to sympathizing with similar communities), or persuasion because partners are the most frequent targets of political discussions (Daenekindt et al. 2020; Stoker and Jennings 1995, 2005). Partners’ attitudes might also converge over time because partners experience the same life events, which shape their political beliefs (Huber and Malhotra 2017).

Literature on political homogeneity identified more congruence between household members than among other social aggregations. Spouses tend to become politically like-minded not only because of the selection process that brings them together but also through socialization (Johnston, Propper et al. 2005; Klofstad et al. 2012; Nickerson 2008; Stoker and Jennings 2005). In non-UK (Hersh and Ghitza 2018; Zuckerman et al. 2005) and UK contexts (Bélanger and Eagles 2007; Johnston, Jones et al. 2005; Lampard 1997), estimates of political homogamy regarding party preferences in two-voter households range from 40% to 75%.

Recent research has challenged the idea that spouses only reinforce each other's beliefs, indicating that it is common for people to hold different political views (Daenekindt et al. 2020; Kan and Heath 2006). Individuals who are politically engaged tend to discuss politics simply because they enjoy political discussions (Huckfeldt and Mendez 2008), even if it leads to disagreements. However, research in political psychology suggests that political disagreements can be uncomfortable for some, leading them to steer clear of situations where such differences are evident (Huber and Malhotra 2017; Mutz 2002), especially in close social relationships (Levinsen and Yndigegn 2015). It is therefore plausible that couples who do not share political views might experience tension, potentially putting their relationships at risk (Arránz Becker and Lois 2010).

Couples’ Heterogamy and Union Dissolution

Our study contributes to the social science literature on the consequences of couple heterogamy. This research has generally observed that partners’ similarity is associated with higher quality relationships and a lower risk of union dissolution. Specifically, couples with differences in race, ethnicity, nationality, education, or religion—referred to as heterogamous couples—are more likely to experience union dissolution, a pattern that is more pronounced in dating and cohabiting relationships than in marriages (e.g., Blackwell and Lichter 2004; Hwang et al. 2021; Wang et al. 2006). Partners’ similarity on personal characteristics, such as personality and attitudes, is often considered crucial for relationship satisfaction and longevity. However, the evidence for this idea is varied, with some studies supporting it and others presenting null findings (e.g., Luo and Klohnen 2005).

The relationship between political agreement among partners and their union stability has not been thoroughly investigated. Limited research, predominantly from the United States, has explored how couples’ political views correlate with their relationship quality or some of its proxies. Wang (2020) documented reduced happiness in relationships between Republican and Democratic partners relative to those in which both partners are Republican. Wilcox and Wolfinger (2015) reported greater marital satisfaction among Republican voters than Democratic voters. Similarly, Wolfinger (2017) and Twyman (2016) found that right-wing voters in the United States and the United Kingdom claim to have more frequent sexual activity than their left-wing counterparts. This related literature, particularly from the United States, has predominantly examined the relationship between individuals’ political views and relationship quality (e.g., Fangmeier et al. 2020) and has not directly investigated the association with union stability. However, if individuals’ political views influence relationship quality, political similarity might influence union longevity. Still, previous research has not directly assessed the association between political heterogamy and union dissolution. Our study aims to fill this specific gap by quantitatively assessing the risk of union dissolution in the context of partners’ political heterogamy, expanding on research focusing more on subjective measures of relationship quality.

According to the assortative mating literature (Kalmijn et al. 2005; Zhang and Van Hook 2009), two main mechanisms explain the positive association between heterogamy and union dissolution. The cultural distance argument posits that differences in religion, ethnicity, and other characteristics imply a divergence in tastes, values, and communication styles (Kalmijn 1998). Such differences are an obstacle to the formation of shared daily routines and decisions (e.g., child-rearing) and reduce mutual understanding, shared interests, and intimacy (Mahoney et al. 1999; Waite and Lehrer 2003). Therefore, heterogamous couples are more prone to misunderstandings and conflicts. Conversely, homogamous couples likely face lower dissolution risks because of shared traits that make decision-making and life together smoother, ultimately enhancing satisfaction and reducing stress.

A second mechanism explaining why couples might be more likely to split up if the partners have different social characteristics relates to the social boundary argument. Forming a union with someone with a different religion, ethnicity, nationality, or social class implies crossing a social boundary in society. Relationships outside the group are often normatively disapproved, leading heterogamous unions to receive less practical and emotional support from the social networks of each partner relative to other couples (Hohmann-Marriot and Amato 2008; Killian 2001).

Both previous theoretical arguments lead to the same prediction: heterogamy makes union dissolution more likely because of a higher probability of conflicts as a result of cultural distance and less support from the extended family or social network. The aforementioned theoretical arguments applied to heterogamy by political preferences lead to the expectation that partners with different political preferences are more likely to split up than partners with the same political preferences (Hypothesis 1).

Political Heterogamy and Union Dissolution Within the UK Political Context

In the United Kingdom, divorce rates peaked in the early 1990s and then gradually declined (Office for National Statistics 2012b, 2020). The risk of marital dissolution within five years of marriage is roughly 8%, approaching 20% within 10 years. Cohabiting unions are less stable, with one in three ending within five years and roughly 40% dissolving by the 10th year (Office for National Statistics 2012a).

Here we ask, What is the role of partners’ political preferences in union survival in the United Kingdom? The country's political system, which is divided into constituencies that elect the most popular candidate, traditionally benefits the two main national parties: the right-leaning Conservatives (also called Tories) and the left-wing Labour Party. The centrist Liberal Democrats (LibDems) saw a rise in vote share until 2010, when they formed a coalition government with the Conservatives, which lasted until 2015. Despite the presence of other parties, such as regional parties and the far-right UK Independence Party (UKIP), the political landscape in the analysis period of 1991–2019 has largely been a Tory–Labour binary.

Recent UK surveys suggest a shift in negative attitudes toward politically mixed partnerships. Ibbetson (2019) reported that the share of Labour voters who would disapprove of their child marrying a Conservative supporter had doubled from 2008 to 2019 (39% vs. 19%; Khomami 2016). Additionally, another recent poll documented that Labour supporters were less open to date Conservatives than the other way around (35% vs. 49%; Ibbetson 2021). Notably, 40% of individuals who voted against the UK leaving the European Union in the 2016 Brexit referendum expressed disappointment at the prospect of their child marrying someone who voted in favor, a sentiment similar to the rate of Labour–Conservative marriage disapproval (39%).

The cultural distance and social boundary arguments presented earlier further lead to the expectation that the bigger the cultural distance or the stronger the boundary between the two groups represented in the couple, the greater the risk of union dissolution (Zhang and Van Hook 2009). Thus, unions between Conservatives and Labour supporters, who are ideologically distant (Kan and Heath 2006), are hypothesized to be at a greater risk of ending than unions between the Liberal Democrats and either of the other two parties (Hypothesis 2).

In June 2016, the United Kingdom underwent a major political event: the Brexit referendum, which decided its continued permanence in the European Union (EU). This referendum, favoring the departure from the EU, caused the emergence of a new political cleavage and significantly influenced political party platforms in subsequent years (Hobolt et al. 2021; Sanders 2017).

Public sentiment toward the EU fluctuated considerably between 2004 and 2016, with support ranging from a high of 52.3% in June 2005 to a low of 34.7% in June 2011 (Janmaat et al. 2018). In recent years, the number of people who strongly identify with political parties has declined. Instead, studies have shown that a growing share of the British population has adopted a strong Brexit identity (Curtice 2018; Duffy et al. 2019; Hobolt et al. 2021). This shift signifies that Brexit sentiments have transcended traditional party lines, catalyzing a distinct political divide, which motivates our decision to include opinions on Brexit as a potential indicator of homogamy and heterogamy within couples. The social boundary argument might be especially applicable in specific political contexts where opinions are publicly shared and debated, such as Brexit. Therefore, this backdrop leads us to expect a greater risk of dissolution for couples with different views on Brexit (i.e., Remain–Leave), as opposed to partnerships with aligned views (Hypothesis 3).

Method

Data and Analytical Samples

In this study, we employ a discrete-time event-history logit model to analyze the association between political heterogamy and union dissolution. This model allows us to assess the timing of union dissolutions and incorporate time-varying covariates, providing a robust framework for examining the dynamics of relationship stability over time.

Our analyses are based on two British annual, face-to-face, population-representative surveys that provide data on individuals and households: the BHPS, covering 1991–2008; and the UKHLS, for 2009–2019. These surveys focus on aspects such as household composition, labor market participation, and other economic and sociological aspects, including political opinions and voting behavior.

We compiled two couple-year datasets: the first, for the analysis of party heterogamy (and its refined version), includes 28,173 heterosexual couples in 1991–2019; the second, focusing only on opinions on Brexit as gathered in Wave 8 of the UKHLS, consists of 14,857 heterosexual couples interviewed from 2016 to 2018. We consider only couples in which both partners are 18 or older and at least one partner has appeared in the survey for at least two successive waves.

Start dates for unions are measured by the year the couple began cohabiting or got married, whichever came first. Union duration is calculated from this start date to either the year of union dissolution or the end of the observation period. Each household member is individually interviewed annually to gather comprehensive data on their sociodemographic characteristics, political preferences, and union status. Note that proxy interviewing is not employed; individuals directly provide their responses to ensure accuracy and reliability.

Dependent Variable

Our outcome is the dissolution of cohabiting or marital unions, which we identify through annual reports from respondents about whether their partner is present in the household. A couple is classified as separated if one partner leaves the household. If a partner dies or both individuals drop out of the survey, the couple is not considered separated, and the data are treated as right-censored.

Explanatory Variables

We gauge respondents’ political preferences through their party support using three widely accepted questions in the UK context (Tilley 2015). The first asked whether respondents consider themselves to be “supporter[s] of any political party.” Those who said no were asked whether they think of themselves as “a little closer to one political party than to the others.” Those who answered no again were asked, “If there were to be a general election tomorrow, which political party do you think you would be most likely to support.” In essence, these questions are proxies for voting behavior in years with no elections. In election years, these questions can be complemented by the question, “Which party did you vote in the last elections?” For example, individuals interviewed in 2010 (an election year) might be asked to recall their vote in May and the party they supported in the autumn. When questions on both political preference and vote were asked, we follow Tilley (2015) in giving preference to voting intention (although we run a robustness test with the alternative specification).¹ For instance, 90% of respondents who voted Labour in May 2010 reported supporting Labour later in the year. The equivalent percentages for the Conservatives and Liberal Democrats are 89% and 84%, respectively.

Our key independent variables focus on political preferences. Following Huckfeldt et al. (2004), we define political homogamy (heterogamy) as partisan agreement (disagreement) as self-reported by each partner, reducing the potential biases in respondents’ perceptions (Foos and de Rooij 2017; Frödin Gruneau 2020). In this framework, accord in political views is achieved when two people—a couple in our analysis—express the same preference for the same party. This approach registers an absence of agreement when partners’ preferences differ but does not quantify the degree of disagreement.² For instance, two partners favoring Labour and the LibDems, respectively, might be ideologically closer than those supporting Labour and the Tories. However, both couples are categorized as heterogamous in our first operationalization.

First, Party heterogamy identifies whether partners have homogamous (same) or heterogamous (different) party views (homogamy = 0, heterogamy = 1). To avoid dropping observations from our analyses, we also account for residual couple types. Second, Party heterogamy–refined delves into the specifics of party-level homogamy. Our sample covers England, Scotland, Wales, and Northern Ireland, which display a variety of political parties. We separate regional parties because of their association with specific national identities with the United Kingdom's smaller countries. For clarity, in the second specification, we consider three main parties (Conservatives, Labour, and Liberal Democrats) and categorize all other parties as “Other.” The resulting categorical variable captures party homogamy (both partners are Tory, Labour, LibDem, or Other), heterogamy (couples are Tory–Labour, Tory–LibDem, Labour–LibDem, or Other mixed), and residual couple types (with at least one nonaffiliated partner or with missing values on party preference). The number of couple-year observations and couples for each category are shown in Table 1.

Third, Brexit homogamy is based on respondents’ answers to the following question in Wave 8 (2016–2018): “Should UK remain in the EU?” This variable reflects partners’ homogamous (Remain or Leave) or discordant (Remain–Leave) views on Brexit. Also, in this case, specific categories are created to address missing responses and individuals with no opinion on Brexit. The number of couple-year observations and couples for each category are shown in Table 2. Information on Brexit opinions was not available in Wave 9, and we do not use data from Wave 10 because, at the time of writing, we could not observe union dissolutions after that point, given that it marks the end of our observation period.

Figure 1 displays the prevalence of couple types according to the first two classifications, and Figure 2 displays the prevalence for the third one. As shown in Figure 1, roughly a third of couple-observations are categorized as homogamous (different types of homogamous couples are in different shades of red) based on the first same-party criterion, and 13% are categorized as heterogamous (different types of heterogamous couples are in different shades of blue). Our second measure of homogamy, which captures specific party affinities, shows that Labour-affiliated couples are the most prevalent homogamous couple, representing 14.7% of all observations. Among couples with different party preferences, Tory–Labour couples are the most prevalent, representing 3.4% of all couples and 26.8% of couples within the group of heterogamous couples (Figure 1). Regarding Brexit homogamy, 14% of all couples reported opposing opinions on Brexit (Figure 2). Tables S1 and S2 (shown in the online appendix, along with all other tables and figures designated with an “S”) report the yearly transition probabilities for two of our main explanatory variables (Party heterogamy and Party heterogamy–refined). Political homogamy shows some stability: couples reporting the same-party preference at one point have approximately a 70% likelihood of maintaining the same status in the following year (Table S1), especially for Tory and Labour couples (Table S2).

Control Variables

Previous empirical studies have examined the determinants of party identity (Dassonneville 2016; Evans and Tilley 2012) and opinions on the Brexit referendum (Alabrese et al. 2019; Becker et al. 2017; Hobolt 2016). Older individuals and those with less education, with lower skills, and living in poorer households are more likely to support “Leave,” whereas young individuals and women are more likely to favor “Remain.” Consequently, we control for these factors, which are also determinants of union dissolution (Boertien and Härkönen 2018; Matysiak et al. 2014). Specifically, we use information on both partners’ age, education, and occupational class to construct indicators of homogamy and heterogamy (Grow et al. 2017). Likewise, we consider ethnicity and religiosity, dimensions frequently examined in previous studies on union dissolution (Wong 2016; Wright et al. 2017). Considering these factors enables a comparative analysis of the associations of the newly considered heterogamy regarding political preferences against previously studied dimensions. Following previous studies, we control not only for homogamy and heterogamy in certain dimensions but also for the levels of these variables.

Specifically, the baseline control variables (in Model 1) include both partners’ age (linear and quadratic) and birth cohort (five-year groups); a function of union duration with a linear, a quadratic, and a cubic term; and two partnership characteristics, the presence of children in the household and marital status (cohabiting vs. married). We also considered an alternative specification of union duration (using a set of dummy variables) and obtained very similar results. We then complement the baseline specification with a stepwise approach, adding measures of partners’ homogamy on dimensions other than political preferences: partners’ age gap (up to two-year age gap [reference], female partner is two or more years older, or male partner is two or more years older) in Model 2; the highest level of education between the partners (lower secondary or less [reference], upper secondary, or degree or higher) and educational homogamy (same education [reference], male partner is more educated, or female partner is more educated) in Model 3; ethnicity homogamy (same ethnicity [reference] or different ethnicity) in Model 4; religion homogamy (same religion [reference] or different religion) in Model 5; and occupational class (NSSEC-5: management and professional [reference], intermediate, small employers/own account, lower supervisory/technical, semi-routine/routine, or both partners are out of the labor force) and occupational homogamy (same job class [reference]; both partners working, male partner's occupational class is higher; both partners working, female partner's occupational class is higher; female partner is out of the labor force; or male partner is out of the labor force) in Model 6. Missing cases are also reported in flag categories for each measure of homogamy. All regressors are measured with a one-year lag. Table S3 describes the control variables at couples’ survey debut by partnership outcome and analysis sample.

Model

We estimate random-effect, discrete-time event-history logit models. The model is represented as follows:

where Y is a dichotomous indicator for the union status of a couple j (0 = intact, 1 = dissolved) at time t, and P(Y)_jt is the probability of a union separation during the interval (t, t + 1). The term t represents the time in the union, and γ(t) is a function of time elapsed since union formation up to the cubic term. Our specification links the probability of dissolution in the interval (t, t + 1) with independent variables measured at t. X_jt is a vector of covariates that can vary across unions and over time; ν_j captures unobservable couple-specific characteristics assumed to be uncorrelated with the independent variables, normally distributed with a mean of 0 and a variance that is estimated; and ε_jt is the idiosyncratic error term. All event-history models consider right-censoring of data.

Although a fixed-effects model would allow for the control of time-invariant confounders, we opted for a random-effect model because of the stability observed in certain couple types, particularly in the political party combinations of partners, over time. This approach exploits differences between couple types rather than only within-couple changes, which is appropriate given the relatively high stability in political affiliations over time (see Tables S1 and S2). Nonetheless, we estimated fixed-effects models, which address whether changes in the concordance of spouses’ political affiliation are associated with dissolution, as a robustness check (see the Additional Analyses and Robustness Checks section).

To enhance the interpretation of the substantive significance of the results, we report estimated odds ratios in the tables and graphically display predicted yearly probabilities of separation by couple types. The predicted probabilities are calculated using the model's estimated coefficients applied to the observed values for the control variables. This approach involves computing individual predicted probabilities and then averaging them, which provides a summary measure of the effect of the independent variable on the probability of the outcome, averaged across all observations in the sample (Williams 2012). Predicted probabilities are presented graphically with confidence intervals for pairwise comparisons. Following methodological studies (e.g., Goldstein and Healy 1995), we center the confidence intervals on the estimated average predictions with an 84% confidence level. Doing so allows us to test statistical differences between any pair of predictions at the 5% level, analogous to conducting a t test for specific pairwise comparisons. Nonoverlapping confidence intervals suggest significantly different predictions, whereas overlapping confidence intervals indicate no significant difference between the predicted values.

Results

Tables 1 and 2 report the incidence of union dissolution by party and Brexit homogamy, respectively. Couples with the same party preferences have a yearly incidence of dissolution of 0.71% (Table 1). This rate is 0.97% when partners belong to different parties. Among the three most common homogamous combinations of party preferences (Tory, Labour, and LibDem), the incidence of separation does not exceed 0.84%. Heterogeneous political combinations exhibit separation incidences ranging from 0.94% to 1.08%. Notably, Labour–Tory couples do not have the highest rates of divorce, which is inconsistent with Hypothesis 2. The incidence of separation in relation to Brexit opinions, as recorded in Wave 8 of UKHLS (Table 2), ranges from 0.87% (partners both adopt a Remain opinion) to roughly 1.8% (couples with Remain–Leave and Leave–Don't know opinions). These figures suggest that couples with heterogamous political beliefs are more likely to separate.

Party Heterogamy and Union Dissolution

Here, we discuss the association between party homogamy/heterogamy and union dissolution. The complete estimates of the six event-history logit models that vary for the set of controls are reported in Table S4. Given that the estimated associations of party heterogamy are very similar across the several specifications, we focus on Model 6, which includes all controls. Couples with the same party preference are less prone to breaking up than those with heterogamous preferences (different party), supporting Hypothesis 1. Specifically, estimates from the fully adjusted model (Model 6) indicate that the odds of union dissolution for both types of politically heterogamous couples are roughly 39% higher than for couples who prefer the same party (Table S4). Results from the full model are represented graphically in Figure 3, which shows the yearly predicted probability of union dissolution. The predicted probability of separation for couples with homogamous political preferences (same party) is slightly below 0.8% per year, which is significantly lower (p < .05) than that for politically heterogamous couples (1.1%).

The pattern for other measures of partnership homogamy/heterogamy is also in the expected direction: heterogamy is always associated with a higher risk of union dissolution (Table S4). Notably, the odds ratios for political heterogamy are similar to or bigger than those for other established forms of heterogamy. For example, the odds of dissolution for couples with different religious beliefs are 34% higher than for those sharing the same religious views. Heterogamous couples by education face higher odds of union dissolution (17% or 19% higher, respectively, when the male or female partner has the highest education) than homogamous couples. Figure S1 shows the predicted probability by groups based on dimensions of heterogamy other than the political one, confirming that heterogamous couples tend to be at higher risk of dissolution than homogamous couples. For example, heterogamous couples by ethnicity or religiosity exhibit predicted annual dissolution probabilities of roughly 1.1%, paralleling values found for politically heterogamous couples.

Party Heterogamy–Refined and Union Dissolution

We now investigate in more detail the combinations of party preferences (full estimates are presented in Table S5). Heterogamous couples (Tory–Labour, Tory–LibDem, Labour–LibDem, and Other mixed) are statistically more likely to separate than the homogamous Tory couples (Table S5), again supporting Hypothesis 1. Because of the more granular classification, the estimates in Figure 4 have wider confidence intervals than those in Figure 3, leading to several pairwise comparisons that are not statistically significant (see Table 1 for the category-specific number of observations). Nevertheless, Figure 4 confirms that all types of heterogamous couples are at higher risk of dissolution than all types of homogamous couples.

The estimated odds ratio of union dissolution is greater for Tory–LibDem than for Tory–Labour (odds ratio = 1.54 vs. 1.38 in the fully adjusted Model 6; Table S5), but a direct comparison between these two groups reveals no statistically significant difference in the probability of union dissolution (Figure 4). In fact, predicted yearly probabilities of union dissolution are similar across all types of heterogamous couples, and differences are never statistically significant. These findings do not support Hypothesis 2, which expected the highest risk of dissolution for Tory–Labour couples.

Among homogamous couples, Labour couples exhibit a marginally higher likelihood of dissolution relative to Tory couples, but this difference is statistically significant only at the 10% level in the fully adjusted model (odds ratio = 1.19; Model 6, Table S5). The probability of union dissolution for Tory couples does not statistically differ from that of homogamous LibDem and Other party couples.

Brexit Homogamy and Union Dissolution

Finally, we address the role of homogamy on Brexit views on union dissolution. The analysis focuses on a subset of couples who disclosed their opinions on Brexit in Wave 8 of the UKHLS. We estimate the transition to union dissolution over the subsequent waves using the same set of control variables as in the earlier analyses. Consistent with the previous findings, partners’ concordance on Brexit is associated with lower odds of union dissolution, corroborating Hypothesis 3 (full estimates are reported in Table S6). The influence of Brexit heterogamy on dissolution is markedly pronounced. Specifically, the odds of dissolution are 2.3 times as high for couples with opposite Brexit opinions (Remain and Leave) than for Brexit-homogamous couples (Table S6). Homogamous Remain and Leave couples display a similar predicted probability of union dissolution, which is as low as 1.1% annually (Figure 5). Couples reporting divergent views on Brexit (Remain–Leave) lie substantially above this rate, with a predicted probability of dissolution of 1.8% per year. Note that Table 2 shows that Leave couples have higher dissolution rates. This finding indicates that controlling for other variables—such as education, which is lower among Leave voters and associated with higher dissolution rates—is crucial for accurately isolating the association between Brexit opinions and union stability.

The association of Brexit heterogamy with union dissolution is substantial relative to other forms of heterogamy. Figure S2 shows the predicted probabilities of union dissolution by groups based on dimensions of heterogamy other than the political one, estimated on the Brexit sample. Couples in which the woman is older than the man display a considerably high yearly predicted probability of dissolution—as high as 2.8%. However, predicted probabilities for the other types of heterogamous couples are consistent with or smaller than for Brexit heterogamous couples.

The association of Brexit heterogamy also appears stronger than those of the two measures of party heterogamy. To better compare these associations, we implement additional regression analyses simultaneously testing the role of Party heterogamy (Table S7) or Party heterogamy–refined (Table S8) and Brexit heterogamy. These models are tested on the same set of couples used in the main analysis of Brexit from Waves 8–10 of UKHLS, ensuring consistency in sample selection. The results from these analyses indicate that the odds ratios associated with Brexit heterogamy are indeed larger in magnitude than those associated with any form of party heterogamy. For instance, the odds ratio for the Remain–Leave category shows a more pronounced risk of union dissolution than those for any combination of heterogamous party preferences, such as Labour–LibDem.

Results for Residual Couple Types

Our models also include categories for nonaffiliated members and for individuals with a missing value on party identification or no opinion on Brexit. Although findings for the resulting couple types are not central to our study, we briefly mention the most interesting estimates. Couples in which one partner supports a specific party but the other does not (nonaffiliated) and those in which both partners are nonaffiliated have a higher risk of union dissolution than homogamous couples (with both partners supporting the same party; Table S4 and Figure S3). Similarly, the Brexit analyses reveal that couples in which only one partner declares an opinion on Brexit have a higher risk of dissolution than homogamous couples (Table S6 and Figure S5).

Additional Analyses and Robustness Checks

We conducted several analyses to assess the robustness of the findings with respect to the consistency of couples’ types over time; gender composition in the Party heterogamy–refined measure; the use of an alternative measure of political heterogamy based on a left–right scale; measuring political heterogamy at the union formation; the inclusion of additional controls for personality traits and attitudes toward gender roles; the use of fixed-effects models; age, period, and cohort effects; and the inclusion of additional controls for both partners’ education levels and job classes. Results from all robustness checks confirmed the findings reported earlier. A detailed description of these analyses and the corresponding tables and figures is in the online appendix.

Discussion

This work contributes to the literature on assortative mating and partnership dissolution (Kalmijn 1994; Schwartz and Mare 2005, 2012) by being the first to examine the influence of a previously neglected aspect of heterogamy—political preferences—on union stability. Analyzing data on up to 29,000 British couples, we considered three indicators of political heterogamy related to party preferences and Brexit opinions. Our empirical approach demonstrated that couples in which partners hold different party preferences or divergent Brexit opinions are at higher risk of union dissolution than homogamous ones. The association between different party preferences and union dissolution is independent of the specific party considered: any combination of different parties tends to be (similarly) associated with a higher risk of union dissolution.

In our analyses, the implications of political heterogamy are important and sometimes exceed those associated with other types of heterogamy examined in previous research, such as differences in age, education, ethnicity, religiosity, and job class (e.g., England et al. 2016; Schwartz and Mare 2012; Wong 2016). These findings challenge prior evidence suggesting that political agreement was considered important for relationship success by only a small fraction of couples and ranked lower in importance than other homogamy dimensions, such as social background and religion (Lampard 1997). Our findings might reflect the possible underestimation of public perception of political issues for a couple's stability despite the high prevalence of observed political homogamy.

Our findings can be contextualized within the theoretical framework of assortative mating (Kalmijn 1998; Schwartz 2013). In line with the cultural distance argument, political homogamy can be considered a marker of partners’ cultural homogeneity, which is correlated with partnership stability independent of other proxies for partners’ homogamy. Partners with shared political values might have stronger bonds than partners with different views. Based on the social boundary argument, a second general mechanism for the associations of couples’ heterogamy with union dissolution, our results might suggest that individuals who partner with someone with different political preferences might suffer from social disapproval by family, friends, and the community. Compared with other, more visible forms of heterogamy (e.g., based on age or ethnicity), social disapproval might be less relevant for party preference, which is socially revealed only when individuals choose to. Although social disapproval might be a weaker determinant of political heterogamy than other forms of heterogamy, the Brexit referendum represented a unique sociopolitical event with markedly polarized and publicly debated political views. The heightened visibility and salience of political attitudes during this period could have amplified the social disapproval mechanism, explaining why Brexit heterogamy appears to be a more potent determinant of dissolution. Indeed, given that the political climate surrounding Brexit was characterized by a vocal public discourse and heightened divisions, couples with differing Brexit opinions might have experienced greater external pressure and conflict, increasing the risk of dissolution. Future research could explore these mechanisms in depth or investigate whether they concurrently influence union dissolution.

A key difference between political heterogamy and most of the previously examined forms of heterogamy must be acknowledged. Unlike time-invariant forms of heterogamy, such as those related to age and ethnicity differences, political preferences can shift throughout a union and adapt to those of the partner, potentially leading to a selected sample of couples when political heterogamy is assessed at any point after union formation (Arránz Becker and Lois 2010). A couple's political similarity might increase over time even if partners do not actively influence each other because they might respond to common experiences and shared environments (Stoker and Jennings 2005). Politically adapted and unadapted individuals might also differ in their personal attitudes, which could be the real factors of the observed associations of political heterogamy. Analyses that controlled for individuals’ personality traits and the differences between partners revealed a substantial robustness of our findings, which were also confirmed by analyses focused on political heterogamy measured at union formation. These findings support the idea that although political preferences might converge over time within a union, the role of political homogamy in union dissolution is consistent, whether assessed at the beginning of the union or after potential adaptive processes.

Our study showed that union survival can be associated with general divergences between partners’ political views as measured by party preferences. Negative associations with union stability can also be found when considering punctual markers of political cleavage, such as opinions on the Brexit referendum. Thus, political heterogamy is not limited to stable identities and long-standing values; it also aligns with emerging political discourses. Notably, we found the association of Brexit heterogamy to be more pronounced than heterogamy in party preference. This finding points to a demographically unintended effect of Brexit: it elevated the risk of dissolution for politically heterogamous couples.

Contrary to our expectations, we found that the risk of union dissolution does not vary significantly among various types of heterogamous couples (Tory–Labour, Tory–LibDem, Labour–LibDem). One possible explanation for this finding could be the nuanced and evolving differences in political culture and social identity between these parties and their supporters. Whereas Labour and Conservative parties have historically represented clear opposite ends of the political spectrum, historically defining the British political landscape, the Liberal Democrats often attract voters who might lean conservative on some issues and liberal on others (Sanders 2017). This ideological crossover could introduce unique interpersonal dynamics and conflicts not present in the more ideologically coherent Tory–Labour couples, potentially increasing dissolution rates.

Furthermore, increasing research has pointed to the changing nature of party politics in Great Britain, where social or cultural issues are becoming as important as economic issues in the political landscape (e.g., see Wager et al. 2022). Since 2010, the Tory party has experienced a disconnect between its neoliberal Members of Parliament and more centrist voters on economic issues. Simultaneously, the Labour Party sees disagreement on cultural issues between its socially liberal Members of Parliament and more authoritarian voters. Party competition has shifted away from class-based and economic cleavages and toward cultural and social issues (British Election Study Team 2016). These changes reflect long-term socioeconomic developments, including the increasing relevance of education, increasing geographic polarization, and more diverse populations. Thus, whereas the Tory party has moved leftward on economic values and become more authoritarian on social values, Labour has moved rightward on economic issues and become more socially liberal. The convergence of the Conservative and Labour parties on certain ideological fronts might make it challenging to predict union dissolution based solely on party preferences. Historically, the Labour and Conservative parties have represented clear-cut left-wing and right-wing ideologies, respectively. However, this distinction is increasingly blurred (Duffy et al 2019).

Additionally, the rise in importance of the Brexit debate has further changed the political landscape. The Liberal Democrats have consistently maintained a pro-European stance, diverging from Conservative and Labour positions, which have also varied over time. This variation was especially true during Jeremy Corbyn's leadership of the Labour Party, when the party's stance on Brexit was often considered ambiguous or indecisive. Such a dynamic scenario can make the EU debate a more salient factor in couple dynamics than traditional party support.

These shifts make assessing the associations of different forms of political heterogamy on union dissolution complex. With Brexit emerging as a divisive issue, it is plausible that differences in opinions on the EU could pose a greater risk to union stability. In this context, couples in which only one partner manifests support for the pro-EU Liberal Democrats could potentially face more conflict because the EU issue cuts across traditional party lines (Hobolt et al. 2021), introducing a new axis of political disagreement that could be more impactful on relationship dynamics.

Our analysis also included a comparison of the risk of dissolution for couples classified as “residual,” whose political alignment did not fit neatly into our primary categories of homogamy or heterogamy. The findings revealed that these residual couples had a higher dissolution risk than homogamous couples but generally had a lower dissolution risk than heterogamous couples. Thus, although residual couples do not benefit from the stabilizing effect of shared political alignment seen in homogamous couples, they also do not face the same level of instability associated with pronounced political differences observed in heterogamous couples. These results highlight the complexity of political alignment within relationships, suggesting that the diversity within the residual category leads to a moderate impact on union stability. This finding underscores the importance of considering various degrees and types of political alignment when assessing their influence on union dissolution.

All in all, measuring political heterogamy among couples based on party identification might not adequately capture potential conflicts among partners arising from politically divergent views. Confirming this possibility, our additional analyses based on the distance between partners’ self-placement on the left–right scale supported the prediction that couples with more politically distant partners are at higher risk of union dissolution.

Our analyses revealed additional findings that deserve future exploration. The finding that couples with one politically oriented partner and one without any affiliation or preference show a greater risk of dissolution than politically homogamous couples might reflect the role of political engagement rather than just political stance. A lack of affiliation or support might signify political apathy, which could be a source of conflict when such individuals are paired with a politically engaged partner, leading to instability. Additionally, politically nonaffiliated individuals might lean toward antisystemic political attitudes, which could also be a cause of conflict in a romantic union with a person affiliated with a systemic party. Unfortunately, the data did not allow us to distinguish among the different reasons for nonaffiliation or lack of support. Future studies could examine in more depth those couples in which one or both partners are nonaffiliated. For instance, union dynamics and stability might vary depending on whether partners intentionally sought a partner with similar political views (preferences) or ended up with a politically similar partner by coincidence (constraints). Understanding whether intentional political alignment (preferences) versus coincidental political alignment (constraints) has distinct effects on union stability could provide deeper insights into the mechanisms driving these associations.

Our analysis also revealed that couples with one member lacking a stance on Brexit face a higher risk of dissolution than those in which partners share the same Brexit opinion. This result might be attributable to the social significance of Brexit, which transcended traditional political boundaries and became a litmus test of social and cultural values. Apathy or avoidance in expressing opinions on Brexit might signal a broader disengagement from public debate and critical social issues, contributing to partnership strain.

Our study has limitations. First, it is important to consider the potential role of selection bias in the associations we found. Individuals might form unions with partners who do not share their political views owing to unobserved traits that might not be desirable in the marriage market. For instance, certain personality traits or social circumstances might lead individuals to cast a wider net, resulting in partnerships displaying political heterogamy. These unmeasured traits could also be associated with a higher risk of union dissolution, suggesting that it is not only political heterogamy per se but also these underlying factors driving the observed associations. Future research should aim to disentangle these effects to better understand the impact of political heterogamy on union stability.

Second, the data did not allow us to measure the strength of party identification in the main analyses. However, a robustness check with a subsample with available information on the self-positioning on the left–right scale confirmed the main results. Future studies can use different data to further corroborate our findings.

Future studies should explore the conditions under which political heterogamy shapes union dissolution. For example, this relationship might vary across historical and political periods, particularly during times of heightened political polarization or significant events, such as elections or referenda. Further, a country's political system can influence this relationship. Political heterogamy might be less impactful in multiparty systems (e.g., in the Netherlands) than in two-party systems (e.g., in the United Kingdom), where political differences are more pronounced. Additionally, other salient political issues, such as congruence on COVID-19 policies, could also influence union stability. Future studies should also investigate the mechanisms underlying the political heterogamy effect, such as the role of political engagement levels, political conflicts, sharing of political activities, and specific values.

Despite these limitations, our study contributes to the growing field of political demography (Goldstone et al. 2012), which has focused on the macro- and micro-level interrelations between demographic and political changes (Arpino and Mogi 2024; Mogi and Arpino 2022; Sommer 2018; Vogl and Freese 2020). In particular, our study offers new perspectives on how political and demographic dimensions can be interrelated. Studies that relate union dynamics and political polarization have mainly explored the dynamics of relationship sorting (or partnership formation; Anderson et al. 2014; Huber and Malhotra 2017). However, we showed that political sorting also continues throughout the relationship. Couples who cannot or choose not to reconcile their political differences could be at higher risk of separating over time. Thus, political sorting is apparent not only in the formation of romantic partnerships (e.g., Anderson et al. 2014) but also in their dissolution, suggesting that contemporary trends of family dynamics might strengthen polarization along the political dimension. Therefore, political homogamy could be another source of partnership sorting alongside age, social class, education, and ethnicity.

Acknowledgments

The authors are grateful for the helpful comments received from Francesco Billari, Danilo Bolano, Gøsta Esping-Andersen, Juho Härkönen, Selin Köksal, the attendants at Dondena seminar and reading groups, Bocconi University; Nicola Barban, Raya Muttarak, and attendants at GENPOP/POPCLIMA seminar series, University of Bologna; and participants at the 2019 Divorce Conference in Florence and the 2021 annual meeting of the Population Association of America. This research was conducted within four projects: CRITEVENTS (Critical Life Events and the Dynamics of Inequality), financed by the NORFACE network; the NCCR LIVES (Overcoming Vulnerability: Life Course Perspectives), financed by the Swiss National Science Foundation; and BIOSFER (101071773) and DisCont (Discontinuities in Household and Family Formation), both financed by the European Research Council.

Notes

References