Enhancing, suppressing or something in between – loneliness and five forms of political participation across Europe

In the last few years, the topic of rising loneliness in western democracies frequently made headlines and got framed as an epidemic at times, which sparked a debate whether politics should tackle the issue of loneliness more directly (Hafner 2016; Easton 2018). In consequence, the interest in research with a particular focus on causes and the long-term consequences of loneliness for mental and physical health grew rapidly (Holt-Lunstad et al. 2010; Leigh-Hunt et al. 2017). However, other outcomes of social interest received less attention and comparatively little is known about the consequences of rising loneliness for outcomes such as social and political participation, social cohesion or policy compliance. This is problematic for two reasons. First, a too narrow focus on the consequences of loneliness for a society underestimates its relevance for policy makers which might partially explain why governments just rarely take concrete action to fight societal loneliness systematically. Secondly, although loneliness is known to exert substantial influence on cognition and behavior, it is rarely subject of sociological and political research and most explanatory models do not account for this dimension of social reality. While a large body of research in political and social sciences investigated the structural component of social embeddedness such as group membership or social influences such as peer pressure extensively, loneliness is yet to be investigated (Smets and Van Ham 2013; Blais et al. 2019). Especially the well-documented tendency of lonely individuals of social withdrawal and social anxiety relates to themes such as political and social participation and suggests a high potential for our explanatory models (Hawkley and Cacioppo 2010; Qualter et al. 2015).

This study aims to fill this gap in the literature by investigating the link between perceived loneliness and the political engagement of citizens in Europe. The study contributes to the literature in two ways in particular. First, by reviewing the relevant literature I derive a bidirectional re-affiliation model of political action, stating that lonely individuals are increasingly likely to participate in political actions if those forms of political actions have a high potential for social re-affiliation. In contrast, political acts that offer little potential are expected to become less likely. With this, the paper offers the first comprehensive framework explaining how perceived loneliness relates to political action. Secondly, by utilizing the European Social Survey, hereafter ESS, I test the proposed relationship for reported voting behavior, participating in demonstrations, signing petitions, working for a political group and contacting politician. By testing the model in a representative, multi-national setting for five prominent forms of political participation, the study highlights that loneliness is a fruitful field of investigation for upcoming studies.

What motivates civic political participation? Discussed in the context of voter turnout, Harder and Krosnick review the economic choice model of political action. From this perspective, the probability to vote can be formalized as a function of how difficult it is to participate, the individuals` motivation to participate, and the ability to enact the participation (Harder and Krosnick 2008). While this framework originated in the literature concerned with voting behavior, other forms of political action are based on similar principles. Given a sufficient motivation, resources and acceptable costs of a political action, a person should be likely to participate (Harder and Krosnick 2008; Huddy 2013).

While loneliness is so far not considered in the political participation literature, other network characteristics are a frequent subject of investigation. Therefore, it is helpful to briefly differentiate between loneliness and social embeddedness in order to clarify how both concepts impact the motivation or ability to participate in different ways. Objective measures of social networks such as memberships in organizations or positions within a network are common characteristics in the political participation literature and are, in combination with interactive measures such as political conversations or peer pressure, common variables to measure social capital (Bhandari and Yasunobu 2009). Authors investigating the link between such measures of social networks and political action often argue that the social network provides resources such as information, material resources, social support and social control. These resources can be utilized in various ways: for instance, information might boost citizens’ ability to make informed decisions, foster self-efficiency and increase motivation to participate in the democratic process. In contrast, social control and norms can increase the motivation to vote in order to prevent conflicts. Therefore, the mechanisms that link objective social embeddedness and political participation revolve around the idea that social networks channel collective assets the individual can utilize increasing the motivation or ability to participate.

In contrast, perceived loneliness is not a characteristic of the network but rather a subjective individual perception and is strongly influenced by the individuals’ evaluation and social forces such as social comparison and socialization (Perlman and Peplau 1981; Jong-Gierveld et al. 2006; Cacioppo and Cacioppo 2016). Consequentially, it cannot be operationalized with measures capturing frequency or type of social interaction (Russell et al. 2012). While most studies review loneliness as a unidimensional construct (Russell 1996), other authors have argued that social relationships fulfil various needs and depending on the kind of missing relationship, individuals experience loneliness differently (DiTommaso and Spinner 1997). However, an essential part of the loneliness experience is the wish to physically or emotionally connect with other individuals and, if not successful, negative psychological consequences such as passivity, social anxiety, distrust and low self-efficacy (Qualter et al. 2015; Spithoven et al. 2017). Correspondingly, two mechanisms explaining how loneliness relates to political participation can be derived from the literature.

The first suggests that political participation will become less likely if people suffer from loneliness due to passivity, alienation from society, distrust and lower self-estee. This expectation is founded in the accumulated evidence that loneliness has severe consequences for the individual’s perception of the trustworthiness of other people, onès own belonging to social groups and society as a whole, as well as the attitude toward self-efficacy (Spithoven et al., 2017). In line with this reasoning, recent correlative studies found evidence that loneliness is associated with a reduced sense of duty to vote and lower reported voter turnout (Langenkamp 2021). As numerous studies have shown that social identity, political self-efficacy and social trust are strong predictive factors for political participation, loneliness is likely to be negatively associated with political participation (Mieriņa 2014; Reichert 2016; Hadjar and Beck 2010).

The second hypothesized mechanism is based on the idea that political participation can be a potent setting for social reconnection. While I argued above that loneliness should exert an inhibiting influence on political participation, one might raise the question of whether loneliness is always related to a general withdrawal from political actions or whether it fosters some, more interactive forms of political participation. The recently revised re-affiliation perspective suggests that lonely individuals have a strong motivation to reconnect with other people (Qualter et al. 2015; Spithoven et al. 2017). Typically, social relationships have to be perceived as meaningful and lasting to reduce loneliness effectively. Therefore, I expect actions that enhance a feeling of belonging and provide a platform for prolonged interaction to be potential settings for lonely individuals to overcome their aversive situation. Research points to the conclusion that successful coping with loneliness is associated with the conscious reflection of the own situation, adopting of behavior and actively seeking for activities and contact. Thus, it is reasonable to assume that political actions are actively and consciously chosen settings for social reconnection (Rokach 1990; Vasileiou et al., 2019).

While all political activities are subject of social influence to some extent, some activities – such as voting – offer little opportunities for sustained or meaningful social interactions while others – such as participating in demonstrations – are more feasible for the goal of prolonged interaction and strengthening the personal belonging to the group (Mazzoni and Cicognani 2013). If the assumption holds true that lonely individuals have a strong motivation to re-affiliate with others in a persistent and meaningful way, and that collective actions are can be a useful mean to achieve this goal, loneliness should be positively associated with the probability of participating in such political actions.

To summarize, the two proposed mechanisms lead to an overall expectation that the relationship between loneliness and political participation is dependent on the potential of the political act for meaningful social encounters. On one hand, loneliness should decrease the probability of participating in political affairs in general due to the psychological consequences of loneliness summarized above. On the other hand, this effect might be weakened or even reversed depending on the potential of the political act for re-affiliation.

In order to investigate the proposed hypotheses, it is necessary to distinguish between forms of political participation that offer a strong platform for re-affiliation and those that are less feasible for such a goal. To be clear, while it is possible to socialize during most political actions, some are interactive by default and make contact very likely while others are more individualistic by concept. This study will focus on five of the most predominant forms of citizen’s political participation, summarized in Table 1. The chosen political acts cover institutionalized as well as non-institutionalized political participation that is exercisable in all considered countries (for potential exceptions, see the section ‘robustness tests’). With that, the analysis covers the most relevant political actions in Western democracies. Furthermore, the political acts offer different potentials for social interaction and, as argued below, can be categorized by their potential for social re-affiliation.

While there are many articles that propose possible classifications of political actions, they typically not discuss the dimension relevant for this paper. Rather, they differentiate the institutionalization of the political act, what person or group the action is supposed to address, whether the action enacted by citizens or other groups within a society such as lobbyists, or whether the action is voluntary (Theocharis and Van Deth 2018; Van Deth 2014). Therefore, this study cannot rely on preexisting classification schemes. Instead, I qualitatively assess the potential of the political acts to foster social re-affiliation based on their potential to (1) enable a person to interact with other individuals for a prolonged period of time and (2) foster a sense of belonging due to meaningful social interaction and to commonly expressed goals and attitudes.

Political demonstrations serve as a very illustrative example. The question what the primary motivating factors for citizens to participate in public demonstrations are was the subject of several qualitative and quantitative studies. The existing social–psychological models emphasize that the motives behind protest movements are manifold (Möller et al. 2009). Especially motives such as social identification and empowerment, a sense of belonging to the group, and the pursuit of shared goals are major motivational factors of participation in protest movements – online as well as offline (Jost et al. 2018). Qualitative studies illustrate that not just the development of new social ties, but also the preservation of intimate community ties, are relevant motives for political activists, an insight in line with the political mobilization literature (Mazzoni and Cicognani 2013). In addition, the literature concerned with recruitment for social movements highlights that informal networks such as friends and families do play an important role in the motivation to join a movement (Jasper and Poulsen 1995). In an investigation of the Fridays for Future movement, scientists confirmed that around 87% of the students protests together with friends, and lone protesters are the exception (Wahlström et al. 2019). Interestingly, this is not limited to friendly or neutral interactions. Studies investigating the Gezi Park protests pointed out that joining social movements can have a lasting impact by creating an atmosphere of exchange and tolerance between previously hostile groups (Budak and Watts 2015; Taştan 2013). Given that joining a social movement or protest does foster social contact, stimulates a sense of belonging and often thrives on political organizations that join the movement and offer a platform for prolonged social encounters, this form of political participation should appeal to lonely people in particular. In respect of the classification scheme, participants of political protests are continuously enabled to participate in the group activity over a substantial amount of time and get in touch with likeminded individuals.

Working in a political party or political action group can provide a space for purposeful interaction as well. While studies investigating motives of party membership are relatively rare, reported motivations do include the goal of sustaining or creating new social contacts. In a study from the 1970s, the scientists found that almost half of the members of a Canadian political party stated that they sustained party activity primarily for social reasons (Clarke et al. 1978. p.147). Similar patterns can be observed in a more recent German case where, alongside other motives, political parties offer the possibility to get in touch with friendly people which similar attitudes and interests (Laux 2011). In respect to other political groups, community service or volunteerism, the overall picture is the same (Holdsworth 2010; Sheldon et al. 2016; Stewart and Weinstein 1997). Consequently, political groups are considered a useful platform for prolonged and purposeful social interactions similar to participating in a protest movement.

While demonstrations and political organizations are classic forms of political engagement, signing petitions experienced increased attention in the last decades due to the rapidly growing frequency of e-petitioning through the internet (Jungherr & Jürgens 2010). In contrast to the first two discussed political actions, petitions provide a less obvious platform for social encounters. As summarized by Lindner and Riehm (2011), petitions fulfil mainly three functions for participants: they are an easy way to (1) express their attitude and influence politics, (2) protect their rights and interest and (3) mobilize others for their cause and increase the individual’s empowerment. While authors do mostly agree that signaling onès political support by signing a petition does foster an individual`s identification with the cause and the success of a petition is evaluated on multiple factors (Wright 2016), there is no evidence that signing a petition is used to foster social interactions in communities or social groups. Social networks are important for the distribution of a petition and mobilization of participants, but the duration of the political act is typically very brief and the act itself not interactive. No study that I am aware of mentions social interaction as key motivator for participating in petition campaigns. While absence of evidence does not prove the opposite, recent studies investigating digital petition platforms show that many contributors sign large numbers of petitions either out of habit or just spontaneously, while most other users remain inactive after just one petition (Puschmann et al. 2017; Halpin et al. 2018). This speaks against a considerable potential for social re-affiliation of lonely people. Rather, participants are motivated by factors such as expression of personal attitudes, empowerment, solidarity with a cause, and sometimes ideology (Wright 2016). Thus, for the purpose of this study, the act of signing petitions is considered as an unlikely context for social re-affiliation compared to protests or activity in political groups.

Another way citizens can influence politics directly is by contacting their elected representative or other relevant government officials. Despite the fact that it would be a short social encounter, contacting political representatives is unlikely to be used as a means to overcome loneliness. While studies have shown that preexisting ties between citizens and representatives are stronger predictors than individual factors (Aars and Strømsnes 2007), there is no in-depth study suggesting that the social interaction itself is a motivating factor for citizens’ to get in touch with political figures or government officials. In contrast to demonstrations and political groups, contacting a politician does typically not allow for sustained or repeated interaction, intimate encounters, or group identity. There is some evidence in other settings that people try to cope with their loneliness by using institutional contacts. For instance, in an interview study of exchange students in Australia, 8% of the students who reported that they felt lonely tried to cope with the situation by contacting student services as well as academic staff (Sawir et al. 2008). In another case, researchers found that loneliness in old age is a predictor for visiting one’s general practitioner despite being healthy (Ellaway et al. 1999). However, such cases are usually found in professions that are meant to deal with personal care or supervision and that can be contacted frequently, without institutional barriers. Given that contacting politicians is usually brief and infrequent, contacting politicians or government officials is considered to have a low potential for reconnection and is expected to become less likely for lonely individuals.

In respect to voting behavior, social forces play a major role in mobilization (Bond et al. 2017; Smets and Van Ham 2013). However, the empirical finding that peer pressure or social contacts motivate citizens to vote does not imply that the act of voting is considered as a useful setting by lonely individuals to revive their social ties. Studies investigating the intrinsic motivation of citizens to vote usually find motives such as fulfilling one’s civic duty, outcome preference due to personal believes or ethical reasons (Ali and Lin 2013; Galais and Blais 2016), but no personal incentives such as enjoyment of the group activity or the aim to socialize. Furthermore, voting is a rather short-lived activity and does not provide a basis for sustained interaction, either. Hence, voting is not considered a potent setting for individuals to socially re-affiliate.

This study investigates the link between loneliness and political participation in a multinational setting. This is important for two reasons. First, multinational datasets are useful to avoid statistical artefacts due to nation-specific characteristics such as differences in the political system, wealth or demographic compositions (Schmidt-Catran et al., 2019). Given the large variation in age composition and economic strength of the countries considered in this sample (Yang and Victor, 2011; Niedzwiedz et al., 2016), the methodological approach increases the reliability of the results by taking the contextual effects into account. At the same time, the sample consists mostly of European, liberal and democratic countries and attempts to generalize the results to other nations with other characteristics should be cautiously made. Second, loneliness as well as some of the considered political actions are comparatively rare events, and multinational samples are useful to avoid problems stemming from insufficient observations and related issues with statistical power. To analyze the derived hypotheses, the data from the third, fifth, sixth, and seventh wave of the European Social Survey (ESS) are utilized. The questionnaires of the other waves do not include a measure of loneliness and cannot be used in this analysis.

After deleting missing values on the individual level as well as the exclusion of participants below age 18, the data consist of 166,815 observations from 34 countries that participated in at least two and up to four waves between 2006 and 2014.

The forms of political participation are all binary measures and report whether an individual has participated in one of the actions listed in Table 1 during the last 12 months. The only exception is the voting variable that measures whether a participant voted in the last national election.

Loneliness is coded on an ordinary scale with four categories. The participants had to report how often they felt lonely in the past week: (1) almost none of the time (2) some of the time (3) most of the time or (4) almost all of the time. Treating loneliness as a metric instead as a categorical variable does not change the results. Likewise, the analysis has been double-checked using a binary coding of loneliness differentiating between people who almost never felt lonely versus categories 2, 3 and 4. In all models, the direction of the effect and significance levels remain unchanged. The effect sizes in the metric operationalization are smaller compared to the other control variables while the effect sizes in the binary operationalization seem to be somewhat inflated. Therefore, the reported results are based on the metric operationalization for reasons of caution, clarity and comprehensibility.

Furthermore, I consider several control variables. It is possible that political actions might be mostly utilized for reconnection by people highly involved with politics. To account for this possibility I control for self-reported political interest. Another important control is general health. Several decades of research has shown that loneliness exerts a negative influence on physical as well as psychological health (Heinrich and Gullone 2006), and that poor health decreases participation (Mattila et al. 2013). The ESS contains a variable that measures the perception of onès subjective general health. Given the lack of more objective health measures, this variable is included as crude control for the general physical and mental wellbeing of the participants. Also, I incorporate income into the model via subjective satisfaction with one’s financial situation, given that financial worries might play a major role in protest, petitioning and party membership. The perception of ones` financial situation has two major advantages compared to the direct measure included in the ESS. First, the direct measure has over 24% missing values which would reduce the sample size substantially. Second, given that objective income is biased due to social desirability, household size, and other social factors, indirect measures capture the financial situation more adequately. Furthermore, I add several sociodemographic variables commonly controlled for in turnout and participation models. Namely, I include gender, age and the highest level of educational attainment. Finally, one might argue that the relationship between loneliness and participation is merely an expression of onès lack of objective social contacts. To preempt such concerns, I add a variable measuring how often the participant socially meets with friends, colleagues or relatives into the model as well as an indicator for unemployment because the workplace is a major source of social interaction in adult lives (Table 2).

To investigate the proposed pattern, I employ three-level logistic regression models to account for the clustering of the individual observations within the participating countries and waves. As a first analytical step, I discuss the valence and significance of the relationship between loneliness and the five outcomes to confirm that the data match the expected pattern. Valance and statistical significance are of course just rough indicators for the substantial relevance of an explanatory variable. As pointed out by Bernardi and colleagues, statistical significance can be misused to claim links between variables that are significant, but so small that they cannot be considered substantial or socially relevant (Bernardi et al. 2017). This consideration is especially important in regression analysis based on large sample sizes such as this one. Therefore, I will discuss the size of the effect in the second part of the analysis. In order to evaluate the substantial strength of the relationship, I standardize all non-binary variables and compare the effect strength of loneliness with other established predictors for political participation.

As expected, loneliness does significantly lower the probability of voting (p < .001), signing a petition (p < .01) and contacting a politician (p < .05). In respect to the assumed positive relationships, loneliness is significantly associated with an increased probability of joining a demonstration (p < .001). In contrast, the effect on the probability of working in political groups is insignificant (p < .791). Overall, the results are in line with the proposed hypothesis. Loneliness seems to decrease political participation in several dimensions, especially those that do not offer many opportunities for social interaction. For political actions that are more interactive, the effect is either insignificant or positive. This might indicate that the general decreasing impact of loneliness is overshadowed or partially attenuated by the motivation to re-affiliate with others.

As can be seen in Figure 3, loneliness predicts voting behavior in a way that is both statistically and substantially significant. One standard deviation increase in loneliness (std = .74 on a 4 point scale) increases probability to vote by 0.129. Translated into average marginal effects, every additional standard deviation increase in loneliness decreases the probability to vote by 1.86%. With that, perceived loneliness correlates with the reported voting behavior more strongly than the frequency of social interactions (std. coef = 0.059; AME = 0.86%; p < .001), and an increase of two deviations is almost as large as the difference between being employed and being unemployed (std. coef. = −0.249; AME = 3.7%; p<0.001). Similar to voting behavior, loneliness is associated with a lower probability of signing a petition (std. coef. = −0.021; AME = 2.09%; p < .007) as well as contacting a politician (std. coef. = −0.017; AME = 0.2%; p < .001). In comparison, frequency of social encounters is associated with a five to six times larger probability of signing a petition compared to loneliness (std. coef. = 0.138; AME=18.6%; p < .001). In respect to the probability of contacting politicians, loneliness is about 10 times smaller than frequency of social interaction (std. coef. = 0.174; AME = 1.9%; p < .001). Therefore, compared to frequency of social contact with peers, the association between loneliness, signing petitions and contacting politicians is comparatively weak, but still exceeds several other predictors in the model as can be seen in the Tables A1 and A2. In contrast, the association between loneliness and voting behavior is substantial.

In respect of the probability of joining a demonstration, every additional standard deviation in loneliness increases the probability to demonstrate by 0.044 (AME = 4.38%; p < .001). In comparison, frequent social interactions are associated with an increase by 0.187 (AME = 18.6% p < .001). Given that one additional standard deviation in loneliness is about 23% of the effect size of the frequency of social encounters, loneliness can only be considered as a moderate predictor for joining a demonstration. Overall, I interpreted the pattern as evidence that loneliness is associated with a lower probability to participate in political acts if they offer little opportunity for affiliation (voting, petition and contacting politicians). While objective social encounters exceed the influence of perceived loneliness in most cases, it is still a considerable predictor. Hence, the analysis provides support for the suppression hypothesis (H1). Considering the insignificant relationship to working for a political group and the significant, but a comparatively small association with the probability of joining demonstrations, the analysis provides only mixed support for the enhancement hypothesis (H2).

One possible concern for the electoral participation model is the amount of time that lies between the interview and the last election held in each of the countries. All forms of political participation tend to become more frequent with nearing elections, and memory biases are less problematic with short time gaps. I have constructed a variable that measures the amount of time between the last national election and the corresponding date of the interview to determine whether the relationship between loneliness and the political acts are confounded by the time measure. Including this variable in the model does not change the results. Thus, this variable is not included in the final models reported in the analysis.

Furthermore, recent studies have established a relationship between well-being and political behavior. Given that loneliness and subjective well-being are logically interlinked, the found relationship between loneliness and political participation might be a confounded effect of well-being. To resolve this concern, I tested whether the link remains robust under control for general life satisfaction. Adding this variable does not change the results except for the ‘contacting politicians’ model where loneliness remains marginally significant after adding the variable (p = .083). However, the strong inter-correlation between loneliness and life satisfaction (corr. = 0.35) is likely to bias the effect strength considerably and the investigation of the substantial effect size would not be possible due to over controlling. Thus, life satisfaction is not controlled for in the final models.

Finally, the two countries included in the analysis might be considered problematic. First, voting is compulsory in Belgium, resulting in atypically high voter turnouts. Second, Russia has a very repressive policy regarding public demonstrations and its elections are characterized by suppressed and barred candidates and repeated allegations of irregularities. Omitting these countries from the voting model and demonstration model respectively does not change the results.

This study provides evidence that loneliness can be seen as a valuable predictor for reported political behavior. I theorized that political participation should become less likely for political acts that are unlikely to foster social engagement (suppression hypothesis), while political acts that foster social engagement were expected to become more likely (enhancement hypothesis). In line with these expectations, all three political actions that offer little opportunity for social re-affiliation are negatively associated with loneliness.

In contrast, loneliness did predict one out of two political actions that were theorized to foster social interaction. In light of the mixed support for the enhancement hypothesis, upcoming research should investigate why loneliness and working in a political group seem to be unrelated. One possible explanation might be that being active in political groups is perceived as more demanding. The higher degree of commitment required to join organized groups such as political parties is especially problematic for lonely individuals who tend to be socially anxious and suffer from depressed mood. Therefore, while providing a platform for social reconnection, demanding institutional barriers might prohibit lonely individuals from join such organizations. Likewise, leaving these groups should happen rarely as well; it is more likely that members become increasingly inactive while trying to preserve the membership in order to reconnect with other members in the future.

Furthermore, the results of the analysis should be read in light of some limitations. First, very little research has shed light on the question of what political action is motivated by the wish to socialize. While there are clear reports that citizens go to demonstrations to meet with like-minded people and to remain in touch with friends, the categorization of the other forms of participation is based on few studies. Upcoming qualitative studies should investigate this question which might allow a more sophisticated categorization. In a similar fashion, one key distinction in the loneliness literature is the difference between emotional and social loneliness (DiTommaso & Spinner, 1997). Investigating how both types of loneliness relate to the proposed hypothesis is a promising field of investigation for upcoming research.

Furthermore, the analysis is based on correlations of cross-sectional multilevel data and cannot account for typical limitations of such designs such as causality or unobserved confounding. Especially unobserved personality traits or social media usage could be confounding factors that cannot be controlled for in this study (Buecker et al., 2020; Ryan & Xenos, 2011). Upcoming research might try to tackle this issue with panel data on the individual level to account for the time constant, unobserved variables.

In the broader picture, this study relates to the core debate in political science about social embeddedness and social capital. While many political scholars do focus on network characteristics such as mutual trust and objective characteristics of one’s social environment, loneliness relates to the subjective perception of these characteristics and offers an interesting additional facet of this phenomenon that is yet to be discussed in the literature. Given that Great Britain has already tasked a minister with combatting loneliness and policies such as social prescriptions are implemented in Great Britain and the Netherlands, a more comprehensive understanding of the phenomenon is needed to provide the groundwork for informed policy decisions.

No potential conflict of interest was reported by the author(s).

Alexander Langenkamp is a PHD Candidate in Sociology at the Goethe University Frankfurt. His research interests include consequences of loneliness and social isolation on social participation.