Everyday Political Talk with Strangers: Evidence on a Neglected Arena of the Deliberative System

According to normative theorists, informal conversations between strangers are the most basic manifestation of the political public sphere and truest to the deliberative democratic ideal. Yet systematic empirical evidence on citizens’ everyday political talk outside their social networks is largely missing. Using a unique survey, we examine citizens’ access to the public discursive sphere of political talk with strangers, as well as the frequency and disagreeableness of the conversations held in this arena of the deliberative system. Although widespread and frequent engagement is desirable from a normative point of view, we find this discursive sphere to be considerably smaller in scope and less vibrant than the private and semi-public discursive spheres of political talk within strong and weak network ties. Contrary to theorists’ equation of strangeness with difference, political conversations between strangers also appear rather harmonious. Furthermore, our findings show that psychological dispositions, most notably social trust and conflict orientations, are important drivers of individuals’ involvement in political conversations with strangers. Their impact exceeds the influence of political dispositions, opportunities, and skills. Some aspects of our results raise doubts about the deliberative quality of these conversations.

from political communication and participation, social networks, and sociological modernization research, we thus aim to contribute to a growing literature that seeks to strengthen the dialogue between abstract theorizing about deliberative democracy and the empirical study of citizens' interpersonal communication (Mutz 2008).
We conceive of everyday political talk with strangers as a specific discursive sphere (Hendriks 2006) of a distinctly public character that can be systematically differentiated from the private discursive sphere of conversations between individuals connected by strong network ties and the semi-public discursive sphere of communication within weak ties (Schmitt-Beck and Grill 2020). Our analysis of this uncharted territory concentrates on three facets of ordinary people's involvement in informal political talk with "anonymous others" (Giddens 1990, p. 120) outside their circles of personal relationships: i) their "access" (Knight and Johnson 1997, p. 281) to this arena of communication, that is, whether they engage in this kind of activity at all; ii) the frequency of such engagement among those who access this arena; and iii) the political disagreement to which they are exposed during these conversations (Klofstad et al. 2013).
We begin with a review of extant scholarship on the role of ordinary citizens' everyday political talk in the public sphere of deliberative democracy. We find that informal conversations between strangers constitute an arena of the deliberative system considered crucially relevant by theorists but neglected by empirical research. We then develop hypotheses concerning two sets of questions: first, how many people access the public discursive sphere of everyday political talk with strangers, how intense is this engagement among those who do, and how often does it lead to experiences of political disagreement? Second, what conditions facilitate or attenuate this involvement? To test these hypotheses, we draw on the Conversations of Democracy study, a unique face-to-face survey specially designed to examine German citizens' everyday political talk.

Deliberative Democracy, the Political Public Sphere, and Citizens' Talk About Politics
Advocates of deliberative democracy praise discussions between those holding opposing views as the most constructive mode to address disagreements between diverse societal interests and value orientations (Habermas 1996;Sunstein 2003). Discussing contrasting worldviews and perspectives is expected to establish a more refined understanding of matters of conflict, and to stimulate adjustments of preferences in the light of reasoned arguments that take the interests of all affected groups into account. Political discussion is thus highlighted as a superior source of democratic legitimacy and societal integration in secularized, highly differentiated, politically plural modern societies (Manin 1987;Habermas 1994Habermas , 1996. To qualify as democratic, deliberative democracy presupposes substantial and effective involvement of the citizenry at large (Barber 1984;Mansbridge 1999;Lafont 2020;Schmitt-Beck 2022b, pp. 13-16). 3 From a systemic perspective (Parkinson and Mansbridge 2012), its politics is envisaged in terms of a multiplicity of interlocking discursive spheres (Hendriks 2006), that is, arenas for discussing public affairs organized in a continuum that extends from citizens' everyday communication all the way to the institutions of government (Neblo 2015, pp. 17-25). The principle of democratic legitimation demands that within this continuum, "communicative power" (Habermas 1996, pp. 359-387) travels bottom-up, and originates from citizens' informal exchanges in their lifeworld. 4 According to Habermas, the state's authoritative decision-making should be guided by "informal public opinion," conceived as the outcome of communication processes constantly taking place within the political public sphere that mediates between citizens' lifeworld and the institutions of government (Habermas 1989(Habermas , 1991(Habermas , 1996. The public character of this system of political communication can be related to two analytically distinct meanings of this complex concept (Weintraub 1997, pp. 4-7): first, the substance it processes consists of topics, perspectives, and views about "public affairs," that is, matters of relevance for the common good of society that require political regulation by means of authoritative decision-making (Dahlgren 2009, pp. 89-90); second, it is open and accessible for everyone (Goffman 1963, p. 9;Gerhards and Neidhardt 1991, pp. 44-47). The latter dimension in particular demarcates the public sphere from the encapsulated intimacy of the private lifeworld of family and friends (Sennett 1977, pp. 16-24;Habermas 1989, p. 30).
Political discussions between ordinary citizens are often highlighted as an archetypical appearance of the public sphere. Habermas, for instance, notes that "[a] portion of the public sphere is constituted in every conversation in which private persons come together to form a public" (Habermas 1991, p. 398). Historical 3 Although theorists have characterized people's everyday political talk as the "centerpiece" of deliberative democracy (Mansbridge 1999, p. 228), empirical research has primarily focused on how citizens discuss politics as invited participants of formalized deliberative forums (Landwehr 2020), rather than informally and spontaneously in their everyday lifeworld. Recent criticism has therefore accused deliberative democratic scholarship of downplaying the democratic element in a shift toward "participatory elitism where citizens who participate in face-to-face deliberative initiatives (and only a small fraction do) have more democratic legitimacy than the mass electorate" (Chambers 2009, p. 344;Lafont 2020). It accordingly requested theory and research on deliberative democracy to revive its "broken link with mass politics" (Bächtiger and Parkinson 2019, p. 56). The present paper is part of a larger research program that takes up this impulse to contribute to a better empirical understanding of ordinary citizens' role in the deliberative system. 4 It bears noting that this does not require such discussions to meet high standards of deliberative quality, such as reason-giving, listening, or respect . Applying such criteria presupposes that people talk about politics in the first place. Theorists of deliberative democracy tend to take a sceptic stance on the potential deliberativeness of casual everyday conversations. Mansbridge (1999), for instance, argues that from a systemic perspective it is more important that people discuss politics at all than that such talk conforms to demanding standards. Accordingly, our analytical focus is on political discussion as such, conceived as a necessary though not sufficient prerequisite of genuine deliberation (see Schmitt-Beck 2022b). While the deliberative character of ordinary citizens' political talk has been examined for structured discussions within formalized mini-publics (Gerber et al. 2018), an empirical verdict about the quality of everyday political talk is still pending. Our systemic perspective also implies that political discussions must not lead directly to decisions in order to be valuable for the overall process of deliberative will-formation (Habermas 1996;Mansbridge 1999). accounts portray political exchanges in face-to-face encounters as their earliest manifestation (Sennett 1977;Habermas 1989). The communicative structure of modern societies' public sphere is of course more complex (Gerhards and Neidhardt 1991, pp. 49-56;Wessler et al. 2020). Nonetheless, everyday political talk is still emphasized as the "fundamental underpinning" and "producer" of the public sphere (Kim and Kim 2008, pp. 51, 53), and accordingly the "main platform" of deliberative democratic politics (Tanasoca 2020, p. 232;Barber 1984;Mansbridge 1999).

Strong Ties, Weak Ties, No Ties
To relate everyday political talk more precisely to the notion of the public sphere, it is useful to think of it as a phenomenon occurring in a distributed way across three arenas of political communication that can be conceptualized as distinct discursive spheres-one private, one semi-public, and one public in nature. They all share a thematic focus on public affairs, but they differ with regard to the structural criterion of openness and accessibility.
Research has thus far concentrated on the first two of these discursive spheres. They both pertain to communication within citizens' social networks (Schmitt-Beck and Lup 2013;Conover and Miller 2018). Of course, people's networks of enduring relationships are an important context for political discussions. Studies suggest that political talk is particularly widespread between persons attached through the strong ties (Straits 1991) of kinship and friendship (Allan 1979). These discussions constitute a distinct discursive sphere of everyday political talk. They usually take place in protected spaces, most notably people's homes, and are therefore of a private nature (Sennett 1977, p. 16). The political views held by their participants are typically very similar (Nir 2017).
The second discursive sphere concerns conversations between acquaintances, such as co-workers or neighbors (Goffman 1963, pp. 112-123). These weak ties (Granovetter 1973) result from contacts made within certain spatial or functional foci (Feld 1981), such as workplaces and neighborhoods, where access is controlled by social norms rather than physical and legal barriers (Goffman 1963, pp. 151-165).
Since individuals' discretion about who to talk to is accordingly incomplete, the character of these conversations is semi-public. Their prevalence and frequency tend to be lower than within strong ties, but due to their more open character they entail a higher likelihood of exposing citizens to political disagreement (Huckfeldt et al. 2004). In the literature they are therefore considered especially valuable for deliberative democracy (Tanasoca 2020).
However, face-to-face interactions are not restricted to strong or weak ties. Mutual "face engagements" may also occur on the part of individuals between whom no ties exist because they are unacquainted with each other (Goffman 1963, pp. 124-148). Modernization theory has indeed identified interactions between strangers as a core feature of social life in contemporary societies (Giddens 1990, pp. 79-83). This creates the possibility of episodic "focused interactions" (Goffman 1963, p. 24) in the specific form of casual conversations about politics between people who do not know K each other. Thus situated outside social networks, this type of citizens' informal political communication constitutes a discursive sphere of its own. Opportunities for this mode of everyday political talk arise in unrestricted spaces where access is open to anyone. It presupposes that individuals who are unacquainted to each other are transient "fellow users" of the same social setting (Goffman 1971, p. 28). Its locus is an impersonal, unrestricted, and thus in the structural sense unequivocally public realm of conviviality that emerges from a complex interplay of spatial and social arrangements (Weintraub 1997, pp. 16-25).
Railroads, buses, airplanes, and other means of public transport are quintessential settings for such episodic communications between individuals that share nothing except their co-presence in the same space at the same time (Goffman 1963, p. 22;Noelle-Neumann 1974;Riesman 2016;Sandstrom and Boothby 2021). They require complete strangers, brought together by some sort of stochastic process, to spend stretches of time with one another, thus opening up the possibility to engage in conversations about, among other things, politics. Parties, pubs, and cafes (Sax 2022), sports events, or doctors' waiting rooms are further examples. What these situational contexts have in common is that they create transient occasions for political exchanges, ranging from a few quick words in passing to lengthy and intense dialogues, with people to whom one has never talked before and whom one will never meet again. They open up fleeting opportunities to approach or be addressed and drawn into conversations about public affairs with persons not belonging to one's social network. Often participants in such conversations won't even exchange names (Riesman 2016, p. 109). In contrast to the private and semi-public conversations between network members, this kind of everyday political talk is public on all accounts, not only with regard to its thematic focus but also regarding its openness and accessibility.

Discursive Spheres of Everyday Political Talk and the Political Public Sphere
How then can these three discursive spheres be related to the notion of the political public sphere? Habermas seems to count all kinds of talk among citizens into this realm, as long as it revolves around political problems and their solutions (Habermas 1989;Dahlgren 2009, pp. 89-90). Such an interpretation is consistent with a purely theme-based understanding of the public sphere. Additionally applying the structural criterion of openness and accessibility leads to a more differentiated view, however. According to this reading, the discursive sphere of strong ties belongs in the realm of citizens' private lifeworld, whereas the semi-public discursive sphere straddles the divide between the lifeworld and the public sphere (Schmitt-Beck and Grill 2020). By establishing bridges between different strong-tie networks, weak ties open up a larger and more diverse web of experiences for individual network members. From a deliberative democratic viewpoint, this renders them more useful than strong ties (Huckfeldt et al. 2004;Tanasoca 2020). By contrast, the discursive sphere of political talk outside social networks has a clear public status with regard to both its topical focus and its openness to everyone. Everyday political talk between strangers is thus the only unambiguously public mode of citizens' informal communication, and accordingly distinct in its unequivocal belongingness to the public sphere. This conceptualization concurs with several strands of theorizing about deliberative democracy and the political public sphere. It is well in line, for instance, with Habermas and Sennett's celebration of 17th and 18th century coffee houses, inns, pubs, and salons as "places where strangers might regularly meet" to discuss public affairs (Sennett 1977, p. 17), which turned these sites into the inaugural institutions of the public sphere (Habermas 1989, pp. 31-43). An analytical model proposed by Gerhards and Neidhardt expressly stipulates episodic "encounters" between strangers as the most basic manifestation of the contemporary public sphere (Gerhards and Neidhardt 1991, pp. 49-56;Habermas 1996, p. 374). Hauser goes even further by very pointedly conceiving of the public sphere as "a discursive space in which strangers discuss issues they perceive to be of consequence for them and their group" (Hauser 1999, p. 64). Young praises the public life of cities as a normative ideal for a functioning democracy because of its "more open public [...] where strangers meet and interact" (Young 1990, p. 237). In a similar vein, Barber claims that democratic politics should be seen as "the art of engaging strangers in talk" (Barber 1984, p. 190). Mini-publics, deliberative democracy's unique institutional innovation (Setälä and Smith 2018), can indeed be seen as formalized attempts to capitalize on the enlightening potential of political discussions outside the boundaries of social networks.
Despite this theoretically elevated status, research on citizens' involvement in the politics of deliberative democracy has not paid any systematic attention to the genuinely public discursive sphere of everyday political talk with unknown others. 5 To address this deficient state of research, we proceed in two steps. We begin with descriptive analyses of citizens' involvement in political conversations with persons they do not know: How many people actually access the public discursive sphere of everyday political talk, and how frequent and how disagreeable is such talk among those that do so? As yardstick for developing testable hypotheses, we refer to everyday political talk within the private and semi-public discursive spheres of strong and weak network ties. Subsequently, we examine the conditions of citizens' involvement in political conversations with strangers. be no less intense than in the private and semi-public discursive spheres. Turned into a testable hypothesis, this normative proposition implies the expectation that access and frequency of talk with strangers are at least as high as within social networks (H1a).
However, from an analytical point of view, this expectation appears rather implausible. Research indicates that contextual circumstances matter for everyday political talk (Carlson and Settle 2022, pp. 48-49). The early public sphere appears to have profited from a remarkable eagerness on the part of the emerging citizenry to discuss public affairs with strangers (Sennett 1977). Yet the socio-economic and cultural changes of the 19th century seem to have altered citizens' outlook toward this activity profoundly. "There grew up the notion that strangers had no right to speak to each other, that each man possessed as a public right an invisible shield, a right to be left alone." (Sennett 1977, p. 27) Goffman describes vividly how in modern societies a "tacit contract" appears to safeguard "the individual's usual right to be unmolested by overtures" from strangers (Goffman 1963, pp. 124-127;Sandstrom and Boothby 2021). A sharp line seems to demarcate "[o]ur 'us'-the family, circle of friends, co-workers-[...] from everything that is 'non-us.' Those who fall into the category of 'non-us' are [...] 'strangers' (a group unknown and different). And everyone knows that you shouldn't talk to strangers." (Poe 2011, p. 52) This suggests that citizens' engagement in the public discursive sphere is lower than in the private and semi-public discursive spheres (H1b).
With regard to their ability to expose individuals to views other than their own, weak ties are believed to be superior to strong ties (Huckfeldt et al. 2004;Tanasoca 2020). However, while perhaps not fully independent from predetermined paths of social structure, conversations outside social networks are in any case less dependent on them, and thus entail a stronger random component. Hence, they should lead to even more normatively desirable encounters with the political plurality of highly differentiated modern society. For this reason, some theorists simply equate interactions with strangers with experiences of otherness (Young 1990, pp. 226-256). It can accordingly be hypothesized that in the public discursive sphere, people encounter more disagreement than in the private and semi-public discursive spheres (H2a).
However, the "tacit contract" that presumably decreases people's likelihood of discussing politics with strangers in the first place might also influence the style of such interactions if they occur. Morey et al. (2012) argue that the intimacy of kinship is more accommodating to dissenting voices than more distant relationships. In a complementary fashion, research on mini-publics suggests that "conversations on matters that can be divisive may be considered out of place in public" (Tatarchevskiy 2012, p. 220). Various communication strategies can prevent latent opinion differences in interlocutors' minds from becoming manifest in political discussions (Peacock 2019; Carlson and Settle 2022, pp. 161-164;Schmitt-Beck 2022a). Together, these findings give rise to the rival expectation that experiences of disagreement are less pronounced outside than inside social networks (H2b).

What are the Conditions of Citizens' Involvement in the Public
Discursive Sphere?
What renders individuals more or less likely to access the public discursive sphere? What determines the frequency of such conversations among those who do so? And which factors are related to experiences of political disagreement? Since no systematic research exists on everyday political talk with strangers, we approach these questions in an exploratory fashion. We apply an outcome-centered research design that takes a variety of potentially relevant predictors into account (Sieberer 2007). Our hypotheses are guided by a generalized approach-avoidance perspective 6 for all three aspects of involvement: engagement with regard to i) access and ii) the frequency of political talk among those with access, as well as iii) disagreement experiences during these conversations. Drawing on research on political participation (Verba et al. 1995) and interpersonal communication (Huckfeldt and Sprague 1995), these hypotheses refer to three theoretical building blocks: motivations, skills, and opportunities. When testing these expectations, we aim to separate conditions of specific relevance for conversations with strangers from generic conditions that pertain to everyday political talk as such, that is, regardless of its different arenas. The first theoretical building block is motivational. According to Carlson and Settle (2022, pp. 24-31), individuals' involvement in everyday political talk is driven by information-based motives that create a desire to learn about other people's political views, individual-based motives that give rise to a desire to pursue a positive self-concept, and relationship-based motives that lead to a desire to affiliate with others. A range of political and psychological dispositions can thus be assumed to render political talk with strangers-or its avoidance, for that matter-desirable and gratifying. They should therefore affect involvement in this behavior (Carlson and Settle 2022, pp. 45-48). H3a assumes a positive impact of political dispositions (Jacobs et al. 2009, pp. 55-59;Schmitt-Beck 2022b). The most general one is interest in politics, which can be understood as a stable "expectation that engaging with political content [...] in the future will turn out to be rewarding" (Prior 2019, p. 4). Passionate stances with regard to ideological camps or parties could also let political discussions appear gratifying. Persons with strong attitudes concerning directional alignments like partisanship and ideology should thus also display stronger involvement in conversations with strangers as welcome opportunities to affirm their political identity.
H3b posits partly positive and partly negative effects of psychological dispositions. While the aforementioned political dispositions are well established in research on everyday political talk, psychological traits have only recently begun to attract attention (Carlson and Settle 2022, pp. 45-48). Need for cognition, individuals' enjoyment of effortful cognitive tasks (Cacioppo and Petty 1982); need to evaluate, persons' inclination to think in evaluative terms and develop opinions (Jarvis and Petty 1996); social trust, that is, individuals' basic confidence in the goodwill of people outside their social networks (Nannestad 2008); and need to belong, a basic motivation to form social relationships (Sandstrom and Dunn 2014), might let everyday political talk with strangers appear gratifying and thus lead to approaching stances. By contrast, need for cognitive closure, a preference for clarity and concomitant dislike of ambiguity, uncertainty, and confusion (Kruglanski and Fishman 2009), and social isolation as an interpersonal dimension of general alienation (Dean 1961;Fischer and Kohr 1980, pp. 90-109) might affect such involvement negatively. Individuals' conflict orientation should play a role as well (Mutz 2006), with conflict seeking rendering involvement in the public discursive sphere more likely, and conflict aversion less likely.
Following the civic voluntarism model of political participation (Verba et al. 1995), we furthermore assume that certain skills facilitate discussing public affairs with strangers as well as encountering disagreements while doing so (H4). Internal political efficacy, citizens' confidence in their ability to make a difference in politics (Craig and Maggiotto 1982), might increase their self-assurance and lead to an approaching orientation. Since political discussions demand specific capabilities, such as a basic understanding of the thematized subject matters and some measure of conversational ability, people endowed with a high competence to discuss politics (Rubin et al. 1993) should also display more involvement (Sandstrom and Boothby 2021). In addition, cognitive resources like political knowledge and education may also let such involvement appear more gratifying.
Lastly, opportunity structures for interacting with strangers (Straits 1991;Huckfeldt and Sprague 1995;Sandstrom and Boothby 2021, p. 64) can be expected to play a role (H5), although perhaps more strongly for engagement than disagreement experiences. Workplaces, churches, and organizations like trade unions may serve as such foci (Feld 1981). Sociability in the form of spending leisure time with other people could have a similar effect. Living in a single household might entail stronger incentives to mingle with others outside one's home than living in a multi-person household. Economically well-to-do persons are better able to afford an expansive lifestyle which in turn might endow them with more chances to interact with people they are not acquainted with. The constant exposure to strangers is often depicted as a hallmark of urban life (Sennett 1977;Young 1990, pp. 226-256); accordingly, we expect the likelihood of discussing politics with such persons to rise as a function of urbanization. Moreover, we assume that due to social segregation, persons of migrant descent are less likely to interact with persons they do not know. Lastly and more specifically, prospects of encountering political disagreement can be expected to increase as a direct function of the frequency of political talk (Huckfeldt and Morehouse Mendez 2008).

Data
We draw on data from the Conversations of Democracy study, a survey that was specially designed to examine German citizens' everyday political talk. The study's 1600 respondents were recruited via a register-based one-stage random sample and interviewed face-to-face between 15 May and 24 September 2017. 7

Dependent Variables
To elicit respondents' engagement in the various discursive spheres, the following instrument was used: "If you think about the last six months, how often have you talked with members of your family/friends/acquaintances, such as neighbors or people at work/people you don't personally know, about political topics: Never-once a month or less-several times a month-several times a week-daily or almost daily?" 8 Talking about politics within the strong ties of family and friends refers to the discursive sphere of private political conversations, discussions between acquaintances to the semi-public discursive sphere, and discussions with strangers to the discursive sphere of public everyday political talk. To assess exposure to political heterogeneity within these discursive spheres, we refer to follow-up questions that elicited the amount of general disagreement (Klofstad et al. 2013) encountered during the various kinds of political talk: "If you think back to your conversations about political topics with family members/friends/acquaintances/people you don't personally know, in the last six months, in general, how often would you say there were opinion differences between you and your discussion partner: Never-rarely-sometimes-often-very often?" For modeling the conditions of involvement in everyday political talk with strangers we derive three variables from these measures that we analyze by means of binary and ordered logistic regression models: i) whether or not people accessed the public discursive sphere (1 = discussed politics with strangers, 0 = never discussed politics with strangers); if they did so, ii) how often ("once a month or less" to "daily or almost daily"); and iii) with which amount of disagreement experiences ("never" to "very often"). 7 Following the model of major studies of political communication in citizens' lifeworld (Lazarsfeld et al. 1968;Huckfeldt and Sprague 1995;Huckfeldt et al. 2004), the study was conducted locally. Its site was Mannheim, a city characterized by the variegated social structure, economy, culture, and political life of a typical mid-sized German city. For methodological details see Grill et al. (2018). 8 See Morey and Eveland (2016) for a discussion of this type of measure's reliability and validity.

Independent Variables
Motivations Political interest is measured through self-reports, ideological extremity by means of a left-right scale folded at the midpoint. Whether respondents identify with a political party is indicated by a dummy variable. The psychological measures are scales including between one and six items.

Skills
Internal political efficacy and discussion competence are indicated by additive scales based on two items each. Since our data include no direct measure of political knowledge, we resort to media use as proxy. We focus on news media that offer a rich information diet and have been shown to render their users more knowledgeable (Aalberg and Curran 2012). Accordingly, we rely on self-reports on the frequency of reading newspapers and watching public TV news. Education is a dummy variable contrasting respondents with completed upper secondary education from less educated individuals.
Opportunities Gainful employment, trade union membership, living in a single household, and migration background are indicated by dummy variables, church attendance by a scale. Sociability is an additive scale based on self-reports about the amount of time spent with various categories of associates. Economic well-being is measured by respondents' assessments of their current economic situation. To measure urbanization, we refer to official statistics on the population density of the city districts where respondents reside.

Strategy of Modeling
All models control for age and gender. Before inclusion in the models, all continuous predictors were normalized to the range of 0 to 1. For each of our three dependent variables, three questions are of interest with regard to the models' findings: whether each block of predictors (i.e., political dispositions, psychological dispositions, skills, and opportunities) is relevant as a whole, whether and which individual predictors within each block are associated with the outcome variables in the expected directions, and how these predictors' effects differ in strength. To get a sense of the relevance of each block of predictors, we refer to partial measures of model fit. One of these measures is permissive and indicates the model improvement achieved through all predictors within a block compared to a baseline model that only includes the two demographic controls. The other measure is restrictive and takes the possibility of associations between the various blocks into account. It refers to the improvement in model fit associated with each block relative to a model that includes the two demographic controls as well as all other blocks of predictors. For assessing the role of the individual predictors within the blocks, we refer to these variables' coefficient estimates.
In our analysis we present two models for each dependent variable: one that contains all predictors listed above (M1.1, M2.1, M3.1), and one that additionally controls for the respective modes of involvement in everyday political talk within strong and weak ties (M1.2, M2.2, M3.2). By partialling out the proportion of our predictors' impact that in a generic way relates to everyday political talk as such, the latter models identify those effects (or effect shares) that specifically pertain to conversations with strangers (see Supplementary Materials for technical details and descriptive information for all predictors). 9

Citizens' Involvement in the Public Discursive Sphere
How many people access the public discursive sphere, and how frequent is political talk among those who do? According to Fig. 1, citizens discuss politics considerably less with strangers than with members of their social networks, as expected by H1b. The overall most active arena is the private discursive sphere. Only very few citizens never discuss politics with family members and friends. Echoing extant research (Schmitt-Beck and Lup 2013; Schmitt-Beck 2022b), semi-public settings like workplaces or neighborhoods appear somewhat less conducive to political talk. But crossing the threshold to the public realm leads to a massive reduction in both the prevalence and frequency of political talk. More than half of the respondents do not access this discursive sphere at all. They never discuss political matters with anyone they do not know. The intensity of political conversations outside social networks is also much lower than within them. The share of regular conversations with strangers, held at least several times a month, amounts to about a quarter of those who do engage in this mode of everyday political talk, compared to 75% for strong ties and 61% for weak ties. "Once a month or less" is the clear modal category (35% of all respondents, corresponding to 76% of those with access). Thus, despite its significance from a normative point of view, the public discursive sphere of everyday political talk with strangers is not only smaller, due to the considerably lower share of those accessing it, but also less vibrant.
With regard to disagreement experiences, Fig. 2 indicates a nonlinear pattern. On the one hand, we see somewhat larger shares claiming to experience disagreement "often" or "very often" when talking to strangers. Together they amount to about Engagement in everyday political talk in the private, semi-public, and public discursive spheres (percentages). Data are weighted by gender, age, and city district (case numbers unweighted)

Fig. 2
Experiences of political disagreement in the private, semi-public, and public discursive spheres (percentages). Only accessing respondents; data are weighted by gender, age, and city district (case numbers unweighted) K that participants of such talks tend to avoid clear opinion statements if they entail the prospect of open disagreement (Peacock 2019). Table 1 shows the conditions of citizens' involvement in the public discursive sphere. The partial model fit measures suggest that each of the different blocks of predictors is relevant. However, zooming in on the individual independent variables within each block reveals considerable differences in the predictors' impact. In addition, we also observe some associations that are statistically substantial but whose direction is contrary to our expectations.

Conditions of Involvement in the Public Discursive Sphere
Comparing the blockwise fit measures suggests that whether citizens access the public discursive sphere is overall most strongly affected by psychological dispositions, followed by political dispositions and opportunities (M1.1). Controlling for access to conversations within strong and weak ties, we see little change in the blockwise fit measures, indicating that all blocks' impact is mostly specific to conversations with strangers (M1.2). Among those persons who do engage in everyday political talk with strangers, the frequency of this activity in a similar vein depends primarily on psychological motivations and secondly on political dispositions, whereas skills and opportunities appear less relevant (M2.1). The impact of political dispositions entails a pronounced generic component (M2.2). Experiences of disagreement are affected by all blocks of predictors roughly similarly, though overall less strongly than access and frequency (M3.1). Controlling for disagreement experiences in strong and weak ties diminishes the relevance of political dispositions, skills, and opportunities, but not the psychological dispositions (M3.2).
Inspecting the individual predictors shows that of the political dispositions, only interest in politics is related to individuals' access to the public discursive sphere, and its effect is quite strong. It is also related to the frequency of such conversations and experiences of disagreement, but only due to its generic relevance for everyday political talk as such and not because of a specific function for conversations with strangers. Instead, the directional motivation of left-right extremity takes over. Ideologically more zealous individuals discuss politics much more often with strangers than moderates, and they also encounter more opinion differences, even after controlling for the frequency and disagreement experienced within social networks. Partisanship appears largely irrelevant by contrast. In a quite nuanced way, this evidence is largely in line with H3a.
For the psychological dispositions an even more complex picture emerges. Only the need for cognition is completely unrelated to citizens' involvement in the public discursive sphere. More trusting as well as conflict-seeking persons seem more inclined to access this arena of political talk, whereas social alienation is an obstacle. Access provided, social trust also facilitates more frequent conversations. In addition, political talk with strangers tends to be more frequent among those with a high need to evaluate, which is in line with H3b, but also among those with a low need to belong, which is not. That a high need for cognitive closure is associated with more rather than fewer conversations of this kind likewise contradicts H3b. However, when partialling out disagreement in social networks, a negative associa-

Talk social networks
Access strong ties --2.008 * ----Access weak ties  tion emerges between need for cognitive closure and disagreement experiences with strangers. It also becomes apparent that conflict seeking renders encounters with opinion divergences more likely and conflict avoidance less likely. These patterns are again in line with H3b. Remarkably, with only one exception, these associations hardly change when controlling for social network communication. They are thus of specific relevance for involvement in the public discursive sphere, and do not concern everyday political talk as such.
As noted, skills are overall least relevant for political conversations with strangers. Upon closer inspection, Table 1 additionally reveals that their effects are also mostly not in line with H4. According to M1.2 and M2.2, individuals who do not regularly follow public TV news and who are therefore presumably less knowledgeable about politics are more likely to access this discursive sphere and to engage there more often. Weaker negative effects for education likewise suggest that a higher endowment with cognitive resources is rather detrimental to this engagement. In addition, individuals with limited discussion competence partake in such conversations more rather than less frequently. The only effect consistent with our hypothesis suggests that high internal efficacy goes along with more encounters with disagreement.
With one exception, opportunities are only relevant for engagement in the public discursive sphere but not for exposure to disagreement. As predicted on the basis of network research (Huckfeldt and Morehouse Mendez 2008), persons that discuss politics in this arena more often are also more likely to encounter opinions other than their own. This of course implies that the factors relevant for the frequency of engagement in the public discursive sphere indirectly also affect individuals' chances of encountering disagreement. For engagement in political talk with strangers, our findings are again quite complex, beginning with the fact that hardly any predictors are relevant for both access and frequency. As expected by H5, sociability and residing in highly urbanized areas are associated with more access. We furthermore see a small negative effect for migration background. According to M2.1, the demographic circumstances of being part of the workforce and living alone indicate opportunities for more frequent talks with strangers. Not in line with H5 is, by contrast, a sizable negative effect of economic well-being. A similar association also emerges for the frequency of these discussions, seemingly to a small part originating from a generic role of this predictor for everyday political talk as such. Contradicting our expectations, these observations suggest that economically advantaged individuals engage less rather than more than disadvantaged ones in political conversations with persons they are not acquainted with.

Conclusion
In deliberative democracy, legitimate decision-making presupposes widespread processes of political discussion within civil society whose outcomes crystallize in the form of informal public opinions and are mediated by the political public sphere into the arenas of authoritative will-formation (Manin 1987;Habermas 1996). Theorists have pointed out that political discussions between strangers are the mode of citizens' everyday political talk that is truest to the essence of the deliberative democratic project (Sennett 1977;Barber 1984;Habermas 1996;Hauser 1999). Casual conversations between persons who are unacquainted with one another are the only discursive sphere of everyday political talk that is unambiguously public in character. Presumably, it is here, rather than within the confines of strong or even weak ties (Tanasoca 2020), that society at large most clearly engages in a conversation with itself. What renders political talk between strangers particularly valuable from theorists' point of view is its presumably superior capability to expose individual citizens to society's diversity and political pluralism (Young 1990, pp. 226-256). Empirical research has thus far paid hardly any systematic attention to informal political conversations outside social networks. Accordingly, it does not allow this claim's validity to be assessed. What is more, it is not even known how many people engage in this mode of political talk in the first place, and how intensely they do so.
Our study offered initial insights concerning this blind spot of research on citizens in the deliberative system. We examined three aspects of citizens' involvement in the public arena of everyday political talk: whether they access this discursive sphere at all, the frequency of such engagement among those who do, and the amount of political disagreement they are exposed to while discussing politics with strangers. Although widespread and frequent engagement is desirable from a normative point of view, we found the public discursive sphere to be considerably smaller in scope than the private and semi-public discursive spheres, and also less active. With about one of two citizens engaging in casual conversations with strangers at least occasionally, it is not negligible, to be sure. Clearly, it is a phenomenon that deserves further attention. But the larger part of citizens' everyday political talk occurs within social networks. It is thus more contained and less open than normatively preferable from a deliberative democratic perspective that emphasizes the need for a strong engagement in the political public sphere.
Concerning exposure to society's political diversity-taken for granted by theorists but deemed less likely from a micro-sociological point of view (Goffman 1963)-we arrived at a somewhat paradoxical finding: talk within the public discursive sphere is at the same time more heterogeneous and more homogeneous than conversations within social networks. The amount of pronounced opinion differences encountered during such conversations appears somewhat larger than when communicating with friends and acquaintances, but not within families. At the same time this communication was much more often characterized by the complete absence of disagreement than conversations within both weak and strong ties. To a rather limited extent, the public discursive sphere thus seems to fulfil the function assigned to it by normative theorists: to expose citizens to their society's heterogeneity of interests, values, and political preferences. But the clearly more dominant pattern is almost complete agreement. We assume that this is the case because openly expressing disagreement is often deemed socially inappropriate, especially in interactions with people one does not know (Tatarchevskiy 2012).
The rather narrow scope of the public discursive sphere as well as the low intensity and rather harmonious character of the communication taking place inside this arena of the deliberative system suggest that the reality of everyday political talk between strangers deviates considerably from the normative visions advocated by theorists of deliberative democracy and the political public sphere. To understand what facilitates or impedes this involvement, we examined the role of political and psychological motivations, skills, and opportunities (Verba et al. 1995;Huckfeldt and Sprague 1995).
Concerning political dispositions, our analyses at first sight confirm the standard diagnosis of a crucial role of general interest in politics as a motivator of citizens' talk about public affairs (Schmitt-Beck and Lup 2013). Closer inspection suggests an important relativization, however. The likelihood of access to the public discursive sphere is strongly diminished if individuals lack sufficient interest in politics. But beyond its generic relevance for everyday political talk as such, irrespective of arenas, political interest has no statistically discernible impact on the frequency and disagreeableness of conversations with strangers. Instead, the directional motivation of ideological extremity comes to the fore as the main political driver of involvement in this kind of everyday political talk. Even more importantly, our findings suggest that involvement in political conversations outside social networks depends more strongly on psychological than political dispositions. Individuals who feel socially isolated tend to stay outside this arena of the deliberative system. Socially trusting and conflict-seeking persons appear particularly prone to getting involved in the public discursive sphere of everyday political talk. Persons eager to form strong opinions tend to participate more frequently in such conversations.
Opportunities also appear important, though not similarly strongly. City life seems to facilitate access, as does a sociable lifestyle. Persons residing alone tend to discuss politics more frequently with unknown persons, presumably as a result of a less household-centered way of life. The more frequently such conversations occur, the more likely it in turn becomes that they expose their participants to political disagreements. Skills appear overall least important as conditions of everyday political talk with persons one does not know. Only internal efficacy is associated with more pronounced encounters with disagreement.
Apart from these results in line with our hypotheses, our models also detected several unexpected relationships. For instance, regarding psychological dispositions we found a high need for cognitive closure and a low need to belong to increase rather than decrease citizens' engagement in the public discursive sphere. Concerning skills, it appears that persons feeling not very competent to discuss politics, as well as the less educated and those who rarely follow news programs on public TV and who are therefore presumably not so well informed about politics, are more rather than less involved in the public discursive sphere. In a similar vein, we found economic well-being to be associated with weaker rather than stronger engagement.
The latter observation is interesting because it suggests a reversal of the wellknown, democratically problematic socio-economic upward bias of engagement in political talk for the special case of conversations with strangers (Schmitt-Beck 2022b). But in other respects, these observations raise doubts about the deliberative quality of the communication going on in this arena of the deliberative system. These doubts are further nurtured by some of the findings that were in line with our hypotheses. That ideological zealots and persons who quickly arrive at firm opinions because of their high need to evaluate engage particularly intensely in these conversations might undercut deliberative virtues like listening, compromising, and preference change. That interlocutors often seem to circumvent the clear expression of opinion differences may contribute to a more pleasant communication climate but preempts genuine deliberation's main purpose-constructively addressing the conflicts that divide societies (Martí2017).
Further research is needed to inquire whether these doubts are justified. Future studies should therefore explore the phenomenon of political talk with strangers in greater detail. Considerable, potentially consequential variability might exist in such communication experiences. Under which situational circumstances they arise, what kinds of discussion partners become accessible under these conditions, how the resulting conversations proceed, how interlocutors navigate them, and to which outcomes they lead, socially and politically, deserves closer inspection.

Supplementary Information
The online version of this article (https://doi.org/10.1007/s11615-023-00462-6) contains supplementary material, which is available to authorized users. acknowledges the support of the Deutsche Forschungsgemeinschaft (DFG) for the project whose findings are reported in this paper.
Funding Open Access funding enabled and organized by Projekt DEAL.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4. 0/.