TEORIJA IN PRAKSA let. 59, 1/2022 77 UDK [316.77:32]:077.5 Hajdeja IGLIČ* POSITIONING OF INDIVIDUAL AND THE MOBILISATION POTENTIAL OF POLITICAL DISCUSSION NETWORKS Abstract. In the article, we discuss how the political communication and interpersonal influence found in discussion networks add to fragmentation of the politi- cal space and the strengthening of political parties at the extremes of the political continuum. To this end, we analyse the mobilisation potential held by discussion networks of people who position themselves on the mar- gins of the left–right political spectrum and compare them with the networks of those positioned in the cen- tre and those who do not align themselves politically. The analysis focuses on four features of discussion net- works: size, homogeneity, frequency of political discus- sion, and frequency of trying to persuade others. The results of the analysis show the most extensive mobilisa- tion potential is found among people in the centre of the political continuum, while people on the far right and far left have networks that are the basis of intense politi- cal communication with people holding different politi- cal views but lack opportunities for the broad network dissemination of their political views and attitudes. Keywords: discussion networks, political communica- tion, left-right political orientation Introduction In recent years, major changes have been underway in Europe as con- cerns the structure of the political space in the direction of fragmentation and polarisation (Judis, 2016; Carothers and O’Donohue, 2019). Left and right parties in the political centre have been losing electoral power even though the majority of voters still locate themselves in the centre. The legiti- macy and stability of today’s multi-party systems are challenged by the dis- senting minorities represented by political parties found on the left and right margins of the political spectrum. * Hajdeja Iglič, PhD, Associate Professor, Faculty of Social Sciences, University of Ljubljana, Slovenia. DOI: 10.51936/tip.59.1.77-94 Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 78 The literature relates the changes on the political map we have been witness to the social fabric’s transformation in an age of “great reversal” (Rosanvallon, 2013) in which inequality in both income and wealth has almost returned to the very high levels seen one century ago. But the politi- cal views held by voters are both a reaction to changes in their objective circumstances and a result of political mobilisation engaged in by politi- cal actors, the media, and discussion networks. Studies show that discus- sion networks, namely the focus of this article, influence the individual’s political knowledge, political attitudes, voting behaviour etc. (e.g., Eveland, 2004; Eveland et al., 2005; Holbert et al., 2002; Huckfeldt and Sprague, 1992; Klofstad, 2007; Leighley, 1990; McClurg, 2003; Mutz, 2006; Nieuwbeerta and Flap, 2000; Nir, 2005). Although it is recognised that political discussion in one’s family, circle of friends, co-workers and acquaintances has been gradually losing its role to the mass media as a source of political informa- tion, the informal exchange of political views remains an important mecha- nism for the interpretation and evaluation of political news. This exchange occurs in the context of everyday communication within personal networks in which political topics are intertwined with subjects from personal and professional life (Bennett et al., 2000). In the article, we study the mobilisation potential of discussion networks of people found on the margins of the political spectrum to understand their role in strengthening the electoral power of extreme parties. We assume that an individual’s positioning on the margins of the political spectrum affects the key characteristics of their discussion network: the size and homogene- ity of the network, the frequency of political discussion, and the frequency of political persuasion. These network characteristics are also crucial for understanding mobilisation potential held by discussion networks work- ing via the processes of political influence: the size and heterogeneity of networks reflect the extensity (or range) of political communication, and the frequency of communication and political persuasion its intensity (or strength). Accordingly, we consider two specific questions in the analysis. First, are there any difference in the mobilisation potential of political networks of people at the extremes of the political spectrum compared to those located in the political centre or those that do not align themselves politically? Second, to what extent does the varying mobilisation potential of networks indeed depend on political positioning as opposed to other social and polit- ical characteristics of people, like political interest, age and education, politi- cal and social trust, exposure to the media etc.? An important assumption is made in our analysis that those who position themselves at the extremes of the political spectrum are still in a minority in the community which defines the dynamics of network formation. Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 79 The analysis considers data arising from an online survey carried out in Slovenia in 2017. Questions about political discussion were part of a longer questionnaire designed to collect detailed information about personal net- works in general, complemented with information about political attitudes and behaviour. Characteristics of political discussion networks and their mobilisation potential The analysis of political discussion networks is one area within a wider research programme in political communication. Typically, political discus- sion networks involve informal and two-way conversations in which, how- ever, no exact balance of influence is established between the discussants. Such political communication takes place in everyday situations and is found in the context of the family, friends, neighbours, and co-workers with whom the individual already has long-term and close relationships (Mutz and Mondak, 2006; Kotler-Berkowitz, 2005; Stoker and Jennings, 2005; Zuckerman et al., 2005). While political discussion is considered the most frequent and basic form of political action, it is also part of everyday life. This makes it unsurprising that political discussion on the individual level is influenced by both political and social factors. While our main variable with which we aim to explain the mobilisation potential of political discussion networks is the position in the political space, the control variables include various political factors and social factors. Classical studies of political discussion include the works of Lazarsfeld and colleagues written just after the Second World War (1944 [1968], 1954) and best known for the two-step flow model of communication concept with which they explained voting behaviour, and Festinger’s (1957) discus- sion of cognitive dissonance. Lazarsfeld’s occupation with the social con- ditions of political behaviour later contributed to the development of the discipline of political sociology (Lipset, 1960), while the concept of cogni- tive dissonance gained ground well beyond its original domain of social psychology, for example, in organisation studies (Simon, 1965). In both types of studies, the research assumed that the processes of political influ- ence occurring in social networks work through political discussion and the exchange of political views in which the individuals must confront and compare their political choices with the views held by other members in the network. Situations in which individuals find themselves when discussing politics with others can vary; the views of network members can be conso- nant or dissonant and, when dissonant, the individual’s views might repre- sent a majority or minority position in the network. Political discussion lies in the core of the normative model of democracy Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 80 where the role of citizens’ is to make key decisions in society, whether about concrete policies or about representatives who are to make decisions on their behalf. Mutz (2002) stresses that the beneficial consequences of politi- cal discussion for democratic political decision-making depend on meet- ing the conditions of the ‘ideal communication situation’, which typically involves the equality of partners, openness to interact with people holding different views, and in-depth two-way discussions. Any deviation from the ideal communication situation, due to the discussants being in unequal posi- tions, the network being strongly homogeneous or the discussion being superficial, can negatively impact society, including decision-making based on demagogy or a deepening of existing political disagreements. The mobilisation potential of political discussion is greater the bigger and more heterogeneous the discussion network is, the more frequently the individuals discuss politics with each other, and the more they strive to change their discussion partners’ views. The four characteristics of net- works mentioned above describe the extensiveness as well as the intensity of political networks. Mobilisation thus depends on the reach of political discussion network, as captured by the network’s size and heterogeneity, and on the strength of political discussion determined by the frequency of conversation and the persuasion. The studies by Mutz (2002a, 2002b) inform us that people might be high on some but not on all dimensions of the mobilisation potential and that various types of networks possess dif- ferent kinds of mobilisation potential. She suggests that homogeneous and smaller networks hold high mobilisation potential in the sense of stronger interpersonal pressure, which leads to the cementing of pre-existing views, forming a high level of political cohesion and considerable readiness for political action, including voting. On the other hand, people whose per- sonal networks have higher extensity, i.e., networks that are larger and more heterogenous, are less politically active and less likely to take political posi- tions or take them later in the process. However, their collective action has a greater range and connects different groups, mobilisation potential in this case stems from the cooperation of different groups. The heterogeneity of discussion partners is one of the most important and frequently studied network characteristics. From the perspective of democratic theory, a political discussion can reach its goal mainly when it occurs between individuals holding different views. The theory of cognitive dissonance states that political deliberation in everyday life is threatened by the avoidance of cognitive dissonance (Festinger, 1957), meaning that while talking about politics individuals attempt to lower the degree of diver- gence in their political views. They can do this in many ways: by subjectively minimising or supressing the perception of the extent of divergence, by try- ing to change the dissonant views of their partners, by avoiding political Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 81 discussion with people having opposing views, and by selectively choosing discussion partners with whom they share opinions. Studies confirm that the perceived and actual homogeneity of political viewpoints in networks is relatively high. However, heterogeneity still plays a role in political discussion. Research by Huckfeldt and colleagues (2004) shows that at the end of an election campaign no more than two-thirds of discussion partners held a presidential candidate preference that coincided with the preference of the respondent whose network was being observed (called “ego” in the network literature). Despite the tensions cognitive dissonance causes and the strategies people employ to avoid it, heterogeneity exists in personal networks. This raises the question of which forces enable heterogeneity to be maintained despite the well-known pressures towards conformity. The key part of the answer lies in the structure of personal networks. When networks include many weak ties and have a structure of cross-cutting social ties (Simmel, 1955) with a lot of structural holes (Burt, 1995), individual strategies for avoiding cognitive dissonance are less likely to result in a completely homogeneous network. Sparse and cross-cutting network structures with structural holes and weak social ties reduce the pressure due to the lack of a central ten- dency. While the pressure stemming from cognitive dissonance is universal, the strength of the pressure and thus individuals’ ability to cope with that pressure depend on the structure of their personal networks (Huckfeldt et al., 2005). Classical studies of cognitive dissonance presumed a dense group structure with many overlapping contacts, which allowed the conclusion that pressure for conformity always leads to a high degree of political homo- geneity. Hypotheses The following section present the empirical hypotheses about the rela- tionship between the individual’s position on the left–right political spec- trum and the characteristics of their political discussion networks, focusing on both their intensity and extensity. We shall examine the impact of politi- cal positioning on the four network characteristics mentioned above: size and homogeneity of the network, frequency of discussion, and frequency of persuasion. We should also mention what is not attempted in the arti- cle: our hypotheses do not deal with the content of political orientations, which are obviously different for the far-left and far-right political extremes. Since the preliminary analyses show the left–right distinction does not sig- nificantly affect the characteristics of political networks, we leave this aspect out of the analysis. Instead, we explain the mobilisation potential of net- works of people occupying the extreme political positions regardless of Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 82 ideology. We assume that extreme position is a minority position implying that people are in communities in which they live and work encircled by others who mainly hold different (majority) political views. The Size of the Network The impact of an extreme position on the size of a network must be understood with respect to the structural obstacles created by a smaller number of discussion partners holding similar views in the individual’s local environment. The assumption is that when choosing their discussion part- ners people consider the similarities in views (the cognitive dissonance the- sis). Differences between respondents who are positioned in the centre and those at the poles might be reflected in people in the centre having larger discussion networks because they have a larger choice or fewer structural limitations. On the other hand, as one empirical study reveals (Ikeda and Richey, 2012), people on the poles can compensate for the lack of similar discussion partners in their local environments by choosing partners in a more immediate social context – in the narrow circle of their family and friends instead of neighbours and acquaintances, because strong ties tend to entail a high level of homogeneity. When adding to this the possibility of finding discussants online, the impact of structural limitations might actu- ally prove to not be significant. Nevertheless, we formulate the hypothesis in the line with the classical argument: Hypothesis 1a: People positioned at the extremes of the political spec- trum have smaller political discussion networks than those positioned in the centre. Regarding those who are not politically aligned, we posit a hypothesis based on personal characteristics and spurious effect. We assume these indi- viduals have little political interest and knowledge, explaining why they also lack a clear political identification and are unable to position themselves. Moreover, the comparatively smaller size of their networks may be ascribed to the fact that they are generally less politically engaged. We propose: Hypothesis 1b: People who do not align themselves politically have smaller networks than those who hold a position (either in the centre or at an extreme). Homogeneity of networks The argument about the political positioning and homogeneity of net- works again stems from the assumption that extreme political positions rep- resent a minority. Due to their minority status, people perceive the majority that surrounds them as a threat, and they feel called to defend their views Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 83 and even more strongly connect with those holding similar views. This means that they censor political communication more than people in the centre because they use their political networks as a defence against the majority view (Huckfeldt et al., 1998; Mutz and Martin, 2001). This results in the networks of people at the political extremes being more homogene- ous than those of the people in the centre. On the other hand, it might be that due to the structural limitations they supress their views and engage in conversation with the majority without exposing the difference. Empirical studies corroborate the former view that while political networks are gener- ally homogeneous and tend to strengthen the individual’s prior beliefs, the selection process stands out even more in minority positions. Thus: Hypothesis 2a: The networks of people positioned at one end of the political spectrum or the other are more homogeneous than the networks of those positioned in the centre. The selection process is present the least among those who do not hold a position, which is why their networks are expected to be the least homo- geneous: Hypothesis 2b: The least homogeneous are the networks of people who do not position themselves on the political spectrum. Frequency of political discussion Frequency of political discussion relates to the individual’s politi- cal involvement and their positioning in the political space. According to Noelle-Neumann’s (1984) thesis on the spiral of silence, the minority and in turn those at the extremes of the political spectrum may be expected to silence their voice and contribute to the spiral of silence to avoid conflicts. In this case, social conformity would lead to less political discussion related to a specific position in the political space: Hypothesis 3a: Those positioned at one extreme or the other of the polit- ical spectrum discuss politics less often than those positioned in the centre. Individuals who do not hold any position at all are expected to be the least active in political discussion given their lack of political interest and knowledge. Hence: Hypothesis 3b: Those who do not align themselves politically participate the least frequently in political discussion. Pressure to accept one’s views Those who are the farthest from the majority views and attitudes are the ones who hold the strongest views, from which they do not withdraw easily and are not necessarily that quickly willing to be silenced when confronting Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 84 the majority public opinion. When they engage in political communication, they might be more eager to persuade others with the aim of expanding support for their views among the majority population. This thesis is cor- roborated by an empirical study (Glynn and McLeod, 1984). We thus pro- pose: Hypothesis 4a: Those who are positioned at either extreme of the politi- cal spectrum defend their views more strongly than those positioned in the centre. Like in the previous section on frequency of discussion, those who do not position themselves on the political spectrum are assumed to reveal the lowest intensity of political communication, which can be circumstantially explained by their smaller political interest and knowledge. Hypothesis 4b: Those not positioning themselves on the political spec- trum are the least active in defending their views. The survey Sample The survey was conducted in 2017 on a quota sample through online surveying. Quotas for the sample were prepared based on gender, age, edu- cation, and region, drawing on data for 2016 from the Statistical Office of the Republic of Slovenia. The survey included Slovenian people aged 18 to 65. When first sending the invitations to samples of panellists from the online panel, the sample framework was also defined by considering quota limitations and the expected response rate. When concluding the survey, the sending of additional invitations was adjusted to the information about individual groups that were missing from the survey, with the aim of achiev- ing the maximum correspondence of the sample with the population. The sample was also weighted. The sample consists of 1,001 respondents. Explained variables Questions about political discussion networks formed part of a wider battery of questions on personal or ego-centric networks. To identify the contacts with whom individuals discuss political topics, the following name generator was used: “Please think of the past 6 months and of persons with whom you have discussed politics during this time. (Including, for example, talking about elections, national and foreign politicians, and political events. Who did you talk to?)”. Respondents could list up to eight people, who they tagged with a name or a nickname. They were then asked about the char- acteristics of the people they had named and the frequency of conversing Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 85 with them, which helped us form the indicators that in our statistical models played the role of explained variables, the size and heterogeneity of net- works, frequency of political discussion, and frequency of trying to per- suade others to change their views. The size of a political discussion network is sum of all people that an individual listed while answering the name generator item. The measure of the homogeneity of political discussion networks was a question asking about the extent to which the political views of each of the respondent’s contacts or names that they had listed as partners in political discussions are similar to the respondent’s, namely: “Do you and this per- son hold similar political views? (Your best estimate will suffice)”. Similarity was estimated on a scale from 1 (Not at all similar) to 10 (Very similar). The average level of similarity between the respondent and the people they had mentioned was used for the purpose of the analysis on the respondent level. We measured the frequency of political discussion by asking how often a respondent talks to each person they had named: “How often do you talk to this person?”. Possible answers were chosen on a scale from 1 (Less than once a month), 2 (At least once a month), 3 (At least once a week) to 4 (Every day or almost every day). This variable was treated as an interval vari- able and its average value was used in the analysis. The question about how often the individual puts pressure on others should they hold opposing views was formulated as follows: “When you have a very strong view about a certain political issue, do you try to per- suade your friends, relatives, neighbours, or co-workers to adopt your view? How often do you do this?”. Responses are scaled from 1 (Never or almost never) to 5 (Very often). Explanatory variables The main explanatory variable positioning in the political space was formed based on the following question: “In politics, we typically speak about the “left” and the “right”. Where would you place your views on a scale from 1 (Very left) to 10 (Very right)?”. Respondents who chose answers situ- ated on the far left (values 1 or 2) and far right (values 9 or 10) were defined as holding extreme positions, while the rest were positioned in the centre. Respondents were also allowed to choose “I do not know” and “I do not wish to answer”. Those who answered with “I do not know” are considered as not aligned politically because they do not have a clear political identifi- cation. Respondents who answered “I do not wish to answer” were ascribed the missing value. Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 86 Control variables The analyses also include several control variables shown by previous research to affect the characteristics of political discussion networks. The first set of controls contains demographic variables: gender, age, and educa- tion. Age was re-coded to three age classes: 1 (18 to 30 years of age), 2 (31 to 55 years of age) and 3 (56 to 65 years of age). Education was re-coded to five education classes: 1 (incomplete or complete elementary school), 2 (sec- ondary vocational or technical education), 3 (secondary general education), 4 (undergraduate education) and 5 (postgraduate education). Alongside the demographic variables, a measure of social trust was used, on a scale from 1 (Almost none) to 10 (Very much). Since social trust influences one’s ten- dency to establish contacts with other people generally, it is assumed to also be important for understanding the respondents’ contacts with others for the purpose of political discussion. Political control variables include variables like political interest, follow- ing political news in the media and self-assessment of political knowledge and skills called self-assessed political competencies. The first was measured with the question: “In general, how much are you interested in politics? Please describe your political interest on a scale from 1 (I am not interested at all) to 10 (I am very interested)”. With respect to the self-assessed compe- tencies, respondents were asked: “Do you think you possess adequate com- petence and knowledge to participate in politics?”, which they answered on a scale from 1 (No, not at all) to 10 (Yes, completely). Regarding actual politi- cal competencies, we were interested in how often they were following the political news in the mass media. Answers were on a scale from 1 (Never or almost never) to 5 (Every day). Since they were asked about how often they followed political news in different kinds of media (the press, radio and television, the Internet), data were reduced by conducting principal com- ponents analysis (PCA) which gave us a single dimension. Regression esti- mates of factor scores were used as values of the new variable. Empirical analysis The political discussion networks of our respondents are small, which corresponds to findings in other surveys (Byungkyu and Bearman, 2017). Having one or more political discussion partners was declared by as few as 36.5% of respondents. Of these, 61.4% mentioned one, 23.0% two, and all the others three or more of such partners. On average, each respondent mentioned just 0.5 of a discussion partner. The frequency of discussions, the similarity of views, and frequency of trying to persuade others were calculated only for respondents who had named at least one discussion Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 87 partner. On a scale from 1 to 4, the average value of the frequency of discus- sion of the average respondent whose political network size was estimated at one or more amounts to 1.76. This means that they often talk politics with their discussion partners: ranging between every day and at least once a week. The people with whom they discuss politics hold very similar views to their own, although the similarity is far from complete. On a scale from 1 to 10, the average similarity level is 7.71. The respondents are, on average, quite active in trying to persuade others to adopt their views. Those who mentioned at least one person with whom they discuss politics often try to persuade this particular person that they are right. The average value is 2.21, corresponding to the answer “frequently”. Three analyses were carried out for each characteristic of the political discussion networks, with the results being presented in three columns in Tables 1–4. The first column shows the statistical relationship between the network characteristics and the positioning in the political space. The sec- ond column shows the results when socio-demographic variables are added to the model (gender, age, education, social trust), while the remaining col- umn shows the results when political variables are included (political inter- est, media exposure, self-assessed political competencies). In all statistical analyses, being located at an extreme position (i.e., far-right or far-left) rep- resents a base category. The size of the network The political discussion network is at its greatest size among respond- ents in the centre of the political space and (Table 1). The results of the bivariate analysis seem to support Hypothesis 1a. Adding social factors to the model shows that network size also depends on age, education, and social trust. The expected positive effect of age is confirmed, noting that the upper age limit of our survey is 65 years, which is why, unlike other surveys, a curvilinear effect of age is not obtained. Education also has a posi- tive effect on the size of the network, together with social trust. Thus, older, better educated people and those with a higher level of social trust discuss politics with a larger number of social ties than others. Still, the influence of socio-demographic factors disappears when political factors are added to the model, which means social factors influence political discussion net- works only indirectly by affecting political factors, namely, political interest and media exposure. Also, the effect of political positioning weakens and becomes statisti- cally non-significant when political variables are added to the model. Being well informed about politics and having a strong political interest explain the variation in the size of the networks. This means that the relationships Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 88 between one’s positioning and the size of one’s networks is spurious. Older and more educated people who are positioned in the centre have a stronger political interest and more frequently follow political news in the media, causing them to have larger political discussion networks. This leads us to reject Hypothesis 1a. On the other side there are respondents who do not align themselves politically. As expected, and in line with Hypothesis 1b their networks are the smallest. This is again explained by political variables such as their lack of political interest and low exposure to political news in the media. Table 1: REGRESSION OF THE SIZE OF POLITICAL DISCUSSION NETWORKS ON SOCIO-DEMOGRAPHIC AND POLITICAL VARIABLES Model 1 Model 2 Model 3 ß ß ß Extreme base base base Centre *0.072* 0.072* 0.053 Don’t know **-0.097** -0.069* -0.025 Gender (male = base) 0.009 0.005 Age *0.069* 0.035 Education * 0.063* 0.030 Social trust * 0.062* 0.050 Political competencies 0.009 Media exposure **-0.113** Political interest *** 0.197*** R 2 0.023 0.035 0.093 N 1000 997 997 Source: Own analysis. Homogeneity of networks Next, we look at the factors that impact network homogeneity in terms of similarities of political views between the respondents and their discus- sion partners (Table 2). The analysis of the homogeneity of networks was performed on a smaller number of respondents, namely, only those who have at least one partner in the network with whom they discuss politics. The results of the analysis show that positioning in the political space has a strong impact on the similarity of political discussants. Individuals whose views are defined and located at the poles have more homogenous net- works. The influence of political positioning does not disappear when we Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 89 add social and political variables to the model. This supports Hypothesis 2a stating that people holding extreme political views censor political commu- nication more than those in the centre, enclose themselves in ‘echo cham- bers’ and use political networks as a defence against the majority view. The least homogeneous are the discussion networks of those who do not align themselves politically, as suggested by Hypothesis 2b. The only other variable explaining homogeneity of networks is gender. The networks of women are more homogeneous than those of men. Further analy- sis would show whether this is linked to the choice of discussion partners. It could be that women tend to discuss politics in a narrower circle of close social ties than men, which tend to be more similar than distant social ties. Table 2: REGRESSION OF THE HETEROGENEITY OF POLITICAL DISCUSSION NETWORKS ON SOCIO-DEMOGRAPHIC AND POLITICAL VARIABLES Model 1 Model 2 Model 3 ß ß ß Extreme base base base Centre *-0.139* *-0.139* *-0.137* Don’t know **-0.197** **-0.188** *-0.147* Gender (male = base) *0.112* *0.128* Age *0.091* 0.072 Education 0.040 0.025 Social trust -0.013 -0.014 Political competencies -0.016 Media exposure -0.079 Political interest 0.063 R 2 0.029 0.052 0.062 N 350 349 349 Source: Own analysis. Frequency of political communication The frequency of discussion is the first of the two measures of the inten- sity of political communication. Table 3 shows that communication is most frequent among respondents in the centre, followed by respondents at the extremes of political communication. This seems to support Hypothesis 3a stating that those holding extreme political views are, because of their minority position, silenced in the political discussion compared to those in the centre of the political space. Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 90 However, the differences between groups disappear once we include social and political variables in the model. The frequency of political discus- sion is a result of a stronger political interest, education, and age. Older people and those with a higher education have a stronger political interest and more often discuss politics with others. Hypothesis 3a thus must be rejected. There is no effect of an extreme position on the frequency of political discussion. Non-positioned respondents have a statistically significant lower fre- quency of political discussion than the other two groups, which is due to lower education, trust, and political interest as predicted by Hypothesis 3b. Table 3: REGRESSION OF THE FREQUENCY OF POLITICAL DISCUSSION ON SOCIO-DEMOGRAPHIC AND POLITICAL VARIABLES Model 1 Model 2 Model 5 ß ß ß Extreme base base base Centre *-0.119* -0.032 0.040 Don’t know *-0.110* -0.093 -0.070 Gender (male = base) -0.012 0.028 Age 0.080 * 0.092* Education ** 0.166** ** 0.147** Social trust * 0.093* 0.084 Political competencies 0.007 Media exposure -0.010 Political interest ** 0.226** R 2 0.012 0.048 0.078 N 364 363 363 Source: Own analysis. Frequency of persuading others The frequency with which one tries to persuade others is again a meas- ure of the intensity of political discussion. Respondents at the poles are much more likely to try to persuade others, thereby confirming Hypothesis 4a. Central and especially non-positioned people put much less pressure on others to adopt their own political views, giving support for Hypothesis 4b. Positioning in the political space also influences the effort to persuade others when control variables are included in the model. Socio-demographic variables have no influence on the persuasion effort; however, the fre- quency of trying to persuade others is linked to a stronger political interest Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 91 and a higher level of self-assessed political competence. People who very often try to persuade others believe they have high levels of political knowl- edge and skills, controlled for other factors. Table 4: REGRESSION OF THE PERSUASION EFFORT ON SOCIO-DEMOGRAPHIC AND POLITICAL VARIABLES Model 1 Model 2 Model 3 ß ß ß Extreme base base base centre *-0.129* *-0.131* *-0.110* Don’t know **-0.200** **-0.181** *-0.100* Gender (male = base) 0.008 0.043 Age 0.010 0.007 Education 0.035 0.008 Social trust 0.039 0.030 Political competencies * 0.137* Media exposure -0.052 Political interest ** 0.201** R 2 0.028 0.032 0.074 N 364 363 363 Source: Own analysis. Conclusion The results of the analysis show significant differences among the polit- ical discussion networks of the three groups under study with respect to their mobilisation potential. The networks of people who do not align them- selves politically have small range (extensiveness) and intensity of political communication; their networks are small, political communication is less frequent, and they less often seek to persuade others to adopt their political views. However, their networks are the most heterogeneous of all, meaning that due to them only sporadically engaging in political discussion they are less selective in their choice of political discussants. The political discussion networks of this group of respondents reflect their general political disen- gagement. Additional analysis (not presented here) shows that, compared to the other two groups, they are less educated, have lower social trust, lower political interest, lower self-assessed competence, and less frequently follow political news in the media, which all decrease mobilization potential of their political discussion networks. Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 92 Very differently, networks of people in the centre of political space are large and they frequently discuss politics with the discussion partners who hold different political views. However, in this communication, they avoid putting pressure on their partners to change their views. We can character- ise their networks as having high range and moderate strength. Such net- works are characteristic of people who have higher levels of education and social trust as well as higher levels of political engagement in terms of politi- cal interest and following the media. Finally, people who position themselves on the far-right or far-left have networks with moderate range and high intensity. Their networks are rela- tively small, although not as small as the networks of the first group (those who are not politically aligned), and very homogeneous. They less often discuss politics with others, albeit again more often than the first group. Most important, when they do discuss politics, they very often try to per- suade their discussion partners. In terms of personal characteristics, they have a low level of education, rather weak political interest and are also less likely to follow political news in the media. However, they do have a very high level of self-assessed political competencies, which causes some incon- gruity with respect to the objectively measured political competence indi- cated by education and media exposure. Thus, people on the far right and far left have networks that are the basis of intense political communication with people holding different political views but lack opportunities for the broader network dissemination of their political views and attitudes. Their networks also make them ready for political action, also a high-profile one. We may conclude that positioning in the political space matters for the structuring of one’s political discussion network and the way one conducts political conversations and influences others, although the impact is not as strong as expected. While political orientation affects the heterogeneity of discussion partners and persuasion, it has no effect on the size of the net- works and frequency of political discussion. Positioning at the extremes leads to more homogeneous networks due to the selection process, since people avoid discussing politics with those who hold different political views. Moreover, when those in the extreme political positions who gen- erally less often engage in political discussion, begin a political conversa- tion with someone they do not agree with, they defend their position and attempt to overcome the cognitive dissonance by persuading their discus- sion partners as to the correctness of their views. Both effects speak to the importance of the cognitive dissonance mechanism for the network dynam- ics among people at the extremes of political spectrum. Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 93 BIBLIOGRAPHY Beck, Paul Allen, Russell J. Dalton, Steven Greene and Robert Huckfeldt (2002): The Social Calculus of Voting: Interpersonal, Media, and Organizational Influences on Presidential Choices. American Political Science Review 96: 57–73. Bennett, Stephen E., Richard S. Flickinger and Staci L. Rhine (2000): Political Talk Over Here, Over There, and Over Time. British Journal of Political Science 30: 99–119. Berelson, Bernard R., Paul F. Lazarsfeld and William N. McPhee (1954): Voting: A Study of Opinion Formation in a Presidential Campaign. Chicago: Chicago University Press. Burt, Ronald S. (1995): Structural Holes: The Social Structure of Competition. Cambridge, Mass.: Harvard University Press. Carothers, Thomas and Andrew O’Donohue (2019): Democracies Divided: The Global Challenge of Political Polarization. Washington: The Brookings Institute. Eveland, William P. Jr. (2004): The Effect of Political Discussion in Producing Informed Citizens: The Roles of Information, Motivation, and Elaboration. Political Communication 21: 177–193. Eveland, William P. Jr., Andrew F. Hayes, Dhavan V. Shah and Nojin Kwak (2005): Understanding the Relationship between Communication and Political Know - ledge: A Model Comparison Approach Using Panel Data. Political Commu- nication 22: 423–446. Festinger, Leon (1957): A Theory of Cognitive Dissonance. Stanford: Stanford University Press. Holbert, R. Lance, William Benoit, Glen Hansen and Wei-Chun Wen (2002): The Role of Communication in the Formation of an Issue-based Citizenry. Commu- nication Monographs 69: 296–310. Huckfeldt, Robert and John Sprague (1987): Networks in Contexts: The Social Flow of Political Information. American Political Science Review 81: 1197–1216. Huckfeldt, Robert and John Sprague (1992): Discussant Effects on Vote Choice: Intimacy, Structure and Interdependence. Journal of Politics 53: 122–158. Huckfeldt, Robert and John Sprague (1995): Information, Persuasion and Political Communication Networks. In Russell J. Dalton and Hans-Dieter Klingeman (eds), The Oxford Handbook of Political Behavior, pp. New York: Oxford University Press. Huckfeldt, Robert, Paul E. Johnson and John Sprague (2004): Political Disagreement: The Survival of Diverse Opinions within Communication Networks. Cambridge: Cambridge University Press. Huckfeldt, Robert, Paul E. Johnson and John Sprague (2005): Individuals, Dyads, and Networks: Autoregressive Patterns of Political Influence. In Zuckerman, A.S. (ed.), The Social Logic of Politics: Personal Networks as Contexts for Political Behavior, 21–50. Philadelphia: Temple University Press. Ikeda, Ken’ichi, Sean Richey (2012): Social Networks and Japanese Democracy: The Beneficial Impact of Interpersonal Communication in East Asia. New York: Routledge. Hajdeja IGLIČ TEORIJA IN PRAKSA let. 59, 1/2022 94 Judis, John B. (2016): The Populism Explosion: How the Great Recession Trans- formed American and European Politics. New York: Columbia Global Report. Klofstad, Casey A. (2007): Talk Leads to Recruitment: How Discussions About Politics and Current Events Increase Civic Participation. Political Research Quarterly 60: 180–191. Lazarsfeld, P. F., B. R. Berelson and H. Gaudet ([1944] 1968): The People’s Choice: How the Voter Makes up His Mind in a Presidential Campaign (3 rd ed.). New York: Columbia University Press. Lipset, Seymour Martin (1960): The Political Man: The Social Bases of Politics. London: Doubleday & Company. McClurg, Scott D. (2003): Social Networks and Political Participation: The Role of Social Interaction in Explaining Political Participation. Political Research Quarterly 56: 449–464. Mutz, Diana C. (2002a): The Consequences of Cross-Cutting Networks for Political Participation. American Journal of Political Science 46: 838–855. Mutz, Diana C. (2002b): Cross-cutting Social Networks: Testing Democratic Theory in Practice. American Political Science Review 96: 111–126. Mutz, Diana C. (2006): Hearing the Other Side: Deliberative Versus Participatory Democracy. Cambridge: Cambridge University Press. Mutz, Diana C. and Jeffery J. Mondak (2006): The Workplace as a Context of Cross- cutting Political Discourse. The Journal of Politics 68: 140–155. Neuwbeerta, Paul and Henk Flap (2000): Crosscutting Social Circles and Political Choice: Effects of Personal Network Composition on Voting Behavior in the Netherlands. Social Networks 22: 313–325. Nir, Lilach (2005): Ambivalent Social Networks and Their Consequences for Participation. International Journal of Public Opinion Research 17: 423–442. Simmel, George (1955): Conflict and the Web of Group Affiliations. New York: Free Press. Simon, Herbert (1965): The Shape of Automation for Men and Management. New York: Harper & Row. Stoker, Laura and M. Kent Jennings (2005): Political Similarity and Influence between Husbands and Wives. In Zuckerman, A. S. (ed.), The Social Logic of Politics: Personal Networks as Contexts for Political Behavior, 51–74. Phila- delphia: Temple University Press. Verba, Sidney, Kay Lehman Schlozman and Nancy Burns (2005): Family Ties: Understanding the Intergenerational Transmission of Political Participation. In Zuckerman, Alan S. (ed.), The Social Logic of Politics: Personal Networks as Contexts for Political Behavior, 95–116. Philadelphia: Temple University Press. Zuckerman, Alan S., Jennifer Fitzgerald and Josip Dasović (2005): Do Couples Support the Same Political Parties? Sometimes: Evidence from British and German Household Survey. In Zuckerman, Alan S. (ed.), The Social Logic of Politics: Personal Networks as Contexts for Political Behavior, 75–95. Phila- delphia: Temple University Press.