UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Living rooms and coffee shops as deliberative forums : informal political conversations and the practice… Owen, Andrew 2002

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_2003-0175.pdf [ 4.76MB ]
JSON: 831-1.0090927.json
JSON-LD: 831-1.0090927-ld.json
RDF/XML (Pretty): 831-1.0090927-rdf.xml
RDF/JSON: 831-1.0090927-rdf.json
Turtle: 831-1.0090927-turtle.txt
N-Triples: 831-1.0090927-rdf-ntriples.txt
Original Record: 831-1.0090927-source.json
Full Text

Full Text

LIVING ROOMS A N D COFFEE SHOPS AS DELIBERATIVE FORUMS: INFORMAL POLITICAL CONVERSATIONS A N D THE PRACTICE OF DELIBERATIVE D E M O C R A C Y IN THE 1997 C A N A D I A N F E D E R A L ELECTION by A N D R E W O W E N B A . (Honours), Queen's University at Kingston, 1999 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF THE REQUIREMENTS FOR THE DEGREE OF M A S T E R OF ARTS in THE F A C U L T Y OF G R A D U A T E STUDIES (Department of Political Science) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH C O L U M B I A October 2002 © Andrew Owen, 2002 ABSTRACT Theories of deliberative democracy perceive decision-making processes as truly democratic only when choices are publicly considered by all members of a society in an effort to serve the interests of a community as a whole. The size of modern political units, however, presents a tremendous obstacle to the practice of such an approach since participation by all members of a society in a single formal deliberative setting is unrealistic. This paper seeks to assess the degree to which informal political conversations emulate those envisioned in formal deliberative theory. Data collected in the 1997 Canadian Election Study is employed to assess the degree to which the processes and impacts of casual discussions about the campaign match the expectations described in deliberative theory. The results presented are equivocal as informal political discussions do have some effects on relevant political behaviours but these effects are weak and inconsistent. The concluding chapter discusses the value of competing explanations for these findings. Namely, informal conversations are significantly different from formal deliberations, and/or, the methods employed failed to adequately measure the key concept under investigation. T A B L E OF C O N T E N T S Abstract 1 1 Table of Contents i i i List of Tables iv List of Figures v Preface 1 CHAPTER I Introduction 2 CHAPTER II Theoretical Foundations 5 Democracy and Deliberation 5 Good Talk/Bad Talk 11 Talk and Cognition 13 CHAPTER III The Antecedents of Informal Discussion 21 CHAPTER IV The Nature of Informal Discussion 34 CHAPTER V The Impacts of Informal Discussion 46 Political Knowledge 48 Information Acquisition During the Campaign 54 Opinionation 55 Opinion Consistency 59 Efficacy 62 Interest in the Campaign 65 Political Participation 68 Tolerance 71 CHAPTER VI Conclusion 75 Bibliography 82 LIST OF T A B L E S Table 3.1 Discussion frequency .25 Table 3.2 Antecedents of discussion frequency, demographics only 28 Table 3.3 Antecedents of discussion frequency 29 Table 3.1 Media use and discussion frequency 31 Table 4.1 Explaining frequency of disagreement 40 Table 4.2 Gender and perceptions of political competence 41 Table 4.3 Party supporters and perception of local electoral success 43 Table 5.1 Determinants of political knowledge 52 Table 5.2 Estimates effects of discussion on opinionation 58 Table 5.3 Estimating effects of discussion on opinion consistency 61 Table 5.4 Estimating effects of discussion on internal efficacy 64 Table 5.5 Estimating effects of discussion on interest in the campaign 67 Table 5.6 Estimating effects of discussion on voting behaviour 71 Table 5.7 Estimating effects of discussion on tolerance 73 LIST OF F I G U R E S Figure 3.1 Frequency of discussion 27 Figure 4.1 Frequency of disagreement 38 Figure 5.1 Comparing frequency of discussion 53 "Yet talk remains central to politics, which would ossify completely without its creativity, its variety, its openness and flexibility, its inventiveness, its capacity for discovery, its subtlety and complexity, its eloquence, its potential for empathy and affective expression, and its deeply paradoxical (some would say dialectical) character that displays man's full nature as a purposive, interdependent, and active being." Benjamin Barber, Strong Democracy (Berkeley: University of California Press), p.173. "Oh, conservative. Liberal and conservative. Liberal and conservative. I haven't given it much thought. I wouldn't know. I don' know what those words mean! Liberal....liberal...liberal...liberal. And conservative. Well, i f a person is liberal with their money they squander their money? Does it fall into the same category? If you're conservative you don't squander so much, you save a little, huh?" Unnamed survey respondent in W. Russell Neuman, The Paradox of Mass Politics (Cambridge, M A : Harvard University Press), p. 20. CHAPTER 1: Introduction Citizens matter. While the behaviour of those subject to political decisions is a legitimate topic of concern in all societies, in a democracy, understanding the citizen is crucial given the impact public opinion has on decision making. The origins and development of public opinion are the subject of a vast body of literature. An important component of this literature deals with the sources of political information citizens tap in constructing their knowledge of, and beliefs about, politics. While the mass media has received a great deal of well-deserved attention, significantly less effort has been devoted to the study of other citizens as a source of political information. Scholars such as Elihu Katz and Robert Huckfeldt raised the profile of interpersonal communication as a worthy subject of investigation and, more recently, debates over deliberative theory have motivated a number of efforts to better understand the effects of political discussions. In the Canadian case, however, political conversations remain largely unexamined. Deliberative theory offers a dispute resolution process distinctly different from both voting and bargaining. Through informed and reasoned debate political actors can reach consensus-based decisions. Deliberative theorists further argue that all those affected by a decision ought to participate in the conversations and debates through which decisions are reached. Apart from being a more honest application of democratic theory, greater participation in decision making is also one solution to growing political apathy and alienation. Concerns about entrusting important decisions to average citizens, however, persist. The underwhelming political knowledge of most citizens is well documented (Delli Carpini& Keeter, 1989). Much of deliberation's value stems from the effects of participation in these forums on the political knowledge and opinions of those who take part. Active participation in the contemplation of a topic brings with it enhanced understanding of the topic. Citizens can become qualified to take part in decision making through the course of deliberation. Some of the most potent criticisms of deliberative theory focus on the logistical impossibility of deliberations among an entire society. Representative democracy seeks to solve this problem of scale by selecting certain individuals and providing them with access to deliberative forums through which decisions are made. The selection of these 2 participants, however, occurs through the aggregation of privately expressed preferences. While some creative ideas such as deliberative polling (Fishkin, 1995), and citizen juries offer potential solutions to the problem of scale, a fully functioning deliberative society seems beyond reach. It is still possible, however, that informal conversations about politics occurring outside the purview of the state might, to a greater or lesser degree, act as a substitute for formal deliberations. Again, deliberation can be of value both by producing more informed and sophisticated citizens who are therefore better suited to influence political decisions, and by resulting in individual opinions which are more truly democratic since the lives of diverse fellow citizens inform the process of opinion development. This paper seeks to assess the extent to which informal conversations accomplish these broad goals. In what follows I draw on deliberative theory and cognitive psychology to identify aspects of formal deliberation against which informal discussions can be empirically measured. The evidence reported here is equivocal. The results suggest that discussion is indeed a worthwhile topic of study as frequency of participation in political conversations predict a number of politically relevant behaviours. Not surprisingly, however, the consequences of casual political conversations prove to be distinctly different from those expected of the formal deliberations prescribed in deliberative theory. The following chapter presents the theoretical underpinnings of my research. I briefly summarize the central features of deliberative democracy, attempt to define what type of conversations are truly deliberative, and comment on the cognitive implications of deliberative discussions. Chapters 3-5 contain my empirical research. The third chapter considers the antecedents of conversation and seeks to idenitfy the characteristics of those who participate in casual political conversations. Chapter 4 deals with the process of informal discussion and assesses the extent to which conversations contain disagreement and involve minority points of view. In chapter 5,1 conduct a number of different tests to measure the effects of participation in these conversations. Using deliberative theory as a guide, I identify a number of expected outcomes which serve as dependent variables. Chapter 6 surveys the results and presents my conclusions. Before 3 moving to a consideration of deliberative theory, I first provide details of the data used to conduct this research. Data: In this paper I employ data gathered during the 1997 Canadian Election Study (Nevitte, Blais, GidengikM, & Nadeau, 1999). The Institute for Social Research at York (ISR) conducted this three part survey in which the same respondents are asked political questions once during, and twice after, the federal election campaign. The first wave entails a rolling cross section with around 80 individuals surveyed on each of the 36 day election campaign. The ISR interviewed a total of 3,949 people between April 27 to June 1, 1997. The second wave involved a telephone reinterview of initial respondents with a response rate of just over 80% totalling 3,170 cases. Finally, a third component asked respondents to complete and return a mail back questionnaire. Of the 2600 who provided their address during the post-election wave, 1,857 completed the survey. As shall become clear in the following discussion of the variables used to operationalize certain key concepts, the CES is a rather blunt tool for investigating this particular topic. This said, the CES is a valuable resource for a number of reasons. The fairly large number of cases allows for examination of specific sub-populations without diminishing the potential for statistically significant results. In addition, the panel aspect of the survey enables measurement of change within respondents over the course of the campaign. Finally, the breadth of the survey presents opportunities for empirical research to those prohibited from conducting their own survey due to a lack of resources. Given the scope of the present study, the CES serves as a useful tool. This said, I shall be careful in drawing conclusions based on the findings contained here since better research instruments exist. 4 CHAPTER 2 ; Theoretical Foundations In this chapter I seek to I outline the theoretical foundations upon which my subsequent empirical work is based. After briefly summarizing the central features of deliberative theory, I offer four ways through which political conversations enhance democracy. In order to further illuminate the ideal type of interpersonal communication against which actual conversations are to be measures, I then speak to two fundamental aspects of conversation. Clarifying the 'who' and the 'how' of political discussions provides an improved understanding of this paper's key concept. Finally, I devote specific attention to the relationship between deliberation and cognition since much of the proposed value of talk stems from the cognitive implications of participation in a discussion. Democracy and Deliberation Emphasis on the important contribution of talk to democracy dates as far back as the thought of Aristotle and received specific attention by modern thinkers including Bryce, Tarde, and Dewey. The theoretical debates surrounding notions of deliberative democracy, however, have increased rather dramatically in more recent years. In the mid to late 1980's, Jurgen Habermas' work on the early bourgeois sphere first drew attention to public deliberation while Joshua Cohen helped expose these ideas to the North American audience. Since that time, a significant number of political theorists have devoted time to the study of public deliberation and deliberative democracy. Despite the fact that there are a number of intriguing debates within the field, space does not permit a full consideration of these topics. Rather, the following section summarizes the work of Cohen and Barber in an effort to tease out the aspects of deliberative theory that inform the empirical analysis at the heart of this research. Joshua Cohen argues deliberation is a precondition of democracy: "the fundamental idea of democratic legitimacy is that the authorization to exercise state power must arise from the collective decisions of the members of a society who are governed by that power" (Cohen 1996, p.95). This definition of democracy does not appear immediately controversial. What makes deliberative theory different, however, is the belief that the process through which collective decisions are made must be collective itself. Decisions about the operation of society are not to arise from the mere aggregation 5 of individual preferences, but through collective discussion resulting in decisions made by the citizenry as a whole. Through such a process the wishes of 'the people' become something greater than the sum of all individuals' desires. The content of these discussions must also be motivated by a public rather than private concern. Citizens are to offer and defend competing versions of the common good rather than bargain over policies from the perspective of their own self-interest. In short, democracy for Cohen exists when the people reach decisions by transforming preferences informed by self interest into preferences motivated by a concern for the good of society as a whole. Cohen acknowledges that such a conception of democracy rests in part on an ideal of active citizenship for a deliberative democracy cannot function unless each citizen devotes a considerable amount of time and effort to politics. (Cohen, 1989). Benjamin Barber shares with Cohen a belief in the importance of active citizenship by advocating 'strong democracy'. While Barber is rarely referred to as a deliberative theorist, formal and informal deliberation between citizens is central to the conception of democracy he advocates in Strong Democracy: "at the heart of strong democracy is talk" (Barber, 1984, p. 173). Barber and Cohen also share the belief that public decisions should be motivated by a concern for the society as a whole and not by private interests. In critiquing Hobbes, Bentham, and Rawls, Barber writes "These formulas, based on right, utility, and fairness, denied the possibility of a public good that was more than an aggregate of individual and particular goods." (Barber, 1984, p. 171). Like Cohen, Barber asserts than an individual must understand 'we' as more important than T . Citizens are able to arrive at decisions in their collective interest by talking with one another. For Barber, Cohen, and others covered by the deliberative democracy rubric, interpersonal communication between citizens is of fundamental importance to the effective functioning of democracy. Thus, any effort to test the impact of talk on democracy must first specify the manner in which talk improves democracy. Below I outline four ways in which political discussion between citizens ought to improve the quality of democracy in an ideal setting. The four points are inherently interrelated. 6 Preference Transformation The act of discussing politics enhances the potential for the transformation of preferences central to deliberative democracy. In order for an opinion or attitude to change, a person must be cognitively engaged with the subject to some degree. Talking about politics requires thought about political matters which is a first, and rather obvious, step to transformation. Beyond simply directing one's attention to politics, the specific nature of talk facilitates opinion transformation in a number of ways. To talk is to be actively engaged with the subject matter. Since people tend to be cognitive misers, motivation to expend the cognitive effort to consider a complex topic such as politics is necessary in order to create the conditions for attitude change. Talking with another person provides such motivation in various ways including the need to respond to a discussion partner's comment and the social desirability of providing a useful response. With respect to the later point, the penalties for failing to offer a reasonable response and/or ignoring a valid point that counters one's initial opinion are much greater in interpersonal communication than when receiving messages from a source to which one need not, and cannot, respond. A later section of this chapter considers the relationship between cognition and discussion in greater depth. Public Opinion: In addition to providing a situation favourable to preference transformation, discussion involves exposure to the kind of information an individual requires when seeking to construct opinions motivated by the common good. Talking with others is of fundamental importance for the citizen in her efforts to consider political matters from a perspective distinct from that of her own private interests. Public opinion, broadly conceived, is the set of preferences that are to inform and affect decisions in a democracy. As mentioned above, public opinion in a deliberative democracy is established when citizens discuss and agree upon how best to serve the needs of the polity as a whole. At the opposite end of the spectrum, public opinion is simply the aggregation of privately developed and self interested preferences. Price and Neijens point out that classical theorists such as Bryce, Park, Laswell, and Lazarsfeld conceived of public opinion as "an emergent product of widespread discussion—emanating ideally 7 from debate that is open to wide popular participation, free flowing and uncensored, and well informed" (Price and Neijens, 1997, 336). In other words, public opinion can only exist after onions are subject to public deliberation. Deliberation itself serves as an important source of information for individuals. Classical democratic theory assumes informed citizens. Deliberative theory assumes that to become informed, citizens must interact with each other. Information gathered through deliberation is information about the collective. By discussing an issue with others, an individual is better able to understand the manner in which an issue affects those different from him and thus the collectivity as a whole. While discussion may certainly help, an individual need not talk with others in order to decide what policy is in her best interest. On the other hand, deciding what is in the best interest of the polity as a whole is not possible without interpersonal communication with others. But there are at least two possible objections to this line of argument: First, some authors such as Diana Mutz suggest that the mass media can replace conversation in this respect (Mutz, 1998). Second, others, such as Schudson, note that not all conversation enhances public opinion since discussion between those with similar life experiences may serve to reinforce beliefs without subjecting them to counter arguments (Schudson, 1997). Both sets of concern shall be discussed shortly. Empathy is a primary means through which the interests of others become relevant to an individual's political opinions. In a deliberative setting where differing group perspectives comprise much of the discussion, empathy becomes of crucial importance for it enables those of different social positions to attempt to understand the positions and perspectives of others: Empathy makes it easier to understand other's needs, and by reflected understanding, one's own. Empathy helps surmount conflicting interests by allowing participants to make another's good their own. Empathy even helps in reaching agreement on principled convictions because understanding the other's feelings makes it easier to forgive their intransigence or slips of logic and to make one's point again in language they can hear. Empathy is critical in good negotiations, because making suggestions that meet another's needs at less cost to one's own requires emotional and cognitive insight into the other's needs (Mansbridge, 124). 8 While one might empathize with abstract individuals affected by a given problem, the potential for empathy seems much greater when one is directly interacting with a concrete other experiencing the same problem. Moreover, talking allows for feedback which can result in improved understanding of an other's experiences and perspectives. In praise of empathy, Barber writes: "there is perhaps no stronger bond and no more significant ally of public thinking than the one fashioned by empathy" (Barber, 1984, p. 188). In summary, the process through which an opinion is constructed or altered is of significant importance to the functioning of democracy. Huckfeldt et al refer to preferences not communicated in public as private opinions "secluded from the bright light of public scrutiny and socially communicated information" (Huckfeldt et al 1995, p. 1049). For Barber, the walls of the voting booth prevent this bright light from illuminating the private opinions that inform voting decisions. In 'thin' democracy, the citizen's most symbolic and fundamental democratic act, voting, is private: "we do not need to explain or justify it to others (or, indeed, to ourselves) in a fashion that would require us to think publicly or politically" (Barber, 1984, p. 188). By talking about politics with others holding diverse and differing initial preferences, citizens are best able to make choices in the interest of the whole. The collective decisions that result from this process more closely approximate the ideal of public opinion. Knowledge and Politics: Talking about politics with diverse others is an important element of the citizen's search for political information. Theories of both deliberative and representative democracy emphasize the need for citizens to be knowledgeable about politics. Berelson, Lazarfeld and McPhee provide the classic expression of this point: The democratic citizen is expected to be well informed about political affairs. He is supposed to know what the issues are, what their history is, what the relevant facts are, what alternatives are proposed, what the party stands for, what the likely consequences are. (Berelson, Lazarsfeld and McPhee, 1984, p. 308). In defining political sophistication, Robert Luskin identifies three dimensions of an individual's political belief system (PBS). Size refers to the number of cognitions 9 one's PBS contains. Range is the extent to which a PBS covers the various topics that politics entails. Finally, constraint, or organization, involves the "extent to which the PBS's cognitions are interconnected" (Luskin, 1987, p.859). While discussion with others should frequently enhance both the size and range of one's PBS, talking is of particular importance to the development of constraint. The process of talking to another person requires an individual to identify connections between political considerations both in order to construct an argument and to respond to assertions made by one's discussion partner. A later section on discussion and deliberative thought investigates this mechanism more closely. At this point, however, it is sufficient to note that deliberation about politics increases a person's political sophistication and thus better enables one to fulfill the criteria of the democratic citizen. Beyond providing citizens with a format conducive to political learning, formal deliberation also enhances citizens' motivation to gather information and become more politically informed. Since talk in a deliberative democracy is political participation which has a direct impact on policy decisions, the incentives to become informed are tangible. In a representative democracy, however, most citizens have very little impact on government decisions and as such becoming knowledgeable about political matters may not seem worthwhile. Political Tolerance Talking about politics with diverse others also promotes tolerance of those ideas with which one disagrees and the people holding those opinions. Even i f a citizen is not fully persuaded by an opposing position, participation in the discussion exposes the citizen to the reasons supporting such a view. Public deliberation displays conflicting positions and in doing so ensures "that no one [can] see the end result as arbitrary rather than reasonable and justifiable, even i f not what he or she [happens] to see as most justifiable" (Fearon, 1998, p. 62). Moreover, continued exposure to the reasons behind alternative positions enhances one's general feelings of tolerance which should be of benefit to future deliberations involving opposing preferences. (Mutz 2002, p. l 12). 10 Good Talk/Bad Talk The desirable democratic implications of discussion that I outlined above assume an ideal type of political conversation. While political conversations can significantly contribute to democracy, it does not necessarily follow that all talk is good. Indeed, some scholars such as Michael Schudson argue that certain kinds of political discussions may detract from democracy. The potential contribution of a political conversation to democracy depends on both who is participating and how the conversation occurs. After a summary of Schudson's arguments on these two points I offer some comments on the implications of these propositions for this study. In an article entitled 'Why Conversation is Not the Soul of Democracy", Michael Schudson offers two dichotomies in an effort to provide an improved conception of the relationship between talk and democracy. He first differentiates homogeneous conversations in which those of similar social situations and holding similar political perspectives engage in discussion from public conversations characterized by interaction between participants with conflicting views (Schudson 1997, p. 302). MacKuen makes a similar distinction arguing that heterogeneous 'public dialogues' enhance democracy while homogenous conversation, characterized by social consensus, "restricts political freedom in the sense that it smoothers the spark to political imagination with the uniformity of accepted reality" (MacKuen, 1990, p. 60). The talk of which deliberative theorists speak is immune to this concern since deliberation involves heterogeneous talk by definition. But political conversations in most democracies as they actually occur are surely susceptible to homogeneity and as such may in fact do more harm than good. Heterogeneous conversations also facilitate the diffusion of information throughout society. In an influential piece, Granovetter suggests that information is more widely communicated when conversation involves 'weak ties' since interaction with those who do not belong to one's cohesive social group is more likely to introduce new information (Granovetter, 1973). Such diffusion contributes significantly to the development of truly public opinion. Talking with those outside one's immediate social network contributes to democracy both by providing individuals with conflicting views which they must consider and by facilitating the spread of information between social groups. 11 Having reflected on the 'who' of conversation, one must also consider the 'how'. Schudson's second distinction emphasizes the norms and rules governing different sorts of conversation. For him, 'sociable conversation' is non-utilitarian; the end is simply the pleasure of interacting with others. In contrast, 'problem solving' conversation involves efforts to arrive at a common end which, in the case of politics, is good government. Schudson argues that only problem solving conversation contributes to democracy since it involves argumentation in an effort to arrive at a shared decision. Such conversation, however, must be rule governed in order to ensure participant equality. These rules are especially necessary when citizens with diverse experiences and perspectives engage in discussion and debate. These rules include: "equal access to the floor, equal participation in setting the ground rules for discussion, and a set of ground rules designed to encourage pertinent speaking, attentive listening, appropriate simplifications and widely apportioned speaking rights" (Schudson 1997, p. 307). Deliberative theory assumes that discussions will be rule governed efforts at political problem solving. Everyday informal political conversations, however, often fail to meet such requirements completely, and frequently miss the mark by a great deal. Schudson's work is valuable in noting the disconnect between the prescribed talk of deliberative theorists and the actual conversations in which citizens participate. This said, his strict differentiation is troubling i f employed to specify the sorts of conversation that are democratic and those which are not. While specification of the two extremes is useful, failure to consider conversations that fall somewhere between the two will prove problematic. As an example, some conversations between intimates surely involve a greater amount of argumentation and divergent preferences than others. Just as the social norms of some informal discussions may accomplish the same goals, to a greater or lesser extent, as those of the rules prescribed for formal deliberation. In addition, casual conversation can have important implications which Schudson overlooks. Scheufele points out that informal conversations lacking in spirited disagreement are crucial to the creation of social capital. He also notes that casual conversations facilitate the recruitment of some citizens into more formal types of political participation (Scheufele, 2000, p. 738). This is not to suggest Schudson's distinctions are not useful. Just as one cannot classify all sorts of discussions about politics as democratic or not, one cannot 12 identify all democracies as deliberative or not without ignoring important differences. Instead, one is better served to simply assess the extent to which the actual phenomenon conforms to its ideal type. In this case, Schudson's work provides ideal types of two important aspects of political discussion against which actual discussions can be compared. Talk and Cognition Deliberative Thought: The effect of discussion on an individual's cognitive processes explains a considerable amount of talk's potential democratic benefits. The act of talking with others fosters deliberative thought. By deliberative thought I am referring to the highly cognitively demanding consideration of the numerous facts, arguments, and perspectives surrounding a given political issue. Deliberative thinking is crucial to the development of constraint in and individual's political belief system. When engaged in deliberative thought an individual actively considers a piece of information and its relation to others. Individuals search their memory for relevant considerations and deliberate over the value of each and how each relates to others. This process is reminiscent of Zaller's model of survey response as both are examples of memory based models of information processing. Zaller, however, suggests people simply average across considerations that come to mind while the deliberative thinker undertakes a significantly more complex set of tasks, not the least of which would includes an effort to reconcile divergent considerations (Zaller, 1992). By thinking deliberatively and developing constraint, individuals also become better able to meaningfully incorporate new information. In many ways, deliberative thought is for the individual what deliberation is for society. The idea of deliberative thought is not a novel one; a number of authors have identified similar concepts. Manin speaks of a process involving "the formation of the [individual] will, the particular moment that precedes choice, and in which the individual ponders different solutions before settling for one of them" (Manin, 1987, p. 348). 'Reflexive integration' is a similar process specific to the acquisition of new information in which an individual reflects on incoming messages and seeks to integrate these into 13 their cognitive structure. (Sotorovic & McLeod, 2001, p. 274). Bonham suggests that individuals make sense of information through a 'dynamic process of reflection' (Bonham, 1996, p. 59). Finally, Diana Mutz asserts that individuals incorporate new information with existing knowledge and reconcile inconsistencies through an internalized conversation between different points of view (Mutz, 1998). Stimulating Deliberative Thought: Motivation is a crucial in explaining why individuals behave as cognitive misers in some situations and deliberative thinkers in others. Most people construct opinions only when required since the benefits of holding reasoned and informed opinions do not outweigh the significant cognitive costs of engaging in deliberative thought: "The most economical way to operate is not too think much about resolving complex considerations until pressed to do so by the nice lady in the living room, itself a low probability event" (Stimson, 1995, p. 183). Talking about politics is important as it stimulates and improves deliberative thinking in a number of ways. A critique of Mutz's suggestion that media use encourages this sort of thinking and an explanation of how talk triggers deliberative thought comprises the remainder of this section. In Impersonal Influence, Diana Mutz responds to critics of the mass media by arguing that media is an important source of political information necessary in a large and diverse society. Mutz shares the deliberative theorist's concern for the degree to which the lives of others inform an individual's political decisions. She argues the mass media is a necessity since the size of our political communities make direct knowledge of others impossible. The mass media provides information about the 'impersonal other' which she defines as "the anonymous 'others' that exist outside an individual's realm of personal contacts (Mutz, 1998, p. 4). By definition, only through mediated communication can these others influence an individual's political decisions. Mutz assumes citizens actively engage new information about these others: "much of the deliberation that may once have occurred in face to face meetings of people with differing views may now occur in an individual's internalized conversation with generalized others" (Mutz, 1998, p.295). This assertion either assumes that individuals are not lazy organisms or that significant incentives exist to warrant the cognitive effort 14 such an internal discussion requires. The first approach is seriously undermined by the vast bulk of psychological research which suggests individuals do indeed hesitate to engage in difficult thought. Mutz also fails to offer evidence of incentives that would justify the cognitive activity required in internalized discussions. In essence, Mutz is suggesting that deliberative thought is reflexive upon exposure to new information. In an earlier article, she identifies 'cognitive response theory' as the mechanism motivating consideration of new information. This theory states that upon learning of divergent views, individuals generate reasons in support of this opposing view, and then reconcile these reasons with their existing opinion. She cites one study in which results confirmed the theory noting: "attitude change was induced by exposure to information about the positions only when respondents both were told and reflected upon others' positions." (Mutz, 1997,p. 105) This statement appears somewhat at odds with her underlying thesis. The italicized only indicates that exposure alone is not sufficient to cause attitude change; reflection must also occur. If simply hearing alternative views triggers an internalized conversation then she need not emphasize the fact that reflection must also occur. That the two events are treated as separate suggests exposure to new information alone does not motivate deliberative thought. While Mutz appears to offer additional extensive support for this theory in her book, the conclusions she reaches are flawed due to the challenges of measuring cognitive processes. Much of the evidence Mutz employs to support her notion that deliberation occurs internally with the impersonal others is of limited utility since the act of measuring her key concept affects the results observed. In one experiment, Mutz finds that those subjects who received cues as to levels of support for a political candidate were more likely to respond to a thought listing question with candidate-relevant thoughts (Mutz, 1998, p. 222). Specifically, the question asks "As you were thinking about your choice of candidate, what kinds of thoughts occurred to you?" (Mutz ,1998, p. 221). While Mutz admits that we cannot be certain that the thoughts reported actually informed the original decision, she fails to recognize that the question itself effectively asks subjects to engage in deliberative thought. We cannot be sure whether the initial information or the investigator's questions prompted participants to engage in thought about the subject more broadly. That those subjects who received cues reported a greater number of 15 thoughts is far from surprising since the cue serves as an additional starting point for the cognitive exploration reported regardless of which aspect of the experiment incited this exploration. The. problem of measuring how people think is by no means unique to Mutz's study and certainly impinges on my ability to prove that discussion motivates deliberative thought. This said, by wrongly attributing reflection inspired by the investigator's questions as thought motivated by simply being exposed to the views of others, Mutz reaches the flawed conclusion that intrapersonal discussion informed by the media effectively replaces interpersonal discussion. Since the preponderance of evidence supports the image of cognitive miserliness, Mutz must do significantly more to prove that simply being exposed to new information generates deliberative thought. Rather than assuming that deliberative thought is reflexive, this study assumes that such thought requires significant motivation and that talking about politics can provide such motivation. Indeed, one of the central values of talk is that deliberative thought occurs as a by-product of engagement in a conversation. One might compare the thought occurring during conversation with the aerobic exercise undertaken during a team sport such as basketball or soccer. Just as there are some content to jog alone there are surely some people who enjoy contemplation of political matters without the presence of others. Many, however, are more likely to enjoy exercise when it is contained within team sports since the social nature of the sport distracts one's attention from the demanding work of running. In short, one's team mates, competitors, and discussion partners provide incentives to engage in the difficult work of running and thinking deliberatively. Unlike conversation, the conditions of most media reception do not encourage deliberative thought for one can quite easily and uncritically accept or reject a message. Discussion is two-way and thus far more complex. Sniderman notes that we do not simply record the initial position of our discussion partner but "we respond to what he or she says in a whole variety of ways, mentioning points of information at odds with it, calling attention to shared values it contradicts, pointing out likely consequences it overlooks" (Sniderman, Jackman, & Tiberj, 2001, p.l). This is the very stuff of deliberative thought. In this case, however, the presence of the other person makes such 16 considerations explicit and forces a one to more fully contemplate the topic at hand. There are at least five unique aspects of interpersonal communication that facilitate deliberative thought. First, discussion acts as a source of new information which serves as grist for the deliberative mill. Discussion is certainly not the sole source of novel information. It is the manner in which this new information is considered, however, that is unique to discussion. Psychologists refer to the thought that occurs during interpersonal communication as 'distributed cognition': "Different heads have different prior knowledge. Moreover the different heads attend to different aspects of the information being considered. As the talk in the group proceeds, connections are made." (Pressley, 254). Not only is information shared within a discussion, but so too is the task of thinking about this information in a deliberative way. An individual's deliberative thinking is enhanced by the ability of others to connect considerations in way that the individual thinking in isolation would not have identified. Simply put, when it comes to the quality of deliberative thought, two heads are better than one, four better than two, and so on. Second, the individual's desire to manage the impression others have of her provides motivation for the individual to engage in deliberative thought. The presence of others encourages purposeful thought since individuals seek to impress others by offering reasoned arguments and statements. Similarly, discussion partners are better able to critique the validity of an individual's assertion than is the individual himself. When deliberating internally one might ignore or discount a conflicting piece of information with relative ease. In social deliberation, however, the fact that others will surely raise the this piece of information forces the individual to generate legitimate opposing reasons. Finally, as anyone who has attempted to solve a difficult problem knows, the mind tends to wander when faced with a difficult task. Thus the thoughts of an isolated individual engaged in an intrapersonal discussion might easily stray to more pleasant matters while the presence of others in interpersonal discussion encourages the individual to return to the topic at hand. Third, the two-way nature of discussion increases the potential that a given individual will understand a particular political consideration. Since "new information is most likely to be learned when it is meaningful; when it can be related in a substantive 17 way to previously known information", it becomes clear that arriving at similar levels of understanding of a given issue will require a different strategy for different individuals (Glover and Corkill, 1990, p. 109). Thus a discussion partner can tailor communication of a consideration to meet the specific cognitive structure of the person with whom he is talking. Difference theorists would likely emphasize this point since individuals with differing social experiences and perspectives will be better able to reach a mutual understanding though trial and error attempts to explain their position. Fourth, when talking about politics, participants are actively engaged with material and not merely passive recipients of incoming information. Indeed, the act of talking about a consideration forces deliberative thought. Much of the utility of acts such as journal writing, counselling, and tutorials stems from the manner in which communicating an idea tends to encourage purposeful thought about the idea. One study of group work in schools finds that when students are involved in group work, those students who do the explaining learn significantly more than those who simply listen (Pressley, 1995, p. 256). Here we have an apparent parallel between media message reception and participation in discussions for those who receive messages passively tend to learn less. Despite the apparent benefit of discussion, however, this study also suggests we must attend to the specific dynamics of a conversation and can probably safely assume that each discussion has a different impact on the different participants. Fifth, and finally, the social nature of discussion can enhance each participant's ability to engage in deliberative thought in the future. In the course of a political conversation, participants do not merely expand their knowledge of the topic of discussion but can also expand their cognitive strategies: "opportunities to engage in interpersonal problem-solving and thinking stimulate the development of personal problem-solving and thinking skills, in that many of the processes that the group carries out together eventually are internalized by individual members of the group" (Pressley, 1995, p. 259). Individuals can acquire both the strategies employed by other participants and the approaches generated and applied by the group as a whole for use in future moments of deliberative thought. Thus of the numerous ways in which talk contributes to democracy, one of the most important is the influence discussion can have on the cognitive effort, 18 content, and ability of citizens. It must be noted, however, that the processes outlined above are grounded in the type of talk expected of formal deliberation. Informal political discussion of the sort studied here may not produce the same benefits to the same extent. Deliberation, Persuasion, and Political Sophistication: Before shifting to a discussion of the hypotheses to be tested in this paper, the relationship between deliberation and political sophistication deserves attention. It seems possible that persuading a citizen who has become quite sophisticated through extensive participation in deliberations could prove rather difficult. For Luskin, an ideology "is a particularly sophisticated PBS- large, wide ranging, and highly organized" (Luskin, 1987, p. 862). Most who have tried have likely found that attempts to persuade a highly ideological person are often pointless given the rigidity of their thinking. The idea of a deliberative society may therefore prove problematic if the cognitive development motivated by deliberation results in more fixed and permanent attitudes. I can offer two points in response to this question. First, the near infinite range of political considerations suggests that no one can truly consider, in an in-depth way, everything that relates to a given opinion. Thus presentation of a new consideration, or even a previously considered idea presented from a new perspective, necessitates continued deliberative thought which can lead to opinion change. Moreover, it seems clear that most people hold a number of deeply held beliefs and values which may coincide with respect to one issue but which will certainly conflict when considering a different issue. Despite the collective efforts of a vast amount of intelligent people, there still remains no widely accepted definition of concepts such as democracy and justice. If over a century of political science has not resulted in a permanent position on such matters, it seems safe to suggest that the political belief systems of individuals devoting dramatically less time to deliberative thought about political matters will remain open to change. The vast number of essentially contested concepts involved in politics ensures that people's minds will always be open to change. Second, despite this first point, i f social norms dictate that sticking to one's opinion is more desirable than being persuaded to adopt a new position, then 19 deliberation may still prove difficult. Rather, participants in a deliberation must understand their task as finding the position upon which all agree and cannot, therefore, conceive of their primary purpose as the defence of their initial preference. Thus while citizens of a deliberative democracy would likely be quite politically sophisticated, they must understand their political belief systems as impermanent and constantly subject to change. This point highlights another important difference between formal deliberation and informal political conversations. Summary The goal of this chapter is to provide the reader with an enhanced understanding of the relationship between talk and democracy. This understanding should provide the context required to interpret the empirical research explored in the next three chapters. Rather than reaching decisions through a process of either voting or bargaining, citizens of deliberative theory do so through shared contemplation and debate of the issue. Deliberation results not only in decisions at the society level, but also in individual level decisions. Thus while assessing the extent to which Canada as a society reached decisions through deliberation is a non-starter, I do intend to investigate the extent to which individuals' political opinions occur as the result of participation in discussion. Do informal discussion result in preference transformation? Do casual conversations act as sites of political learning? Some of the ideas raised in this chapter also indicate the importance of a conversation's participants and content. Formal deliberation requires the presence of dissimilar views and equal participation by all regardless of initial preferences. Do informal conversations meet these criteria? Finally, while formal deliberation is to result in agreement, one by-product of the informed consideration of alternative perspectives is greater tolerance of those holding different opinions. Do people who participate in informal discussions exhibit higher levels of tolerance? The next three chapters seek to supply answers to these questions. 20 CHAPTER 3: Antecedents of Discussion The conditions that give rise to political conversation are the focus of this chapter. In order to gain an understanding of the role of political conversations I first assess the characteristics of those engaging in differing frequencies of political discussions. The first section of this chapter highlights the sort of results expected if discussions closely approximate the deliberative ideal and provides context by summarizing the results of previous research. The second section outlines the measures and methods employed to test the hypotheses outlined in the first section. Presentation and interpretation of the results comprise the final segment. Hypotheses: Since the ideal deliberative setting entails full and equal participation by all citizens, the specific characteristics of an individual are unrelated to the frequency of his participation in deliberation. Informal discussions are not, however, governed by such rules and thus we should be concerned with the extent to which one's demographic, motivational, or behavioural characteristics affect the likelihood that one will participate in discussion. Since equal participation lies at the heart of formal deliberation, i f informal discussion in a society serves as an adequate substitute for formal deliberation then there should be essentially no difference in discussion frequency amongst different sub-groups in the population. When the differences in discussion behaviour between sub-groups in a population are great, informal discussions fails to meet the first criteria of formal deliberation: participant equality. Identifying the first cluster of characteristics to include in a model predicting discussion necessitates rounding up 'the usual suspects' including: gender, age, education, income, language spoken at home, and the presence of a spouse or live in partner. A second set of characteristics includes motivational factors such as general interest in politics, internal and external efficacy, a measure of general political knowledge. The third, and final, component involves media use variables. Below I report some previous findings on the antecedents of political conversation in order to provide some context within which the results reported later can be interpreted. 21 In a study using data from 1996 nationwide survey of the US, Kim et al find that of the demographic variables considered, only education was a significant predictor of political talk (Kim et al, 1999, p. 12). Dietram Scheufele reports a similar finding as both education and income explain participation in political conversations. (Scheufele, 2000, p. 733). Bennett, Flickinger, and Rhine compare the results of OLS regressions of political conversation in the 1990's in Britain and America. They find significant coefficients for race, age, education and income in Britain while gender, age and income achieve significance in the U.S. (Bennett et al, 2000, p. 109). The same authors also include marital status in their model but find it unrelated to discussion frequency. In light of these findings, each of the variables used in these other studies appear in my model along with the respondent's home language included in an effort to identify newer immigrants and take into account the differences between French and English speaking Canadians. A much larger amount of research focuses on the relationship between media use and participation in conversations about politics specifically and the news more generally. Perhaps the most intuitive link between the two is the media's provision of content for interpersonal discussions. Even those who do not consult media sources likely discuss topics raised by the media. In Tarde's words, "one pen suffices to set off a million tongues" (Tarde, 1969, p. 304). Scholars seem to disagree over the normative value of the media's role in setting the topics, boundaries, and language of debate. Noelle-Neuman suggests that by providing "people with words and phrases they can use to defend a point of view" the media enhance democracy (Noelle-Neumann, 1993, p. 178). Barber, on the other hand, argues this media effect hinders democracy since deliberation is not truly democratic when the terms of debate are set by an elite few: "those who control language thus control the communal we.. .give each citizen some control over what the community will mean by the crucial terms it uses to define all the citizens' selves and lives in public and private, and other forms of equality will follow" (Barber, 1984, p. 193). As this quotation points out, the stumbling point between the two sides stems from the question of how democratic media sources are. Mutz and Martin applaud existing media outlets as they compare the current media processes and content with more elitist media of the not so distant past (Mutz & Martin, 2001). Barber finds the 22 same media lacking as he contrasts the current media with the ideal media of a 'strong democracy'. This more critical approach to the democratic nature of the mass media informs this paper since the aim here is to contrast actual phenomena with the theoretical sort expected of a radically participatory democracy. The empirical evidence accumulated thus far suggests a rather complex relationship between these two sources of political information. Scholars appear to have shifted from a comparison of the relative benefits of each source to consideration of the relationship between media and conversation. In a unique investigation of the effect of an eight month newspaper strike on interpersonal discussion, Jeffery Mondak finds that access to news media is positively related to discussion frequency while at the same time the two sources of information compete for influence on voting decisions (Mondak, 1995). In their study of news comprehension, however, Robinson & Levy conclude that discussion of news stories is just as important in explaining comprehension as exposure to news media alone (Robinson & Levy, 1986). This finding presents a considerable challenge to researchers such as Mutz and Martin who argue that media provide significantly greater exposure to conflicting opinions since Robinson and Levy conclude that exposure alone does not explain comprehension. Sotirovic and McLeod similarly suggest that talking about the news results in more sophisticated comprehension of the issues at hand (Sotirovic and McLeod, 2001). Researchers have also sought to disaggregate the concept of mass media into the different communication mediums. When compared to television news, previous work has found a more significant relationship between newspaper reading and political information (Robinson & Davis, 1990), conversations about the news (deBoer, 2001, p. 152) and political conversation (Kim and Wyatt 1999, Scheufele 2000). Scheufele writes: "whereas newspapers mostly used with explicit informational intent, TV is often considered a more passive medium that fosters what has been labelled 'learning without involvement' (Scheufele, 2002, p. 52). In addition to describing the characteristics of people who discuss politics, I will therefore also investigate these two hypotheses in the first set of empirical tests: 23 H3.1: Media use is positively related to participation in political conversations. H3.2: Newspaper use is more strongly related to political conversations than use of television. Measures Political Conversation How frequently people engage in political conversations is the key independent variable of this study. CES investigators asked respondents how frequently they discussed the campaign with two separate groups: 'friends and relatives' and 'others'. These questions were asked in both the campaign and post-election waves and presented respondents with 'often', 'occasionally', and 'not at all ' as possible responses. Since my focus is not specifically on the dynamics of the campaign, I use the post-election question as a measure of overall discussion frequency throughout the campaign. Campaign wave responses to this question were surely affected by date of interview since discussion frequency increases as the campaign matures. While the post-election question treats everyone equally in this regard, it seems likely that responses to this question are subject to biased recall. Not only might people overestimate their actual behaviour in an effort to appear more desirable, they may also recall only the last few days of the campaign when assessing their own discussion frequency which would result in an estimate unrepresentative of their discussion frequency over the entire campaign. The findings reported later, therefore, are clearly not representative of general political conversation. Campaigns generate heightened political moments which results in more frequent discussion of politics than the amount occurring between elections. Accordingly, my conclusions can only be applied specifically to campaign conversations. I treat the two discussion variables as distinct, and therefore use them separately in regressions for two reasons. First, from a theoretical standpoint, conversations with those other than one's friends and relatives should tend to involve greater consideration of dissimilar views. Talk is important to democracy in so far as it encourages an individual to contemplate life experiences and political preferences of which they are unfamiliar. Second, the variables are not so highly correlated that one can suggest they measure the same thing, (see Table 3.1) 24 Tab le 3.1 D i scuss i on F requency Fr iends and Fami ly Not at all O ther P e o p l e Occas iona l l y Often Total Not at all 77 .15 21.91 0.94 100 Occas iona l l y 21 .75 75.04 3.21 100 Often 6.87 39.89 53 .24 100 Total 27 .55 57 .63 14.28 100 P e a r s o n Ch i2= 1868 .0564 pr=0.000 Kenda l l ' s tau-b= 0 .5795 News Media Use: I use two different measures of media use in this study. The CES asked respondents, on a scale of 0-10, how much attention they paid to news about the election in the newspaper, on television and on radio. Since I expect the relationship between discussion and media use to differ by medium, some analyses employ all three of these measures as independent variables. In most of my analysis, however, I use a general media use variable which represents a respondent's highest response of the three media use questions. This measure should prove more useful than a composite measure comprised of all three types of media since most citizens receive the bulk of their media information from a single medium (Zaller, 2002). Demographics Throughout the study, I use the same set of demographic variables in each regression regardless of whether discussion is the dependent or independent variable. Gender and age are self explanatory, the latter ranging from 18-101 and coded 0-1, as are all other variables. During the campaign wave respondents were asked to select one of eleven responses as their highest level of completed education and to report their household income1. I created dummy variables for those identifying French and a language other than English or French as their home language. Finally, I use a dummy variable to differentiate those who are married or live with a partner since these people 1 1 used an imputed income variable generated by Fred Cutler (2000). It is based on ces variables cpsml6 & 16a. It is recoded to thousands. Cutler offers the following description o f his procedure: "Miss ing data were imputed based on a regression method, with values drawn from a distribution with a mean o f the predicted vale (XP) and a variance equal to the square o f the standard error o f the regression estimate. T w o imputations were made and then one was selected randomly. Thus standard errors for income estimates in all tables are slightly too small because they do not take this imputation into account." (Cutler, 205). 25 are more likely to report higher discussion frequency given the greater opportunity couples have to engage in discussion. Attitudes The three attitudinal variables in this study appear as both dependent and independent variables given their hypothesized non-recursive relationship with discussion. Interest in politics is a powerful predictor of most politically related behaviours and measured here by respondents self identification of their level of interest 'in politics generally' on a scale of 0-10. My two measures of efficacy are the same as those used by Mendelsohn and Cutler in a study investigating the effects of referenda (Mendelsohn & Cutler, 2000). To create a measure of external efficacy, I recoded the four ordinal responses (strongly agree through strongly disagree) to the statement: 'People like me don't have any say about what the government does'. My measure of internal efficacy is the result of a similar process and based on responses to this statement: 'Sometimes politics and government seem so complicated that a person like me can't really understand what's going on?'. Finally, to assess general political knowledge, I use a composite measure based on respondents' answers to four factual information questions: ability to name the current U.S. President, the current Minister of Finance, the current premier of a respondent's province, and the first female Canadian Prime Minister2. In order to test the hypotheses contained in this section I include all the variables listed above in two multivariate OLS regression models predicting reported frequency of discussion with friends and relatives and with others. Results Figure 3.1 displays the frequency distributions of responses to the two survey questions most central to this study. The observations of both variables are distributed in a normal fashion. Responses to the friends and relatives questions are skewed slightly to 2 In Chapter 5 I increase the knowledge measure to an eight point scale. The three questions added deal with campaign promises and are excluded here since I am, at this point, interested in a respondent's general political knowledge. The campaign questions are too specific and ability to answer them speaks directly to the campaign time period. 26 the right with around 5% more respondents selecting 'often' rather than 'not at all ' . The opposite is true of the discussion with 'others' which is skewed to the left with around 13% fewer choosing often. A first point that deserves attention is the fact that the During the campaign, did you talk about the election with friends and relatives often, occasionally, or not at all During the election campaign, did you talk about the election with other people often, occasionally, or not at all? 70 60 50 40 30 20 10 0 59.1 16.9 ^1 Not at all Occasionally Often Not at all Occasionally Often Figure 3.1 wording of these questions affects the results obtained. For example, were respondents provided with the opportunity to select 'rarely' as a response, it seems likely that the number who discussed politics 'occasionally' would shrink since even those who discussed the campaign only once likely responded with occasionally. The stark wording of 'not at all ' suggests that only those extremely apathetic and uninterested individuals would select this response. That 17% of the population never discussed the campaign with those close to them offers the first clear indication that informal discussions differ from formal deliberations in important ways. Moreover, close to a third (27.6%) of respondents never talked about the campaign with an 'other'. This finding is of greater significance assuming that discussions with others are more likely to involve new information and dissimilar political views. Interestingly, however, only 13% of those surveyed responded 'not at all ' to both questions. Therefore, some citizens (around 4%) who are either unwilling or unable to discuss politics at home are able to do so with those beyond their circle of friends and relatives. 27 Tab le 3.2 An teceden ts of d i s c u s s i o n f requency ( O L S Regress ion ) . Dependen t Va r i ab le Fr iends and Re la t i ves Other P e o p l e B S tandard Errors B S tandard Errors Demographics Ma le 0.016 0.012 0.065 0.011 A g e 0.206 0.039 0.140 0.039 Educa t ion 0.262 0.032 0.197 0.032 Income 0.396 0.130 0.450 0.130 S p o u s e / P a r t n e r 0 .009 0.120 0.000 0.012 L a n g u a g e at home F rench -0.041 0.014 -0.036 0.014 Nei ther F rench nor Eng l i sh -0 .012 0.027 -0 .027 0.028 Cons tan t 0.270 0.029 0.214 0.030 N 2796 2796 Ad jus ted R 2 0.041 0.039 A l l var iab les c o d e d 0-1 , for income 1= $1 ,000 ,000 * Indicates p> 0.1. Bo ld ind icates p>0.05. Table 3.2 presents the results of a multivariate regression estimating respondents' frequency of conversations about the campaign with 'friends and relatives' and 'others' . In this model I included only socio-demographic variables in an effort to understand who engages in political conversations. If informal deliberations effectively substitute for formal deliberations then we should expect little differences in discussion frequency based on demographic characteristics. This is clearly not the case as age, education, and income are all positively related to frequency of discussion with both groups of people. The income coefficient is deceiving since the highest reported income is one million 3 In separate models not reported here I included the fol lowing variables, none of which achieved significance: a series o f dummy variables representing the different regions in country, a similar series based on province o f residence, union membership, religion, party identification, strength o f identification, and the presence of children in the respondent's home. 28 dollars. An increase of 50,000 in household income is associated with a 2% increase in likelihood of responding with 'often' when asked about discussion frequency with friends and family. Those belonging to more privileged groups of society appear to talk about politics more often. We must wait until chapter 5 to fully appreciate the importance of this gap since it is not yet clear whether these informal discussions matter in any substantive way. Tab le 3.3 An teceden ts of d i s c u s s i o n f requency ( O L S Reg ress ion ) . Dependen t Va r iab le F r iends and Re la t i ves Other P e o p l e B S tandard Errors B S tandard Errors Demographics M a l e 0.013 0.011 -0.042 0.011 A g e -0 .047 0.037 -0 .068* 0.039 Educa t ion 0.049 0.031 0.028 0 .033 Income 0.256 0.115 0.328 0.121 S p o u s e / P a r t n e r -0 .003 0.011 -0 .012 0.012 L a n g u a g e at home F rench 0.006 0 .013 0.007 0.014 Nei ther F rench nor Eng l i sh 0.004 0 .025 -0 .008 0.027 Attitudes Interest in Pol i t ics 0.280 0.023 0.251 0.024 G e n e r a l Pol i t ical K n o w l e d g e 0.059 0.021 0.024 0.022 Internal Ef f icacy -0 .022 0.017 -0 .010 0.018 Externa l Ef f icacy -0 .005 0.016 -0 .007 0.017 Political Information Sources M e d i a U s e 0.323 0.022 0.268 0.023 Cons tan t 0.135 0.037 0.210 0.039 N 2692 2690 Ad jus ted R 2 0.241 0.174 A l l var iab les c o d e d 0-1. Fo r Income, 1 =$1,000,000 * Indicates p> 0.1. Bo ld ind icates p>0.05. ' Table 3.3 reports the results of an expanded model in which I include explanations for differences between demographic groups. Consistent with previous findings, interest in politics plays an extremely important role in explaining interpersonal communication. Interest likely encourages discussion both by motivating an individual to raise the topics of politics and by motivating the individual to not inhibit conversations 29 raised by another. With a slightly larger coefficient (.32) than interest in politics, the media variable is the most powerful predictor in the equation and explains nearly one third of the possible range of the discussion variable. Since I control for interest in politics, this media effect is distinct from the impact that acquiring new political information has on interest. One possible explanation is that regardless of whether or not an individual media story affects one's general interest in politics, the individual may seek out conversation in an effort to reach a greater understanding of the media story's content. Also, regardless of interest in the subject matter, knowledge of the topic acquired through the media may make individuals more likely to enter into a discussion of the topic since awareness of the topic better qualifies one to participate. For the most part, the demographic characteristics that are significant in Table 3.2 are not in Table 3.3. Thus greater attention paid to the election through the media coupled with a higher interest in politics seem to explain most of the differences between social-demographic groups. Income remains significant but the effect is quite weak: all else equal, an individual whose family income is 80,000 will report a discussion frequency .0135 greater than one whose family income is 30,000. In a separate regression predicting discussion with other people, I included the income variable logged. The discussion frequency of those earning 100,000 is only .04 higher than those earning only $20,000 in a year. When I included a logged income variable in predicting discussion with friends and relatives, the variable does not meet significance requirements. The only other demographic variable to remain significant is gender. Men are slightly more likely to engage in conversations with other people than women with a coefficient of 0.042. By controlling for age I have ruled out the explanation that women raised before second wave feminism are responsible for this gender gap. This result therefore suggests that political socialization still occurs differently for men and women. I am not able here, however, to investigate whether this difference is explained specifically by socialization into the world of politics, or more broadly by internalized expectations of the appropriateness of offering one's opinion. This said, the effects of both income and gender are very weak once variables such as media use are controlled for. 30 Table 3.4 allows more in-depth investigation of the talk-media relationship by reporting the coefficients of a regression predicting discussion with friends and family. A l l three mediums are significantly related to discussion. It is television use, however, that stands apart and is twice as powerful a predictor as newspaper use. This finding Tab le 3.4 M e d i a use and d i scuss i on f requency ( O L S Regress ion ) . Dependen t Va r iab le F r iends and Re la t i ves Other P e o p l e S tandard S tandard B Errors B Errors Media Sources N e w s p a p e r (0-1) 0.142 0.021 0.172 0.021 R a d i o (0-1) 0.103 0.019 0.128 0.020 Te lev is ion (0-1) 0.294 0.021 0.191 0.022 Cons tan t 0.311 0.010 0.240 0.010 N 3124 3122 Ad jus ted R 2 0.174 0.135 Bo ld ind icates p>0.05. stands at odds with existing research and may be explained, in part, by the focus on friends and family. The shared viewing potential of television may encourage conversation since programs expose viewers to the same message at the same time. Asking for clarification, explanation, or offering a critique of a story seems more likely when one knows that the individuals around him have just watched the same program. In order to engage in conversation specific to a newspaper story one must first establish whether the potential discussion partner has read the article. The results of the equation predicting discussion with 'others', also in Table 3.4, offer some support for this perspective as the coefficients between the two media types are much more similar. That few people watch TV news with those beyond their circle of 'friends and family' suggests that discussion with others about information from the TV news is more similar to that of newspaper discussions in so far as the medium itself rarely prompts discussion of the topic. This said, regardless of medium type, it is troubling from a deliberative perspective that citizens engage in these important comprehension-generating discussions only with those who likely hold similar views. Indeed, in this case, discussion may well 31 disrupt the effectiveness of the media in presenting dissimilar views since the discussion about a news story amongst like-minded people can involve immediate refutation of alternative views presented in the story. Given the consistently stronger relationship between newspaper use and discussion in studies of the US, this television finding comes as a surprise. In a study involving British respondents, however, Bennett et al (2000) report a finding similar to that reported here with television exposure, and not newspaper exposure, predicting discussion. Norris and Sanders note that American papers have less explicit biases than their British counterparts. Canadian papers, to a lesser extent, seem to follow the British pattern. It may be that newspaper readers in these two countries more easily comprehend information from their newspaper because issues are conveyed through a shared ideological lens. Readers in the US may seek out conversation to a greater extent since the more even-handed approach of American papers complicates the choice of which side to take on an issue. On the other hand, the presence of a popular national broadcaster on British and Canadian airwaves may explain the differing results. These publicly funded news broadcasts may have a greater informational focus than privately-owned broadcasters in the US. Thus those American citizens who turn to the newspaper in search of more information and more in-depth coverage might well tune into The National were they to move north. In a recent paper, Norris and Sanders argue weaker support for the 'print superiority' hypothesis in Britain than in the US "can be explained by structural features of the public service broadcasting tradition in Britain compared with the commercial ethos of network television in the US (Norris & Sanders, 2002, p. 3). Certainly, this explanation could be extended to cover the Canadian findings reported here. Summary Not all Canadians spoke about the campaign with the same frequency, and a full tenth of the population did not enter into political discussions at all. Such results are not surprising and suggest that while some deliberative conversations may occur, only some members of society participate in them. Why these groups differ is of specific interest. Interest in politics and media use lie behind the greater frequency with which older, more 32 educated, male, and more wealthy respondents engage in political conversations. If informal conversations do result in deliberative outcomes, then the relative silence of less privileged social groups is a specific concern. Media use also explains a large amount of the range of discussion frequency. Media play an important role in fostering discussion by providing content and expertise which viewers can use in subsequent conversations. From a normative perspective, i f political conversation is deemed to be a good, then future research into how different content and different mediums affect the frequency and quality of discussion should prove valuable. Finally, interest in politics, not surprisingly, is an important antecedent of political conversation. Again, research providing an explanation of how people become interested in politics should offer further insights into the relationship between interest and discussion. Given the significant controls for alternatives, the most plausible explanation may lie in the minimal impact most individuals can have on political decisions. Were politics something that most perceive they can affect, interest in political issues would likely increase. Under the present democratic system in Canada, interest in political matters may be more akin to a hobby than something more central to an individual's life. One's choice of hobbies, while surely the result of a complex causal chain, is largely idiosyncratic. 33 CHAPTER 4: The Content and Process of Informal Discussion, Deliberations are formalized when a set of rules and norms govern who participates in conversations and how they do so. Without direct observation it is impossible to fully assess the degree to which informal discussions are characterized by similar processes. The 1997 CES does, however, enable an analysis of two of the most important features of deliberative discussion: exposure to dissimilar views and the presence of minority viewpoints in discussion. In order to accrue the benefits of talk outlined above such as the creation of a more truly public opinion, the need to reconsider initial preferences, and a wider dissemination of information and perspectives, conversations must involve those from different social experiences and holding, at least initially, divergent political preferences. Hypotheses Discussion and Disagreement: Critics of informal discussion point to the negative impact social norms can have on the potential for discussion to foster desirable results. Most notably, Michael MacKuen employs formal modelling to illustrate the detrimental impact of discussion when citizens, fearing the discomfort of disagreement, only engage in conversation with those holding like preferences (MacKuen, 1990, p. 59). DeBoer suggests that individuals engage in discussion of information gathered in media in order to assess whether or not this new information violates "the dominant norms and values in their social groups" (DeBoer, 2001, p. 153). As Schudson notes, the social norms surrounding disagreement make democratic discussion, by definition, uncomfortable. By establishing a set of specific rules and norms, formal deliberation avoids the negative side effects of social desirability and other such impediments to formal deliberation. In casual conversation, on the other hand, only the norms of society dictate behaviour. Thus the value of such discussions hinges on the behaviour of citizens during these discussions. It is worth pointing out, however, that while disagreement and participation by diverse others is crucial to democratic discussion, the conceptual distinction between these concepts and their opposites tend to be overdrawn. Much has been written lately on 34 the importance of citizen engagement in the sort of activities that bring them into contact with those from different social backgrounds and experiences (i.e. Putnam, 2000; Uslaner, 2002). While interaction with diverse others and the existence of heterogeneous social networks is indeed crucial to democracy, we must not overlook the significant differences between people within a certain group. The specific focus on a single identity in order to differentiate between groups tends to pre-empt consideration of the other identities that differ amongst members of the given identity group. For example, while people of a specified class may tend to interact only with members of the same class, this group of individuals also differs by gender, age, race, party identification and so forth. The point here is not to suggest that an individual who interacts with only a close set of associates lives up to the democratic ideal, but that within this set of associates some variation in political preferences is surely present. With respect to disagreement, then, the following hypotheses deserves attention: H4.1: Both discussion with 'friends and family' and 'others' are related to reported frequency of disagreement. The relationship between disagreement and the latter is stronger than it is with the former. In both cases, greater frequency of discussion results in greater frequency of encountering disagreement. Are people engaging in the sort of conflict-riddled interaction expected of formal deliberations when they participate in informal discussions? Since it is difficult to identify a specific amount or frequency of disagreement that must occur in informal discussion in order for it to mimic the effects of formal deliberation, an answer to this question is more likely to arise through an investigation of the outcomes of informal discussion. This said, the frequency with which Canadians encounter disagreement is still of interest as is an understanding of whether certain types of people are more likely to have experienced disagreement. Previous efforts to answer these questions offer some guidance in interpreting the results. In studies of both the 1984 and the 1992 presidential elections, Huckfeldt and his colleagues find that no less than one third of a respondent's political discussion network have candidate preferences different from the main respondent noting, "disagreement and heterogeneous preferences are the rule rather than the exception within microenvironments surrounding individual citizens" (Huckfeldt et al, 2002, p. 3). With 35 respect to the idea that political conversations serve as the means to enforce social conformity, Huckfeldt and his co-authors conclude that conformity is not imposed upon minority opinion holders and that disagreement is not reduced through interpersonal communication (Huckfeldt et al, 2002). These authors suggest that disagreement is able to survive due to the perceived unimportance of political opinions and general acceptance of the subjective nature of politics (Huckfeldt et al 1996; Huckfeldt et al 2002). Those with more diverse social networks, however, talk more often, with more disagreement, and are better able to accurately judge the distribution of preferences than are those with more homogenous contacts (Huckfeldt et al, 1995). Ulbig and Funk approach the issue from a different angle by investigating different levels of conflict avoidance in individuals. They find that those who seek to avoid conflict the most tend to avoid face to face forms of political participation in which disagreement is most uncomfortable. Moreover, they find that those with lower education and lower incomes tend to score higher on conflict avoidance and assert that "those least equipped to take part in democratic political system may be further deterred from participation to the extent that the avenue for participation involves social conflict" (Ulbig & Funk, 1999, p. 277). Keeping these results in mind, I shall provide results describing the characteristics of those experiencing more frequent disagreement. The Spiral of Silence: Noelle-Neumann's 'spiral of silence' theory is one of the more widely researched concepts in the study of public opinion. For effective deliberation to occur, participants must be willing to express views that conflict with those of the majority. Formal deliberation involves a set of rules established to ensure that such minority perspectives receive equal consideration by all participants. The 'spiral of silence' theory suggests that individuals employ an inferred opinion distribution when deciding whether or not to raise a particular point of view. (Noelle-Neuman, 1974). A recent meta-analysis covering 25 investigations of the 'spiral of silence' concluded that research thus far provides very limited confirmation of the theory. The authors find an overall correlation between willingness to speak and perceptions of opinion climate that, while statistically 36 significant, is only 0.054. They suggest that future efforts should measure actual willingness to speak as opposed to hypothetical readiness (Glynn et al, 1997). The current study meets this criterion as these discussion questions seek to measure actual discussion behaviour. In a study of people's feelings of 'freedom to talk', Kim et al find that neither perceptions of local nor national opinion climate explain 'freedom to talk'(Kim et al, 2000). To see whether or not the spiral of silence applies to the 1997 Canadian federal election I shall test the following hypothesis: H4.2: A respondent's willingness to engage in political conversation is, in part, the result of his perceptions of his preferred party's likelihood of electoral success at the riding level. These perceptions of opinion climate are positively related to discussion frequency. Method To explore the determinants of disagreement frequency I estimate a regression model in which disagreement serves as the dependent variable. I include both types of discussion in order to assess the whether or not talking with certain groups of people is more likely to involve disagreement. Since the disagreement question was asked only during the campaign wave, I use the discussion questions from the same wave and control for date of interview. This chapter's second hypotheses seeks to test the applicability of the spiral of silence theory. In order to test this hypotheses, I created a variable that reflects the opinion climate in which a respondent resides. During the campaign wave, respondents were asked to report their perception of the chances of each party winning their riding on a scale of 1-100. For all respondents reporting for which party they intended to vote, the measure of opinion climate is their perceptions of the likelihood of electoral success for their preferred party in their riding. I then conducted a series of bivariate regressions with opinion climate predicting discussion frequency for each of the parties separately as well as a single pooled regression involving all those providing vote intention responses. Results Disagreement: 37 Discussions that sustain a deliberative democracy depend on disagreement. Conflicting political preferences are identified, defined, and debated over in an effort to arrive at a collective decisions. If casual conversations about politics are to be understood as 'democratic' in this way, they must involve disagreement. Only 8% of respondents begin to approach the ideal deliberative citizen by participating in discussions that 'often' involve disagreement, (see Figure 4.1). While 54% reported disagreeing occasionally, over a third of those surveyed (38%) never disagreed with their discussion partners. By shielding themselves from exposure to dissimilar views, these citizens deny themselves access to new political information, lose the opportunity to confront their own opinions and must therefore rely entirely on the media and subsequent internalized conversations in order for their opinions to be considered democratic in the deliberative sense. Figure 4.1 Have you disagreed with the people you talked with, most of the time, some of the time, or not at all? 60 50 40 30 20 10 0 37.71 n 7.85 Not at all Some of the time Most of the Time This frequency of agreement, however, may be over-reported for a number of reasons. First, since disagreement produces an uncomfortable mental state, individuals may unconsciously misperceive prior conversations, remembering them as less conflictual. Second, people may forget that a conversation ending in agreement may well have begun in disagreement. Third, Huckfeldt et al note that individuals often misperceive minority preferences and wrongly assume that a minority discussion partner supports the majority position (Huckfeldt et al, 1998). A subsequent article, however, points out that individuals are better able to accurately recognize others' preferences 38 during the heightened politics of a campaign. (Huckfeldt et al, 2000). This second article speaks only to partisan identification and says nothing of individuals' ability to identify agreement and disagreement about issues and policies. People may lack the political sophistication to fully realize when and how they disagree with conversation partners. Indeed, given the electorate's generally limited understanding of politics, failure to recognize disagreement seems the most plausible explanation for inaccurate reports of disagreement frequency. This said, accurately perceiving the preferences of one's discussion partner is crucial to the value of conversation and thus regardless of whether or not disagreement occurred, failure to recognize it suggests political conversations of limited value. A lack of disagreement might also result from more one-way political conversations. Some individuals may discuss politics with a more informed associate as a simplified way of gathering political information. The less informed participant may well have no opinion or trust the source to the extent that he still accepts his partner's interpretation regardless of whether or not he initially agreed with this perspective. If this is the case, political conversation serves more as an information gathering technique rather than a forum for information processing. Questions which ask respondents if they learned about a new issue or perspective would better gauge the degree to which some political conversations more closely match the former of the two conceptions of conversation. Since the range of disagreement frequency is so great between citizens, it makes sense to examine which characteristics explain participation in conflictual conversations. As expected, the frequency of discussions with both 'others' and one's 'friends and family' are related to the incidence of disagreement. In a simple regression predicting disagreement in which the types of discussion are the only independent variables, the friends and family coefficient is 0.09 while discussion with others is more likely to result in disagreement, with a coefficient of .14 (Table 4.1, Panel A). Controlling for other variables, (Panel B), reduces the apparent effects of both types of conversation. Discussion with others still exhibits a strong and significant effect while talking with friends and family just fails to meet significance requirements (p=.07). Given that the 39 Tab le 4.1 Exp la in ing f requency of d i sag reemen t ( O L S Reg ress ion ) . A Dependen t Va r iab le F requency of D isag reemen t B Dependen t Va r iab le F requency of D i sag reemen t S tandard B Errors B S tandard Errors Political Information Sources Discuss i on F requency Fr iends and Re la t i ves 0.085 0.021 0 .042* 0 .023 Other P e o p l e 0.140 0 .019 0.116 0.021 M e d i a U s e - 0.060 0.025 Demographics -M a l e - 0.037 0.012 A g e - 0.019 0.042 Educa t ion - -0 .034 0.036 Income - 0.055 0.128 S p o u s e / P a r t n e r - -0.001 0 .013 L a n g u a g e S p o k e n at H o m e F rench - -0.037 0.015 Nei ther F rench nor Eng l i sh - 0.034 0.028 Attitudes Interest in Pol i t ics - 0.075 0.028 G e n e r a l Pol i t ical K n o w l e d g e - 0.018 0 .023 Internal Ef f icacy - 0.008 0.020 Externa l Ef f icacy - -0 .009 0.019 Date of Interview (1-36) 0 .000 0.000 0.001 0.000 Cons tan t 0 .230 0 .017 0.166 0 .043 N 2818 2401 Ad jus ted R 2 0.030 0.043 Al l va r iab les c o d e d 0-1 Bo ld ind icates p<0.05 * Indicates p<0.1 dependent variable is reported disagreement, it is no surprise that discussion with others is a much more powerful predictor than other important variables such as media use and interest in politics. The substantial difference between these two types of discussions lends support to the distinction between democratic conversations and homogenous discussions. Disagreement activates deliberative thought and in doing so contributes to the quality of individual opinions. While some conversations without disagreement may 40 involve learning and exposure to new ideas, the majority involve affirmation of one's initial preference and as such are of limited democratic value. Some of the other variables achieving significance in predicting disagreement frequency are also of interest. Gender is the only demographic variable to achieve significance in this model with men more likely to report disagreement. Two related points offer an explanation for this finding. First, women are socialized to avoid conflict to a greater extent than men. Thus upon hearing a position with which she disagrees, a woman may be more likely to either 'clam' and change topics (MacKuen 1990) or offer ambiguous statements to avoid disagreement (Huckfeldt et al, 2002)~the ultimate end of both strategies being the avoidance of disagreement. Second, some women may also have less confidence in their opinions and might therefore be more easily persuaded, thus avoiding disagreement. Table 4.2 reports the difference between responses provided by the two genders to a measure of internal efficacy for those with like scores on a eight-point political information scale. Women report less confidence in their knowledge of politics when compared to men answering the same number of questions correctly. The latter explanation suggests that women are therefore more susceptible to persuasion while the implications for attitude change of the former remain unclear. More broadly, this finding points to the existence of heterogeneous conversational norms and suggests that the norms governing informal discussions differ depending on the identity of participants. Table 4.2 Gender and perceptions of political competence. Number of Correctly Answered Political Knowledge Questions 0 1 2 3 4 5 6 7 Mean Internal Efficacy Men 0.71 0.66 0.62 0.58 0.5 0.45 0.4 0.31 Women 0.77 0.7 0.72 0.65 0.57 0.48 0.4 0.41 Difference 0.06 0.04 0.1 0.07 0.07 0.03 0 0.1 For Internal Efficacy, 1= Strongly Agree that "sometimes politics and government seem so complicated that a person like me can't really understand what's going on." The lack of an education or income effect seems at odds with Ulbig and Funk's findings on conflict avoidance. These authors, however, base their conclusions on the correlation between conflict avoidance and both income and education. Since such an approach fails to control for the effects of other variables, the conclusions drawn are 41 rather suspect. More detailed research involving both conflict avoidance as a generalized trait and avoidance of disagreement in political conversations would provide more useful conclusions. From the findings presented here, it appears that these two variables are not related to disagreement avoidance. General conflict avoidance may not inform behaviour during political conversations if such discussions are perceived to be of limited importance. Disagreeing about something that directly affects one's life in an important way is quite different than disagreeing about matters perceived to be more abstract or trivial. Table 4.1 also shows that French speaking Canadians report less disagreement. The stakes of politics in Quebec, only two years removed from the divisive 1995 referendum, may offer an explanation. Huckfeldt et al suggest that disagreement is able to survive since most citizens consider politics unimportant and subjective. The salience and emotions of Quebec's independence question, however, may present a very different political climate within which discussion takes place4. Quebeckers may only discuss politics when they know their partner's position on sovereignty in order to avoid conflict. It is also possible, however, that disagreement and debate may play a different role in Quebecois culture causing respondents to report a lower level of disagreement than English speaking Canadians after participating in a conversation involving a similar amount of conflict. The 'Spiral of Silence': While the amount of disagreement in political conversations is important, so too is the degree of disagreement. A conversation involving disagreement only at the margins of a general agreement does little to force participants to reconsider their preferences. Moreover, i f certain opinions and perspectives are silenced by the social discomfort of presenting minority opinions, then conversations are of limited democratic value. To investigate whether the 'spiral of silence' applies to interpersonal communication during the election campaign, I estimated a series of bivariate regressions with either type of discussion as the independent variable and perceived level of support 4 For a related investigation of discusion in Northern Ireland see Miller, Wilford, & Donoghue, 1999. 42 for the respondent's preferred party as dependent variables (Table 4.3). I first grouped all party supporters together to see if perceptions of electoral success explain discussion Tab le 4 .3 Par ty suppor ters and percept ion of local e lectora l s u c c e s s ( O L S Regress ion ) R e s p o n d e n t percept ions of the l ikel ihood that prefered party will w in riding All Parties Liberal P C NDP Reform B Q Dependent Variable. Frequency of Discussion with: Fr iends and Re la t i ves { Q 3 Q ) ( Q 5 5 ) ( Q 7 4 ) ( Q ? 4 ) { Q M ) { Q Q 5 ) Cons tan t 0A4 0.37 0.48 0.47 0.48 0.35 Other P e o p l e 0 0 7 3 2 1 0 0 3 2 0 6 9 0 3 0 1 1 2 P (.028) (.052) (.072) (.074) (.061) (.090) Cons tan t 0 3 3 0.21 0.34 0.37 0.41 0.24 Entries in bold have a p value of less than .05. Bracketed entries are standard errors. frequency regardless of which party one supports. Opinion climate is a significant predictor of discussion with others but does not explain frequency of discussion with friends and family. This result seems logical since one is more likely to know the political preferences of friends and relatives, and can therefore selectively expose oneself only to similar views. The political leanings of 'other people' are more often unknown and thus the decision whether or not to engage in political discussion will more likely involve an inferred opinion distribution. In a regression not reported here, I included all the demographic, attitudinal, and media variables present in Table 3.3 along with perceived opinion climate to predict discussion frequency with 'other people'. Despite these additional controls, the opinion climate coefficient is 0.06 with a p value of 0.054. Overall, then, it appears that an individual's perception of the likelihood that a potential discussion partner will share his party preference plays an important role in the decision of whether or not to engage in political discussion. In order to see i f supporters of different parties differ in their approach to conversations, I treated those intending to vote for different parties as distinct groups, 43 (see Table 4.3). Of the ten estimations, perceived opinion climate only once achieves significance. Quite surprisingly, those intending to vote for the Liberals were more likely to discuss politics with 'others' when they perceived a better chance of a liberal victory in their riding. Given that the Liberal party occupies the centre of the Canadian party system, one might expect supporters to feel free to engage in conversations with others. In the same way, supporters of more extreme parties are not only less likely to encounter others with shared preferences but are also more likely to confront substantial disagreement. This approach aligns the parties with respect to economic and social policies. It may be that Liberal voters keep to themselves in certain parts of the country in which the party is vilified for its policies from a regional perspective. Regardless, the effect is surprisingly large.5 The spiral of silence hypotheses does not appear to apply to supporters of other parties. Huckfeldt et al's assertion that citizens "see political opinions as idiosyncratic expressions of personal preference that do not require explanation, similar to the choice between a Chevrolet and a Ford" offers an explanation for the survival of minority opinions in the face of social conformity (Huckfeldt et al 2000, p. 648). In effect, the perceived unimportance of political opinions keeps the spiral of silence at bay. Were citizens to perceive that their opinions matter (as, perhaps, in Quebec), either through direct participation or an elevation in the collectively perceived importance of politics, then minority opinions might well meet with rejection thus establishing a need for the sorts of rules which govern formal deliberation. I suspect classical democratic theorists did not envision general apathy as the means through which to ensure the free exchange of diverse information and ideas. These results do not reject, however, the possibility that more extreme political views are silenced by the process described by Noelle-Neuman. Even NDP and Reform supporters may speak fairly freely with each other since the policies and ideas presented by these parties still fall within an acceptable range of political ideas. Those holding more extreme opinions such as members of white supremacy or radical environmental groups may silenced to some extent by the norms of acceptable political ideas. 5 In a separate regression I included a squared opinion climate term to test for non-linearity. Since the variable did not achieve significance I have not included the results. 44 Summary The findings presented in this chapter suggest there are some important differences between the content of informal discussions and that of formal deliberation. As expected, frequency of discussion predicts frequency of encountering disagreement. That discussions with other people are more strongly related to disagreement offers confirmation for the hypothesized importance of such conversations and also offers indirect evidence suggesting conversations with friends and relatives do tend to involve shared preferences. The weak but significant effect of perceived opinion climate on willingness to express minority views also indicates the distinct difference between casual conversations and rule-governed deliberation. Optimal social decisions are possible only when all perspectives receive full consideration. That social norms silence some views indicates that opinions formed, confirmed, or revised through informal discussions will be different than those that would result from formal deliberation. Perceived opinion climate, however, is not presently a strong determinant of discussion frequency. Apathy and belief in the insignificance of a fellow citizen's views may well stave off a more powerful 'spiral of silence' one would expect to find in societies where political views are conceived of as distinctly important. 45 CHAPTER 5: The Impact of Discussion Did informal discussions during the campaign result in the sort of outcomes predicted by deliberative theory? This question motivates the third component of this study of political conversations during the 1997 election. Information and Sophistication: At the heart of both classical and deliberative theories of democracy lies the assumption that knowledgeable citizens improve democracy. Theories of deliberative democracy, however, are expressly concerned with interpersonal communication as the means through which citizens become increasingly knowledgeable about politics. The effect of participation in political conversations on an individual's knowledge of politics is therefore of significant interest. Before specifying the sorts of information to be investigated, I first consider proposed revisions to democratic theory motivated by empirical findings of information levels in the electorate. Bartels writes: "one of the most striking contributions to political science of half a century of survey research has been to document how poorly ordinary citizens approximate a classical ideal of informed democratic citizenship" (Bartels, 1996 p. 195). 'Minimalism' is one body of research which proposes corrections to democratic theory based on these findings. Supporters of this approach argue that despite their lack of in-depth knowledge, citizens are able to make reasonable decisions by employing various heuristics (i.e. Popkin (1991); Sniderman, Brady, and Tetlock (1991); Page and Shapiro (1992)). Research so far suggests that employing such shortcuts is not equally as effective as becoming informed. Bartels finds that citizens who employ various heuristics "do significantly better than they would by chance, but significantly less well than they would with complete information, despite the availability of cues and shortcuts" (Bartels, 1996, p. 217). Johnston et al similarly report that intervener cues, polls, feelings, and aggregation "did not close the gap, in translating interests into expressed preferences, between the poorly and the well-informed" (Johnston et al, 250, original emphasis). I am not suggesting here that citizens do not employ heuristics, but that these shortcuts do not result in the same choices as those that would result from formal deliberation. 46 In their widely cited study Voting, Berelson, Lazarsfeld, and McPhee also propose amendments to democratic theory in light of their empirical findings. The authors argue that apathy on the part of a significant portion of the electorate is, valuable to the democratic system since "extreme interest goes with extreme partisanship and might culminate in rigid fanaticism that could destroy democratic processes i f generalized throughout the community"(Berelson, Lazarsfeld, & McPhee, 1954, p. 314). From a deliberative perspective, these authors may have a point. At the elite level of representative democracy, deliberation occurs frequently. Discussions play an important role in decision making in legislative debates, committees, cabinet and caucus meetings and so forth. A n informed citizenry that lacks access to such deliberative forums would surely hold preferences different from those agreed upon during elite negotiations. If interested and informed citizens do not see their preferences reflected in policy decisions, such a democratic system may be undermined. Apathy and uninformedness serve as a buffer between the privately-devised preferences voters hold and the deliberatively produced policy decisions made by the elite. Voters either do not notice or do not care when their preferences are not reflected in government decisions. This said, deliberative theorists would surely object to this "implicit division of labour within the electorate" on the grounds that informed 'public' decisions can only come about when all are able to participate in the deliberations informing such decisions. While apathy may make some forms of democracy possible, and may even be necessary for the survival of some democratic systems, it does not follow that democracy cannot function with an informed and interested citizenry. That an informed and participatory public results in more truly democratic outcomes is the normative claim which lies at the heart of deliberative theory. Since deliberative theory serves as the foundation of this investigation, understanding the relationship between political discussions and knowledge is essential. This relationship is rather complex since discussion can affect the size, range, and constraint of one's political belief system to different degrees. While separating these three concepts is useful for analytical purposes, operationalizing these distinctions is more difficult given the data available. Some notable researchers suggest ability to answer factual questions about politics provides a measure of all three aspects of one's political belief system (Luskin, 1987; Zaller, 1992). The present study adopts this 47 approach to information holding and supplements it with two additional attempts to measure constraint by looking at opinionation and opinion consistency. Knowledge of Political Information: Hypotheses Luskin identifies opportunity, ability, and motivation as the conditions which encourage a given behaviour: "To become highly sophisticated, we must encounter a certain quality of political information, be intellectually able enough to retain and organize large portions of the information we encounter, and have reason enough to make the effort." (Luskin, 1990, p. 335). Conversations about politics can be beneficial since such discussions contribute to all three of these conditions. With respect to opportunity, political conversations can obviously be an important source of both factual information, and personalized experiences, about political matters. Advance knowledge that one will, or is likely to, engage in future political conversations also generates motivation to become informed so one can fully participate in conversations. Some support for this notion can be found in an article by Levine and Russo in which they find that individuals expecting to participate in group discussion of an issue attend more closely to relevant information. They also find that the degree of effort involved in information acquisition varies with respect to the positions held by those with whom discussion shall occur. (Levine and Russo, 1995). These findings again highlight the importance of conversations with those holding divergent preferences since anticipation of conflict produces greater efforts to acquire useful information. As mentioned, conversations motivate the deliberative thought so vital to the development of constraint. While an individual might uncritically accept a specific point acquired from media sources, she may be forced to reject the same point in the face of valid counterarguments supplied by discussion partners. Thus, not only is the exposure to information that occurs in a discussion important, equally important is the way that discussion fosters more extensive consideration of this information. Finally, engaging in conversations should also enhance one's ability to deal with political information in a sophisticated manner. Luskin argues that the limited ability of many citizens denies the possibility of a significantly more politically sophisticated 48 citizenry: "Even under a more facilitative regime, the combination of limited cognitive resources and competing attentional demands may keep politics a minority pursuit" (Luskin, 1990, p. 353). Indeed, Luskin recommends that researchers be more concerned with intelligence than education, implicitly suggesting that the ability to become fairly sophisticated is innately denied to many. Luskin seems to underestimate the effect of environmental factors for surely most citizens, given the right conditions, can become fairly politically sophisticated. Luskin may have reached this troubling conclusion by interpreting limited of motivation as a lack of ability. The limited social value of political sophistication, coupled with the extremely restricted potential for the average citizen to experience a direct impact on political decisions, makes the decision to remain unsophisticated rather logical. Indeed, Luskin's finding that interest is the most powerful predictor of information suggests that one's motivation to become informed is of greater importance than one's ability. Luskin argues that adjustments in the variables of the democratic system will not result in a widely sophisticated public and that only by changing the system's 'parameters' can such an end be achieved. Thus, in the end, perhaps I agree with Luskin i f he understands the direct impact citizens in deliberative democracy have on policy outcomes, and the elevated social importance of politics, as parameter changes. Participation in political discussion can enhance one's ability to become politically sophisticated. Barber notes that when citizens affect government through talk, "politics becomes its own university, citizenship its own training ground, and participation its own tutor" (Barber, 1984, p. 152). Since discussion motivates one's own deliberative thought and provides new informational strategies through distributed cognition, those who participate in extensive and informed conversations likely enhance their cognitive ability to learn about politics. Having outlined the ways through which talk affects sophistication, I now consider some of the other variables to be included in a model predicting political knowledge. In a study of information levels in the Canadian electorate, Fournier identifies education as the most powerful predictor of information. Luskin's work on sophistication suggests that education may only appear significant due to a failure to control for other variables, of which intelligence most notable. Given the lack of 49 intelligence measure in the data set both Fournier and myself employ, any interpretation of education's effects should involve a fair dose of caution. Other demographic variables worthy of inclusion are age, gender, income, and language spoken at home each of which predicts political knowledge in Fournier's work (Fournier 2002, 101). Interest in politics is a consistently powerful predictor of information in numerous studies (Luskin, Delli-Carpini & Keeter, 1989; Bennett et al, 2000). In the Canadian case, interest is second only to education (Fournier, 2002). Simply put, we learn about things in which we are interested. At the same time, however, discussion may also lead to increased interest. The relationship between media use, discussion, and information also deserves attention. It is rather obvious that the frequency with which one attends to media reports of politics should be related to one's knowledge of the topic. Fournier finds that respondents who identify newspapers as their main source of political information tend to score higher on information scales than those identifying television or conversations with friends and family (Fournier, 2002). This finding is questionable in so far as it relies on a suspect measure of media use. Respondent's identify which source they use most frequently but do not report the relative frequnecy that they consult the source. Thus, one who reads the newspaper once a month an one who reads a paper everyday are treated as identical. In a study of both Britain and the USA, Bennett et al find exposure to newspapers and radio news both predict political information while exposure to news on the television does not. (Bennett et al, 2000). Of greater interest to the current study, however, are the number of articles concluding that conversation plays an important mediating role between media use and political knowledge. Robinson and Levy find that in a national sample, "discussion was associated with almost twice as much spread in news comprehension as was media exposure" (Robinson & Levy, 1986, p. 171). Sotorovic & McLeod (2001) report similar results. In an article published in 2002, Scheufele employs a media discussion interaction term which returns a coefficient larger than either media use or political discussion (Scheufele, 2002). This overview of the relationship between discussion and information gives rise to the following hypotheses: 50 H5.1: Both types of discussion increase a person's level of political knowledge. Conversations with others exert a stronger effect than do those with friends and family. H5.2 Much of the impact of media use on information is explained by participation in conversations about political matters. The two behaviours together have a positive interactive effect, which is to say that together the two have a greater effect on knowledge than simply adding their separate effects. Method To examine the relationship between political knowledge and discussion I constructed two information variables in addition to the general measure mentioned in Chapter 3. In order to measure overall political knowledge, I generated a eight point information scale based on respondents ability to answer seven factual political questions. I use the same questions as Fournier except I use those asked in the post-election wave when the same questions was asked in both the pre and post- election surveys. Three such questions ask respondents to identify the party making a specific campaign promise: the party against Quebec as a distinct society [Reform], the party promising to lower income taxes by 10% [Progressive Conservatives], and the party promising to cut unemployment by half by 2001 [NDP]. I use the post-election responses since a respondent's ability to answer these questions depends on the day on which one was interviewed. The other four questions involve basic political facts: name of the current U.S. President, name of the Premier of the respondent's province, name of the current Minister of Finance, and the name of the first female Canadian Prime Minister. This composite scale of political information serves as the dependent variable in an OLS regression. Results Based on the assumption that discussions about politics serve as a source of information, present opportunities to consider political matters, and can instil a desire to gather further information, I hypothesized that frequency of political discussion is a strong predictor of political knowledge. The results reported in Table 5.1, however, do not provide overwhelming evidence in support of this assertion. Contrary to expectations, discussions with others was not the more effective of the two types of discussion at predicting information and the variable does not pass the standard test of 51 s i g n i f i c a n c e . F r o m a d e l i b e r a t i v e p e r s p e c t i v e , t h e n , i n f o r m a l p o l i t i c a l d i s c u s s i o n s w i t h o the rs are o f l i m i t e d v a l u e s i n c e s u c h c o n v e r s a t i o n s h a v e n o i m p a c t o n the p o l i t i c a l Tab le 5.1 Determinants of poli t ical know ledge ( O L S Reg ress ion ) Dependen t Va r iab le Dependen t Va r iab le S c o r e on know ledge Information ga ined or m e a s u r e (0-7). lost (0-1). S tandard S tandard B Errors B Errors Political Information Sources Discuss i on F requency Fr iends and Re la t i ves 0.632 0.196 0 .027* 0.014 Other P e o p l e 0.130 0.188 -0.001 0 .013 F requency of D i sag reemen t -0 .023 0.110 -0 .005 0.011 M e d i a U s e 0.934 0.262 -0 .017 0.014 D i scuss ion F requency X M e d i a U s e -0 .525 0.434 - -Demographics A g e 1.872 0.228 0.015 0 .023 M a l e 0.499 0.067 0.005 0.007 Income 0.820 0.664 0.040 0 .065 Educa t ion 2.484 0.188 0.058 0.019 S p o u s e / P a r t n e r 0.172 0.069 0.007 0.007 L a n g u a g e S p o k e n at H o m e F rench 0.002 0.081 0.019 0.008 Nei ther F rench nor Eng l i sh - .277* 0.157 0.007 0.016 Attitudes Interest in Pol i t ics 1.061 0.152 0.020 0.015 Internal Ef f icacy -0.804 0.106 0.006 0.011 Externa l Ef f icacy 0.251 0.101 0.020 0.010 Cons tan t 1.034 0.263 0.451 0.022 N 2039 1811 Ad jus ted R 2 0.287 0.009 A l l va r iab les c o d e d 0-1 Bo ld Indicates p<0.05 * Indicates p<0.1 k n o w l e d g e o f t h o s e i n v o l v e d . C o n v e r s a t i o n s w i t h o n e ' s ' f r i e n d s a n d f a m i l y ' , h o w e v e r , d o a f f ec t i n f o r m a t i o n l e v e l s . W h i l e t he e f f ec t i s l a r g e e n o u g h to s u g g e s t that s u c h c o n v e r s a t i o n s a re r e l e v a n t , the v a r i a b l e ' s c o e f f i c i e n t o f 0.63 i s q u i t e s m a l l g i v e n tha t the d e p e n d e n t v a r i a b l e s r a n g e s 0-7. I n d e e d the e f f e c t i s s m a l l e r t h a n t h o s e o f e d u c a t i o n , a g e , in te res t i n p o l i t i c s , m e d i a u s e , i n t e r n a l e f f i c a c y , a n d g e n d e r . 52 These results are somewhat confusing in light of the theoretical expectations and some of the findings mentioned above. Despite the fact that conversations with other people tend to involve more disagreement, and that disagreement should motivate more extensive deliberative thought, participation in these conversations are not related to political knowledge. The difference between the two variables might simply be a matter of volume in so far as a greater number of political conversations occur between friends and relatives. While 19% of respondents reported more frequent discussion with friends and relatives and only 5% spoke with other people more frequently, the vast majority (76%) reported similar discussion frequency (see Figure 5.1). These results, then, appear to refute the idea that volume alone accounts for the difference between the two types of discussion. Intuitively, however, these results seem surprising since one wonders whether most people really do talk about the campaign with 'other people' as often as they do with friends and relatives. The results are surely in part due to the limited range of response options presented to respondents on the discussion questions. Figure 5.1 Comparing frequency of discussions with 'friends and relatives' and 'other people'. Friends and Friends and Other people> relatives> relatives = Friends and Other people Other people relatives Even if people do talk with others as frequently as they do with friends and family, the characteristics of these discussions surely differ significantly. Discussions with others may be much shorter, frequently involving an exchange of a few general offhand comments. Moreover, people may feel more comfortable appearing uninformed and testing ideas with those closer to them. As mentioned, friends and family are more 53 likely to be present during and after media exposure, increasing the chances that conversations about new information shall occur. In short, frequency alone does not tell the whole story. The duration and content of discussions with friends and family members likely explains the difference between the two types of conversations in predicting political knowledge. A second explanation of this difference lies with the dependent variable. The type of information measured here may not speak to the sort of things people learn from talking with others. Price, Capella, and Nir find a significant relationship between 'argument repertoire', (the number of reasons offered by a respondent supporting her opinion and reasons that someone else might hold the opposite view), and discussions involving disagreements with acquaintances (Capella, Price, & Nir, 2002). This type of knowledge measure seems a better way of getting at the kind of information of concern to deliberative theorists. This said, knowledge of basic political facts remains important and speaks to more general political knowledge that affects an individual's political behaviour (Zaller, 1992). The findings presented here suggest informal political discussions with others are unlike deliberations prescribed in theory. The results contained in Table 5.1 also fail to support the second hypothesis tested in this section. The discussion - media use interaction term is not statistically significant. While it is conceivable that cultural differences are responsible for the previously mentioned U.S. findings which suggest that discussion plays an important role in comprehension of news stories, difference in measures seems a more reasonable explanation. For example, Scheufele's use of a discussion measure with a five-point scale and more meaningful categories may explain his positive results. (Scheufele, 2002). Overall, informal discussions play a much weaker role in explaining the political knowledge than expected. Information Acquisition During the Campaign: Hypothesis The panel aspect of the Canadian Election Study presents an opportunity to measure gains (and losses) of information over the course of the campaign. For the same reasons offered above with respect to overall political knowledge: 54 H5.3: Those discussing politics more frequently are more likely to have gained information over the course of the campaign. Method In order to assess the role of discussion in information acquisition during the campaign, I created a variable which measures the difference in number of correct responses to the campaign platform questions between the campaign and post-election waves for each respondent. I then used this change in information as the dependent variable in a regression with discussion and the aforementioned control variables entered as independent. Results While 30% of respondents correctly answered more of these platform questions after the election and 56% scored the same in both waves, 14% of respondents actually supplied fewer correct answers in the post-election wave. The results of a regression predicting information gains offer very weak confirmation that discussions with friends and relatives explain acquisition of information over the course of the campaign (Table 5.1). The coefficient is quite small (0.027) given that the dependent variable ranges 0 to 1. The overall hypothesis, however, must be rejected since conversations with other people do not predict gains in information more powerfully and the coefficient is not statistically significant. Only education, speaking French at home, and external efficacy achieved significance at the .05 level. Interest in politics, media use, and age fail to exert a significant effect, an interesting finding given their important role in predicting overall knowledge. The fact that conversations with friends and family are significant while these other variables are not suggests this finding is important. There is insufficient information here, however, to assess whether these effects occur through a process similar to deliberation or one more analogous to media consumption. That conversations matter with respect to people's knowledge of politics is an important finding. From the deliberative perspective, however, unless such learning occurs through consideration of, and deliberation over, a number of differing perspectives, these conversations may not result in the sort of political knowledge deliberative theorists desire. 55 Opinionation: Hypotheses While information scores may reflect constraint to some degree, such an approach likely speaks more to the range and size of one's political belief system. With this in mind, I offer two alternative approaches to measuring sophistication in the hopes of achieving a firmer grasp on constraint. Kim and Wyatt adopt the term 'opinionation' to refer to whether or not respondents provide answers to questions asked of them by survey researchers. The logic behind this approach fits well with much of the discussion contained earlier in this paper. Briefly, the authors argue that only through stimulus such as political conversations do individuals realize they hold a number of conflicting considerations relevant to a specific topic and that this realization, and the resulting reconciliation and/or elimination of considerations, results in more coherent and consistent opinions (Kim and Wyatt, 1999). Those with limited constraint lack either the possession of a previously constructed opinion or the capacity to construct an opinion on the spot. A maximally constrained person should have a readily available opinion for a wide range of political topics and thus would not likely provide a 'don't know' response. There are, however, competing explanations for 'don't know' responses that deserve attention. For those lacking existing opinions, the capacity to produce one is a function of their cognitive ability and the motivation they have to undertake the task. Certain aspects of the interview or of the interviewee will impact this capacity. For example, face to face interviewing may increase the respondent's motivation thus making some more likely to engage in the task of constructing an opinion (Green, Holbrook, & Krosnick, 2002). A respondent's lack of interest in the survey may also make her more likely to report 'don't know' in an effort to finish the survey faster. While there may be alternative explanations for 'don't know' responses, these explanations are related to the respondent's political sophistication since the condition of one's political belief system is a partial cause of these behaviours. Not all opinion questions are the same, however, and as a result I investigate two bundles of questions. Value-related questions require less cognitive effort since they tap easily accessible and important value schemas. The second type of question, which I have 56 labelled policy related questions, require respondents to offer an opinion on an issue less easily related to dominant cognitive structures like values. To answer these questions, individuals must rely on issue specific information upon which the majority of people have not likely reflected. Thus those who provide an answer either recall an opinion constructed previously or have the cognitive ability to generate one upon hearing the question. People able to answer this latter type of question therefore are likely more politically sophisticated. H5.4: Both types of discussion are negatively related to frequency of 'don't know' responses. Conversations with other people exert a stronger effect. H5.5: The discussion variables more powerfully predict opinionation on policy questions than value questions. Method I selected eight questions asking respondents to provide an opinion and created a composite measure of opinionation by summing the number of times a respondent selected 'don't know' to each of the questions. I also created two similar measures based on the two types (values and policy) of questions included in these eight. I then ran two separate regressions predicting the frequency of don't know responses for each of the two bundles of questions. The questions used are: 'Value questions': i)'How much more do you think should be done for racial minorities: more, less, about the same as now.', Do you strongly agree, somewhat agree, somewhat disagree, or strongly disagree with the following statements: ii)'Only people who are married should be having children, iii) Society would be better off if more women stayed home with their children, iv) Which is the best way to deal with young offenders who commit violent crime: give them tougher sentences or spend more time rehabilitating them. 'Policy questions': v) There's not much any government can do theses days to solve the unemployment problem, vi) To maintain our social programmes we must eliminate the deficit, vii) The government should leave it entirely to the private sector to create jobs, viii) Which is the best way to fight unemployment: eliminate the deficit, or reduce taxes? 57 Results The shared and in-depth deliberative thought that occurs in formal deliberation results in citizens with more constrained political belief systems. Informal discussion, however, does not appear to have the same effect when constraint is operationalized as opinionation. The results contained in Table 5.2 illustrate that both sorts of discussion are not related to respondent's willingness to provide answers to two sets of opinion questions. Tab le 5.2: Es t imates ef fects of d i scuss ion on opin ionat ion ( O L S Reg ress i on ) Dependen t Va r iab le Dependen t Va r iab le Don't know' answe rs Don't know' a n s w e r s to 'value' ques t ions to 'pol icy' ques t ions M (o-D Standard S tandard B Errors B Errors Political Information Sources Discuss ion F requency Fr iends and Re la t i ves 0 .015* 0.008 -0.0070.01 0 Other P e o p l e -0 .0050.00 7 -0 .0030.00 9 F requency of D isag reemen t -0 .0070.00 6 -0.0230.00 8 M e d i a U s e -0 .0020.00 8 -0.0160.01 0 Demographics A g e 0.058 0.013 0.049 0.017 M a l e ' -0 .0020.00 4 -0.0260.00 5 Income -0 .0490 .03 7 -0 .0530.04 8 Educa t ion -0.0050.01 1 -0.0500.01 4 S p o u s e / P a r t n e r -0 .0030.00 4 -0 .0050.00 5 L a n g u a g e S p o k e n at H o m e F rench 0.003 0.005 -0.0140.00 6 Nei ther F rench nor Eng l i sh -0 .0070.00 9 0 .005 0.011 Attitudes Interest in Pol i t ics -0 .0120.00 9 -0.0250.01 1 Internal Ef f icacy -0 .0020.00 6 0 .013* 0.008 Externa l Ef f icacy -0 .0070.00 6 -0 .0090.00 7 Cons tan t 0.017 0.012 0.108 0.016 N 2001 2021 Ad jus ted R 2 0.009 0.050 A l l var iab les c o d e d 0-1 Bo ld Indicates p<0.05 * Indicates p<0.1 58 These findings support the distinction between the two sorts of questions as the coefficients for most variables differ substantially in predicting 'don't know' responses to the two sorts of questions. While 88% of respondents offered an opinion on all the 'value' related questions, only 79% did so when asked for their opinion on the 'policy' questions. That only one variable (age) significantly, albeit rather weakly, predicts opinionation on value questions suggests that most people are able to recall or quickly construct opinions on these questions, likely because of the ease with which one can access the values that inform responses to these questions. With a p value of 0.066, conversations with friends and family appear to exert a weak effect on 'value' opinionation. The coefficient, however, is extremely small at 0.015. Both being male and discussion with friends and relatives are related to an increased likelihood answering 'don't know'. For policy related questions, on the other hand, six of the fourteen variables included in the model achieve significance. Disagreement frequency is positively related to 'policy' opinionation with those disagreeing often with their discussion partners were 2.3% less likely to select 'don't know'. When a person encounters disagreement, the thought required to respond should result in greater constraint. Those with more constrained political belief systems are less likely to report no opinion. There is, however, an alternative explanation. This finding may result from the fact that in order to disagree with another, one must first have an opinion. Thus it may not be the disagreement, per say, that explains the impact on opinionation. It is also possible that factors unrelated to discussion, such as interest in the survey or concerns about privacy may explain why some are more likely to select 'don't know'. Overall, however, these findings offer limited support for the idea that informal discussions foster development of constraint. The aspects of formal deliberation that result in greater political sophistication appear to be absent from most informal discussions. Opinion Consistency: Hypothesis 59 A second approach to assessing constraint involves measuring the consistency of respondents' opinions. To deliberate with others is to engage in the sort of thinking that brings organization to the often isolated pieces of information that comprise many individual's knowledge of politics. Those with more constrained political belief systems should be more likely to offer a consistent opinion on a topic regardless of question wording, question order, and other variables which effect the responses of less constrained individuals. In an important piece on measuring sophistication, Luskin is expressly critical of measures of opinion consistency. His main objection lies with the use of aggregate level measures of consistency. He does, however, criticize individual level measures of consistency on the grounds that efforts to suggest that two opinions are consistent is an inherently subjective process (Luskin, 1987). With regard to this concern, the tests employed here use such basic political concepts that the assumptions that underlie their connections should be rather widely accepted. H5.6: Those who participate in more conversations about politics report more consistent attitudes. This discussion-consistency relationship is stronger for discussions with other people than it is for conversations with friends and relatives. Method I measure consistency by comparing an individual's responses to two related opinion questions. The first pair deal with job mobility. Respondents were asked the extent to which they agree with these two questions: 'People have a right to work in the region where they are born' and 'Unemployed people should move to regions where there are jobs'. Those responding in a consistent manner are coded as 1 while those answering inconsistently are coded as zero. Those selecting 'don't know' or 'refused' are treated as missing. The second pair of questions concerns immigration policy. The survey asked respondents how strongly they agree or disagree that 'Immigrants make an important contribution to this country' and 'do you think Canada should admit more immigrants, fewer immigrants or about the same as now?'. While the link between these two questions is somewhat less direct than for the previous pair, it still seems fair to suggest 60 that those who feel immigrants make a valuable contribution and that the country needs fewer immigrants do not have a consistent attitude about immigration. The final pair is somewhat different than the previous two since one question deals with an abstract principle while the other inquires about a policy based on the principle. Politically sophisticated individuals should display 'principle-policy' consistency (Kim & Wyatt, 1999). During the campaign wave, respondents were asked 'should Quebec be recognized as a distinct society'[cpsj 18] and were subsequently asked in the mail-back survey if they agree that 'Minority groups need special rights' [mbsal4]. Tab le 5.3 Est imat ing effects of d i scuss ion on opin ion cons i s tency ( O L S Reg ress ion ) . Dependen t Va r iab le Dependen t Va r iab le Dependen t Va r i ab le C o n s i s t e n c y on job mobility (0,1). C o n s i s t e n c y on immigrat ion (0,1). Dist inct Soc ie ty Pr inc ip le -Po l i cy C o n s i s t e n c y (0,1). B S tandard Errors B S tandard Errors B S tandard Errors Political Information Sources Discuss ion F requency Fr iends and Re la t i ves 0 .109* 0 .063 -0 .002 0.058 -0.071 0.070 Other P e o p l e -0 .020 0.057 -0 .091* 0 .053 0.038 0.066 F requency of D i sag reemen t 0 .006 0.049 -0 .008 0.045 0.006 0.056 M e d i a U s e -0 .112* 0.064 -0 .047 0.060 0 .15* 0 .078 Demographics A g e -0 .169 0 .103 -0 .127 0.095 0.070 0.111 M a l e -0 .057* 0.030 0.019 0.027 0.077 0.032 Income -0.431 0.327 -0 .346 0.304 0.149 0.338 Educa t ion 0.113 0.086 0.340 0.079 0.217 0.095 S p o u s e / P a r t n e r 0 .046 0.031 -0 .035 0.029 -0 .016 0.034 L a n g u a g e S p o k e n at H o m e F rench -0 .012 0.037 0.006 0.034 0.092 0 .083 Nei ther F rench nor Eng l i sh -0 .100 0.074 0.041 0.070 0.195 0.090 Attitudes Interest in Pol i t ics -0 .095 0.068 0.068 0 .065 0.121 0.078 Internal Ef f icacy -0 .043 0.048 -0.061 0.044 0.019 0.051 External Ef f icacy -0 .040 0.046 -0 .005 0.042 -0.146 0.051 Cons tan t 0.778 0.100 0.647 0 .093 0.060 0.109 N 1115 1150 864 Ad jus ted R 2 0.010 0.036 0.040 A l l var iab les c o d e d 0-1 Bo ld Indicates p<0.05 # Indicates p<0.1 61 Results Only 17% of respondents offered consistent responses to the three sets of related questions. The reader should keep in mind, however, that this figure excludes residents of Quebec since the distinct society question was not asked in that province. With respect to consistency on job location questions, no variables return a p-value of less than 0.05. Being male, media use, and discussion with friends and relatives are all significant at the p<0.1 level (see Table 5.3) Conversations with other people display a similar strength of effect on immigration attitude consistency but are negatively related to consistency. Only education achieves significance and exerts a quite strong effect. Finally, for consistency on the minority rights-distinct society measure, both types of discussion and frequency of disagreement are not significant. The varied and weak effect of discussion variables in these models presents equivocal evidence. Moreover, the inconsistent role of most variables coupled with the rather small adjusted r-squared values suggests that the independent variables are of limited value. If anything, these findings offer support for Luskin's warnings about consistency measures (Luskin, 1987). Overall, there is a little support here for the idea that participation in conversations result in more consistent responses. It seems quite possible that conversations might enhance ambivalence and inconsistency because of exposure to new perspectives. Thus, until people perceive adequate incentives to hold consistent and well developed opinions, such as different conversational norms or significant and meaningful participation in decision making, conversations involving new perspectives may do little to enhance constraint. Efficacy: Hypothesis An individual's sense of his effect on the political system plays an important role in the outcomes of interpersonal discussion. External efficacy refers to a person's belief in the political system's sensitivity to her preferences while internal efficacy describes one's perception of her own capacity to affect the political system. With respect to external efficacy, a significant difference between formal deliberations and informal political conversations is immediately apparent. The link between talk and policy outcomes is far more tangible in a deliberative system than in representative democracies. 6 2 The first-time participant in a conversational forum resulting in a policy decision will surely experience enhanced feelings of external efficacy. Moreover, this belief that one's views will inform decisions affects future choices about participation in deliberations. Thus the relationship between external efficacy and formal deliberation appears to be non-recursive. It seems particularly unlikely that the same holds true for casual conversation since there is no obvious link between the content of one's discussions and policy decisions. Since I therefore expect discussion to have no impact on external efficacy, this variable appears only in the first set of tests contained in Chapter 3. The relationship between internal efficacy and informal political conversation is more complex. An individual's perception of his political competence not only affects the likelihood of his participation in politics, but should also be affected by these conversations. Those who feel they lack the knowledge or intelligence to participate fully in conversations are less likely to participate when such discussions arise. On the other hand, participation in political conversations should result in an improved sense of internal efficacy as a result of the information and constraint acquired in conversations (Praktis & Turner, 1996). Some general comments on the relationship between internal efficacy and discussion may be possible after testing the following hypothesis: H5.7: Discussions about politics result in enhanced feelings of internal efficacy. Method To test whether or not participating in political conversations affects an individual's sense of his ability to understand politics I shift internal efficacy to the other side of the equation specifying it as the dependent variable. I continue to use level of agreement with the following statement as my measure of internal efficacy: "Sometimes politics and government seem so complicated that a person like me can't really understand what's going on". Results If discussion does in fact involve and encourage deliberative thought then those who participate in conversations should have more positive feelings of internal efficacy. The results reported in Table 5.4, however, fail to offer support for this assertion. Neither 63 sort of discussion is related to reported feelings of internal efficacy. A number of other variables do serve as significant predictors including gender, income, interest in politics, and external efficacy. Education is also a significant and powerful predictor with a coefficient of-0.33. If some conversations affect ability in the same way as education, then the connection between discussion and efficacy may still exist. These results, however, suggest that the kind of conversations about politics in which people engage do Tab le 5.4 Est imat ing effects of d i scuss ion on internal ef f icacy ( O L S R e g r e s s i o n ) Dependen t Va r iab le Internal Ef f icacy (0 -D-S tandard B Errors Political Information Sources Discuss i on F requency Fr iends and Re la t i ves -0 .025 0.030 Other P e o p l e 0.016 0.027 F requency of D i sag reemen t 0 .013 0.023 M e d i a U s e 0.003 0.030 Demographics A g e 0.144 0.048 M a l e -0.038 0.014 Income -0.453 0.137 Educa t ion -0.333 0.040 S p o u s e / P a r t n e r 0 .003 0.014 L a n g u a g e S p o k e n at H o m e F rench 0 .023 0.017 Nei ther F rench nor Eng l i sh 0.065 0.032 Attitudes Interest in Pol i t ics -0.179 0.032 Externa l Ef f icacy 0.253 0.020 Pol i t ical K n o w l e d g e -0.030 0.004 Cons tan t 0.786 0.041 N 2029 Ad jus ted R 2 0.243 A l l va ra ib les c o d e d 0-1 Bo ld Indicates p<0.05 "Indicates p<0.1 not effect internal efficacy through ability. Given that discussion, internal efficacy, knowledge, and interest each affect each other, it is not possible to rule out whether or not 64 some of the benefits of discussion are occluded by the inclusion of other variables. Future research designed with the specific intent of sorting out the effects of these variables might be better able to describe the relationship between internal efficacy and political conversations. Based on the evidence reported here, however, informal discussion appear to lack the content or processes behind the effect formal deliberation has on internal efficacy. Interest in the Campaign: Hypotheses The relationship between interest in politics and political discussions is fundamental to many of the outcomes investigated here. People think about, talk about, search out information about, and participate in, things they find interesting. Interest is certainly irrelevant to some behaviours, but is fundamental in explaining optional behaviours such as engagement with politics. Explaining why people become interested in certain things is no easy feat. Possible explanations of how talk enhances interest include: First, knowledge of, and interest in, a topic of concern to others provides individuals with opportunities for social interaction. The Spring visitor to Canada hoping to interact with sports bar patrons would do well to learn about the N H L playoffs. Second, a cognitive approach suggests that the more frequently one activates a packet of information, the more likely those same concepts will become active again. Thus an engaging conversation about politics should motivate participants to enter into further such discussions. Third, conversation about politics may provide previously unknown information which causes an individual to re-evaluate their existing interest in politics. News about a government contract that may be awarded to a firm in one's town might generate a higher level of interest in politics. Similarly, exposure to an alternative view may force one to reconsider a significant number of political considerations and opinions and in doing so stimulate one's interest in other perspectives not previously considered. This list is by no means exhaustive. The reciprocal relationship, that interest affects participation in political conversations, is far more intuitive and has been discussed earlier. To test the validity of this theory: 65 H5.8: Discussion frequency is positively related to interest in the campaign. H5.9: Participation in discussions about the election campaign results in an increase in respondent's reports of their interest in the campaign. Method To investigate the relationship between interest in politics and discussion, I estimated two multivariate regressions. The dependent variable in the first equation is responses to a question quantifying interest in the campaign and asked during the post-election wave. The dependent variable in the second equation is change in reported interest in the campaign between responses to the campaign and post-election waves coded -1 to 1. In addition to the usual controls, I include campaign wave date of interview (0-36). Results From Table 3.3 it is clear that one's interest in politics is related to the frequency with which one enters into political discussions. Again, since we cannot specify general interest as temporally prior to conversation, identifying causality is not possible. Table 5.5 reports the output of a regression predicting post-election interest in the campaign. Both types of discussion are clearly significant; the coefficient for friends and family is .15 and .05 for other people providing support for H5.8. Media use is the most powerful predictor with a coefficient of .49 which is quite large given the independent variable ranges 0-1. Not surprisingly, general interest in politics also plays an important role. By controlling for the most powerful predictors of discussion, namely interest in politics and media use, it seems likely that the conversations themselves are exerting the effect. The causal relationship may become clearer by assessing the effects of discussion on interest levels reported during and after the campaign. The presence of an 'interest in the campaign' question in two of the three survey waves allows me to assess the interest -discussion relationship from a slightly different perspective. Since conversations of interest here took place prior to the respondent's second evaluation of her own interest in the campaign, I can look at the role discussion 66 plays in explaining this measure of interest. Of those surveyed, 46% were less interested in the campaign after the fact than they were during the campaign, 26% Tab le 5.5: Est imat ing effects of d i scuss ion on interest in the campa ign ( O L S Regress ion ) . Dependen t Va r iab le Dependen t Va r i ab le Post -e lec t ion report of interest in the C h a n g e in interest in campa ign (0-1) the c a m p a i g n (0-1). S tandard S tandard B Errors B Errors Political Information Sources Discuss i on F requency Fr iends and Re la t i ves 0.152 0.017 0.0610.02 2 Other P e o p l e 0.046 0.015 0 .0200.02 0 F requency of D i sag reemen t 0 .007 0.013 -0 .033 0.017 M e d i a U s e 0.487 0.017 0.2510.02 2 Demographics A g e 0.042 0.027 -0 .038 0 .035 M a l e 0 .008 0.008 0 .020* 0 .010 Income -0 .0250.07 8 0 .0460.10 1 Educa t ion -0.0810.02 3 -0 .028 0.030 S p o u s e / P a r t n e r -0 .0110.00 8 -0.011 0.011 L a n g u a g e S p o k e n at H o m e F rench -0.0210.00 9 -0.030 0.012 Nei ther F rench nor Eng l i sh 0.005 0.019 -0 .013 0.024 Attitudes Interest in Pol i t ics 0.174 0.018 -0.164 0.024 Internal Ef f icacy 0.002 0.013 -0 .008 0.016 Externa l Ef f icacy -0.0400.01 2 0.0070.01 5 Pol i t ical Know ledge 0.005 0.002 0.0110.00 3 Date of Interview - 0.0010.00 1 Cons tan t 0.109 0 .025 -0.174 0.035 N 2021 2014 Ad jus ted R 2 0.546 0 .113 Al l var iab les c o d e d 0-1 Bo ld Indicates p<0.05 * Indicates p<0.1 maintained the same level of interest, and only 18% became more interested6. These results may be largely due to the specific nature of the 1997 campaign and the style of 6 Given that interest in both cases was measured by a ten point scale, much of the observed variation in pre and post election responses is l ikely the result o f measurement error. 6 7 coverage employed by the media. Upon reflection in the post-election survey most voters may have indicated less overall interest in the campaign since the election results did not differ much from the previous election and the outcome seemed easily predicted at the outset. Higher interest in the campaign may well speak to the effectiveness of the various news outlets in framing the election as a competitive affair subject to significant changes. Horserace coverage may have provided the appearance of an election with an uncertain outcome while the ultimate results confirmed what most had expected all along. Table 5.5 reports that the higher one's interest in politics generally, the more likely one's reported campaign interest declined by the end of the campaign. One very interested in politics generally may have been more likely to evaluate the 1997 campaign as a comparatively boring election. On the other end of the general interest scale, many likely saw this election as no different from others because they pay little attention to all elections. That media use is by far the most powerful predictor of change in interest lends further support to this idea. If this explanation holds any water, then it becomes difficult to interpret the effects of discussion. If people reported higher levels of interest while the campaign was happening because of stirring media coverage, discussion could have diminished interest if one's discussion partner presented the idea that the composition of Parliament after the election would be virtually the same as it was before all the fuss. Having offered this complex bit of speculation, the results suggest just the opposite. Discussion with friends and family did in fact contribute to increased interest in the campaign. Those speaking with friends and relatives often were 6.7% more likely to report increased interest in the campaign. Again, however, discussion with others is not significant, reinforcing the emerging picture that talking about politics with one's close associates matters, while talking with 'other people' rarely does. Political Participation: Hypotheses Participation lies at the heart of deliberative democracy. What makes deliberative democracy unique is both the extensive and direct participation of citizens and that this participation consists, above all else, of talking about politics. The democratic citizen 68 must participate in politics extensively and citizenship stands "for something more than taxpaying and voting" (Barber, 1984, p. xvi). Verba and Nie offer one of the more widely cited definitions of this concept: "political participation is activity that is intended to or has the consequence of affecting either directly or indirectly, government action" (Verba & Nie, 9). Despite such a broad definition few scholars have conceived of discussions about politics as participation. The presence of intent to affect government action seems to inform most efforts to operationalize participation and thus activities such as writing letters to representatives, campaigning for parties, and signing petitions are the sort of participation usually studied. If, however, we are concerned with participation more broadly conceived, interested in possible impacts, and less concerned with intent, talking about politics surely qualifies as participation. The relationship between talk and other forms of participation can be explained in at least three ways. First, talking with others can foster social capital which brings with it a desire to become more involved in one's community. In addition, political conversations may serve as useful tools for those engaged citizens to recruit others to participate (Scheufele 2002). A second possible explanation understands interest as the variable mediating the talk-participation relationship. Finally, discussion may also affect participation through information: "knowledge allows citizens to make informed decisions, it creates a sense of civic duty among citizens, and it increases familiarity with bureaucratic institutions and political processes" (Scheufele, 2002, p.52). The same author reports evidence in support of both the direct relationship and the talk-knowledge-participation explanation (Scheufele, 2000, 2002). In the present study, I consider political participation by measuring whether or not respondents cast a ballot. While there are numerous theories of non-voting behaviour7, discussion might influence this decision in a number of ways. First, discussion of politics enhances one's sense of membership in a political community. Those who understand themselves as part of the world of politics are more likely to undertake the most symbolic act of citizenship: voting. Second, people may identify their preferred party during discussion and subsequently will vote in order to ensure cognitive consistency. Third, 7 See for example: C R I C , Pammet, Blais, Nevitte. 69 the enhanced interest discussion fosters increases the likelihood that one will vote. Thus for a number of reasons: H5.10: 1993 non-voters that engage in more political conversations are more likely to cast a ballot in 1997. H5.11 Those voting in the 1993 election will be less likely to be a 1997 non-voter the higher their reported frequency of discussion. Method I divided the sample into those reporting they had cast a ballot in the 1993 Federal election and those who were non-voters. I then created two variables to measure the behaviour of members of these two groups in the 1997 election based on their self-reported voting behaviour from the post-election wave. By doing so, I am able to assess the relationship between discussion frequency with voting behaviour. The results of the regression are reported in Table 5.6. A l l those who did not vote in the 1993 election are included in the model reported in panel A . Those who voted in 1997 are coded as 0 with those who remained non voters coded as 1. Panel B refers to those who voted in 1993 with those who cast a ballot in 1997 coded as 0 and those who became a non-voter coded as 1. Results In both models, none of the three discussion variables achieve significance. While significant variables in the second model such as age, education, and external efficacy figure importantly into explanations of non-voting behaviour, none are significant in the first model. This fact, coupled with the quite low number of cases, suggests the results in panel A are of little use. While the second model presents more interesting results, there are at least two reasons in favour of a cautious interpretation. First, the measure of 1993 voting behaviour may not be accurate since self-reports are subject to incorrect recall with more respondents who are unsure selecting the more socially desirable response. Second, the various distinct reasons for non-voting may result in noise that obscures the impact of discussion. In some instances, such as habitual non-voting, a lack of discussion may matter. In others, such as being outside one's riding 70 on election day, discussion likely plays no role. Voting behaviour is surely the result of a complex, and heterogeneous, chain of causality. A more in-depth battery of voting and discussion related questions is necessary for an improved analysis of the relationship between the two. Tab le 5.6 Est imat ing effects of d i scuss ion on voting behav iour ( O L S Regress ion ) . A B Dependen t Va r iab le Dependen t Va r iab le 1993 Non-voters . 0=Voted in 1997 1=Remained a non -voter. S tandard B Errors 1993 Vo te rs . O V o t e d in 1997 1 = B e c a m e a non -voter. S tandard B Errors Political Information Sources Discuss i on F r e q u e n c y Fr iends and Re la t i ves 0.208 0.230 -0.001 0 .045 Other P e o p l e -0 .055 0.215 -0 .013 0.041 F requency of D i sag reemen t -0 .222 0.189 -0 .032 0.034 M e d i a U s e -0.964 0.223 -0.142 0.044 Demographics A g e 0 .612* 0.630 -0.211 0.074 M a l e 0.178 0.100 0.016 ' 0.021 Income -1.164 1.657 0.167 0.178 Educa t ion -0 .417 0.332 -0.130 0.062 S p o u s e / P a r t n e r -0 .017 0 .115 0.011 0.021 L a n g u a g e S p o k e n at H o m e F rench -0.343 0.142 -0 .007 0.024 Nei ther F rench nor Eng l i sh 0.172 0.158 -0 .027 0.048 Attitudes Interest in Pol i t ics 0.078 0.249 -0 .006 0.049 Internal Ef f icacy 0.004 0.175 -0.076 0.034 Externa l Ef f icacy -0.024 0.178 0.100 0.031 Pol i t ical K n o w l e d g e 0.033 0.039 -0.016 0.006 Cons tan t 0.840 0.320 0.413 0.069 N 80 854 Ad jus ted R 2 0.265 0 .065 A l l va r iab les c o d e d 0-1 Bo ld Indicates p<0.05 * Indicates p<0.1 71 Tolerance: Hypothesis Mutz and Martin suggest one benefit of exposure to dissimilar views is the legitimation of unwanted political outcomes (Mutz & Martin, 2001). Being exposed to the reasons supporting a conflicting opinion should enhance an individual's perception that political decisions in favour of this opinion are legitimate and not groundless. Discussions can enhance this process by exposing the justifications supporting differing views. Hearing about a political issue from a concrete other, especially if the information provided includes personal experiences, may make abstract problems and positions far more real to the receiver. The effect of discussion, however, surely depends on with whom one is engaged in discussion. An individual whose network of political discussion partners contains few dissimilar views may well become less tolerant given their insulation from opposing ideas. As such: H5.12: Those frequently participating in political discussions report higher levels of tolerance. The effect is stronger for conversations with other people than for conversations with friends and family. H5.13: The frequency of reported disagreement is positively related to tolerance. Method In the mail back portion of the survey respondents were asked the extent to which they agree or disagree that 'we should be more tolerant of people who choose to live according to their own standards, even if they are very different from our own'. I then use this tolerance measure as the dependent variable in a multivariate regression involving the same independent variables as appear in the other models in this chapter. Results Table 5.7 presents results useful in highlighting the difference between frequency and content of political conversations. Frequency of discussion with neither group produces significant coefficients. Thus, how often one engages in casual political 72 Tab le 5.7 Est imat ing effects of d i scuss ion on to lerance ( O L S Reg ress ion ) . Dependen t Va r iab le To le rance 0=low, 1 = high S tandard B Errors Political Information Sources Discuss ion F requency F r iends and Re la t i ves 0.008 0.031 Other P e o p l e -0 .039 0.028 F requency of D i sag reemen t 0.066 0.024 M e d i a U s e 0.026 0.032 Demographics A g e -0 .089* 0.051 M a l e -0.031 0.015 Income -0 .110 0.168 Educa t ion 0.133 0.044 S p o u s e / P a r t n e r -0 .018 0.015 L a n g u a g e S p o k e n at H o m e F rench 0.010 0.018 Nei ther F rench nor Eng l i sh 0.003 0.039 Attitudes Interest in Pol i t ics 0 .005 0.034 Internal Ef f icacy 0.005 0.024 Externa l Ef f icacy -0.081 0.022 Pol i t ical K n o w l e d g e 0.002 0.004 Cons tan t 0.580 0.049 N 1243 Ad jus ted R 2 0.033 A l l vara ib les c o d e d 0-1 Bo ld Indicates p<0.05 * Indicates p<0.1 conversations does not affect one's level of tolerance. The substance of conversations does matter, with frequency of disagreement leading to greater tolerance. Those who found disagreement often were 6.6% more likely to agree with the statement affirming tolerance. Disagreement indicates participant acknowledgement of the presence of conflicting views. The reasons offered in support of the various positions will, in some cases, result in persuasion. In most instances, however, the expression and consideration 73 of these facts and arguments enables those who continue to disagree to accept such views as reasonable. The same explanation applies to education's effect. Stouffer argues that education increases tolerance since "schooling puts a person in touch with people whose ideas and values are different from one's own" and teaches one that diversity of opinions and ideas are a good (Stouffer, 1955, p. 127). This argument is borne out by the fact that education is by far the most powerful predictor in the model. That media exerts no effect is surprising and questions the notion that media supply dissimilar views and instigate an internalized conversation with these views. Summary In this chapter I have used a variety of dependent variables in an effort to assess the effects of participation in political conversations. Taken together, the results do not tell a coherent story. Conversations with others do not play an important role in all but one of the outcomes studied here. Disagreement predicts policy opinionation and tolerance, as expected, but fails to affect the other dependent variables included. The most effective predictor of deliberative outcomes is frequency of conversation with friends and family. These conversations play a role in predicting political knowledge, gains in political information over the course of the campaign, opinionation on value issues, and both measures of interest in the campaign. While these conversations do involve some disagreement and effect some deliberative outcomes, their impact is relatively weak. Talking with one's friends and relatives appears to serve a simplified information gathering function rather than the more onerous and widely informed critical processing of information characteristic of formal deliberation. The implications of this fact and an analysis of the overall results contained in this paper are the subject of the following, concluding, chapter. 74 CHAPTER 6: Conclusions This study of conversation among citizens during the 1997 Canadian federal election appears to have raised at least as many questions as it sought to answer. Moreover, these initial questions remain, for the most part, unanswered. This latter fact comes as no surprise since to suggest a small project such as this might resolve the relationship between informal discussion and democracy is distinctly unrealistic. Instead, the findings reported here are simply a small piece in the vastly complex puzzle of the citizen's role in a democracy. With this in mind, I offer a number of broad conclusions in order to both clarify the implications of this study and identify some of the questions arising from it. Informal Discussion in a Canadian Federal Election Earlier in this paper I drew on Schudson's critique of political conversations in highlighting the importance of the 'who' and the 'how' of discussion. For conversations to be of democratic value, they must involve participants from diverse backgrounds and holding conflicting initial preferences. If "politics is the art of engaging strangers in talk" then rather few Canadians engage in politics and they do so infrequently. (Barber, 1984, p. 189.) Discussions with other people exhibit very few effects suggesting, beyond the fact that fewer such discussions took place, that diversity of opinion was either absent, unrecognized, or not confronted. The limited effects of such discussions may be the result of inaccurate reporting of discussion frequency. It might also be the case, however, that the frequency of the two sorts of conversation are similar but that conversations with others involve quite different content and accordingly exhibit weaker effects. Discussions with family and friends did play a role in explaining a number of behaviours. The effects in most cases, however, were fairly weak. Moreover, the less powerful effect of such conversations at predicting disagreement lends further support to the notion that most conversations take place among those with similar preferences. Schudson also argues that democratic discussions are ends oriented and therefore involve significant conflict. That almost 40% of respondents never encounter disagreement indicates that informal discussions fail in this respect as well. 75 What accounts for disjuncture between my hypotheses and these results? Two explanations seem plausible. First, the method of evaluating discussions employed here may obscure what is actually happening. The survey was not designed to answer these questions specifically. Other researchers have designed surveys with the explicit intent of investigating the role of interpersonal communication and accordingly have produced more concrete conclusions, (see for example: Kim & Wyatt, 1999; Wyatt et al, 2000; Scheufele, 2000 & 2002) The key concept itself is also difficult to measure. It is virtually impossible, and decidedly so given ethical limits, to measure casual political conversations as they actually happen. To understand the effects of political conversations, however, we must assess three things. First, what is the actual content of the conversation and how long does the conversation lasts. Second, how frequently does an individual engage in such discussions. Third, who is participating and what are their prior opinions. Taken together, such a measurement task seems overwhelming. Improvements in the right direction, however, are surely possible. One simple improvement would be to ask respondents to report how frequently they discuss a diverse range of topics on the same scale to get a more realistic appraisal of how often people actually talk about political matters. While it is reasonable to suggest that a different method might have produced different results, it seems unlikely, given the weight of other research, that the findings would differ in a truly significant way. The second, and more plausible, explanation is that conversations in which people engaged during the election were distinctly unlike those expected in deliberative theory. People generally talk with others holding similar preferences and do not engage in ends-oriented discussion of issues from a variety of perspectives. These conversations do have important effects; most notably on political knowledge, interest in the campaign, and tolerance. But it seems clear that citizens are not engaging in the sort of talk Benjamin Barber praises in the quotation included at the outset of this paper. What do these findings mean for deliberative theory? These, and other, findings do not mean empirical political scientists should dismiss thoughts on deliberation and return the concept to the less encumbered world of theory. Again, because deliberative theory does not apply to democracy as actually practiced does not mean the theory should be rejected. It does seem safe to say, however, that the chances of existing democracy 76 becoming more deliberative through organic means are distinctly slim. I now take a step back and consider the implications of this study for the broader topic of how citizens process political information. Citizens, Media, and Talk The relationship between media and discussion surfaces repeatedly in this paper. Media use is a strong predictor of both discussion with friends and family and with others suggesting that the two share an important relationship. Since media's effect is independent of interest in politics it seems clear that exposure and acquisition of information from the media motivates discussion. At the very least, acquiring political information from the media makes one better able to participate in conversations since media headlines serve as the starting point for most discussions. The print superiority theory does not apply to Canada. Rather, in instances where previous work in the United States suggested newspaper use should outperform television viewing, the opposite result occurred. Mutz and Martin present an intelligent explanation for the differing effects of print media in different countries by noting that media with clear political biases enable the recipient to expose themselves selectively to information (Mutz & Martin, 2001). Norris also suggests the public broadcasting ethic explains why British television seems to have a greater informational impact than commercially oriented American stations. As is often the case, Canada may well fall between these two cases being home to both newspapers with less explicit bias than Britain yet greater than those in the US, and a television public broadcaster less dominant than the B B C but of significantly greater relevance than PBS. Research that explores how different media channels assume different models should prove quite interesting. Moreover, the relationship between different mediums in different countries and political discussion also deserves attention. More specifically, what type of media content generates discussion is an important question. How does political learning from newscasts and stories differ from Sunday morning talk shows and opinion columns? Experimental and quasi-experimental research hold significant promise for investigations into this topic. Selective perception also explains the significantly different effects of mass media and interpersonal communication. Research here seems to support the idea that most 77 political conversations consist of consensual discussion amongst homogeneous participants. Formal deliberation is thus more truly democratic since it avoids the potential for selective exposure by forcing individuals to confront conflicting opinions. Since casual conversation fails to meet this important criterion one could therefore assume that the media is better able to expose citizens to dissimilar views. Before we can consider the mass media as an effective alternative to the interpersonal communication so central to democratic theory, we must first establish the extent to which exposure to dissimilar views through the media motivates deliberative consideration of these views. Given obstacles to frequent interaction with diverse others such as geography, the media surely has an important role to play in the dissemination of information about the political world. Just as I have used an idealized type of conversation to suggest what discussions can do, one can envision a model media program which motivates viewers to deliberate over a number of diverse perspectives. The enhanced potential for two way communication presented by newer technologies may well improve the media's ability to encourage deliberative thought. For herein lies the key, if democratically informed citizens are a good, then whatever motivates deliberative thought is a worthy avenue of investigation. Deliberative Thought and Democratic Citizens In explaining the democratic value of discussion, Barber identifies the importance of deliberative thought in writing: "Indeed, one measure of healthy political talk is the amount of silence it permits and encourages, for silence is the precious medium in which reflection is nurtured and empathy can grow" (Barber, 1984, p. 175, original emphasis.). Despite the fact that I devoted a considerable amount of earlier chapters to the concept of deliberative thought, my findings provide rather scant evidence in support of the concept's importance. Those speaking more frequently with friends and family scored higher on general information measures and gained more information over the course of the campaign. The same cannot be said of discussion with others. Thus it appears that the range of perspectives that informs the deliberative thought which does occur is rather narrow. Discussion's lack of impact on measures of constraint further suggests that informal political conversations are quite distinct from deliberations. While these 78 measures might be improved, the broader picture presented here supports an image of fairly uninformative political conversations. If citizens voices matter, be it in theory or in practice, then we must be interested in how individuals reach the conclusions they then voice. Beneath the specific questions about informal conversations investigated here is the broader topic of how citizens learn about politics. Deliberative thought is of specific interest since participation in discussions is one of the more effective means though which people learn. People learn more when they cognitively engage with information. While experiential learning is often a more effective pedagogical approach than informed discussion, in the case of politics, the two are in fact the same. To debate ideas, define problems, reconsider opinions, and discuss political matters is to engage in the practice of politics. Surely other communication methods, such as repetition, are both available and frequently employed. Messages from political elites to the masses, especially during elections, rely significantly on repetition and encourage citizens to employ heuristics rather than cognitively confront information. But developing political opinions through this process does not qualify one to meaningfully participate in decision making. If citizens are to participate, they must learn about politics in a more meaningful and public way. Like other behaviours, the occurrence of deliberative information processing depends on three conditions: motivation, ability, and opportunity. I now briefly mention each in turn. It seems fair to suggest that most citizens fail to engage in much deliberative thought about politics due to a lack of motivation. One's interest in politics is an important aspect of motivation to think and learn about politics. The potential for a more deliberative system hinges on the assumption that politics is, in fact, interesting. If this is the case, then talking about politics and becoming more interested in the political world is a reinforcing cycle. Einstein may be right that politics is more complicated than physics, but it had also better be a lot more interesting or people will lack the motivation to continue to become informed and participate. It is useful to conceive of interest as a continuum between two types. On one hand, interest can be grounded in a more abstract appreciation for politics while on the other, interest is motivated by a concern for the effect of government decisions on one's life. A citizen might well become interested in politics based on an understanding of the relationship between political decisions and her 79 own life regardless of whether or not she finds the subject to be of interest when there is no link between a political choice and her own circumstances. Greater participation in political decision making should result in an increase in both sorts of interest; interest being important as it motivates the deliberative thought that enables a citizen to fully participate. This paper indicates, however, that greater participation and interest may well have deleterious results. Berelson, Lazarsfeld, and McPhee's warning must be heeded. Diminished political apathy may well result in a more divisive society. Given the evidence in support of the spiral of silence at present levels of apathy, it seems fair to suggest that informal conversations may be less useful, and perhaps even detrimental, if people care passionately about political matters. If talk matters, then social norms would likely change. Indeed, when people know that their own opinions, and those of others, can impact government decisions, psychological processes such as social conformity and social cohesion will likely result and present obstacles to the effective functioning of a deliberative society. Efforts to improve citizen participation and reduce political apathy must closely consider potential for unwanted effects. The frequency and depth of deliberative thinking also depends on the thinker's ability. Greater constraint should of course facilitate deliberative thought, but so too should critical thinking skills. While most are surely capable of thought that can be described as deliberative, some do so far more efficiently and thoroughly because of various skills. Even the meta-cognitive awareness that one is seeking to solve a problem should improve the quality of one's thinking. Those lacking such skills must find politics extremely confusing given that most any piece of political information is subject to conflicting interpretations. Attempting to make sense of a normative world with a descriptive mind is surely a recipe for confusion. If a more informed citizenry is desirable, then the education system, educational psychologists, and political scientists have much to learn from one another in ensuring that citizens are trained in the skills necessary to be a deliberative thinkers and active citizens. Finally, thinking about politics in a meaningful way requires opportunity. While the two are surely related to motivation in important ways, my comments on both media and discussion effects are grounded in an interest in how different information sources 80 create the conditions for deliberative thought. Earlier on I asserted that discussions are better able to encourage deliberative thought due to the two-way nature of communication. The weak results should not be interpreted as a rejection of this idea. Rather, it seems clear that casual conversations do not contain the content and processes that foster deliberative thought. If citizens are to continue to play an increasingly important role in the political decision making process, based either on normative claims about democratic theory or more practical concerns about growing apathy and declining deference, then we must be interested in how they arrive at political opinions and that these opinions are the result of meaningful consideration of different political perspectives. The content and channel qualities of different political information sources affect how receivers process the information communicated. Efforts to further understand the implications of formal and informal political conversations can make an important contribution to our understanding of how citizens learn about politics. 81 Bibliography Avery, R. K, & McCain, T. (1982). Interpersonal and mediated encounters: A reorientation to the mass communication process'. In G. Grumpert & R. Cathcart (Eds.), Inter/Media: Interpersonal communication in a media world. New York: Oxford UP. Barber, B. R. (1984). Strong democracy: Participatory politics for a new age. Berkeley: University of California Press. Bartels, L . M . (1996). Uninformed votes: Information effects in presidential elections". American Journal of Political Science, 40(1 \ 194-230. Bennett, S. E., Flickinger, R. S., & Rhine, S. (2000). Political talk over here, over there, over time.' British Journal of Political Science, 30, 99-119. Berelson, B. R., Lazarsfeld, P. F . , & McPhee, W. N . (1954). Voting: A study of opinion formation in a presidential campaign. Chicago: University of Chicago Press. Blais, A . (2001) To vote or not to vote: the merits and limits of rational choice theory. Pittsburgh: University of Pittsburgh Press. Bonham, J. (1996). Public deliberation: Pluralism, complexity, and democracy. Cambridge, M A : MIT Press Capella, J. N . , Price, V. , & Nir, L. (2002). Argument repertoire as a reliable and valid measure of opinion quality: Electronic dialogue during campaign 2000. Political Communication, 19. 73-93. Capella, J. N . , Price, V. , & Nir, L. (2002). Does disagreement contribute to more deliberative opinion?. Political Communication, 19. 95-112. Cutler, F. (2000) Explaining geographic variation in electoral behaviour: Local environments and Canadian Voting. (Phd Dissertation), [online] Available: http://www.politics.ubc.ca/fcutler/research/dissertation.htm. [2002, September 20.]. de Boer, C , & Velthuijsen, A . S. (2001). Participation in conversations about the news. International Joirnal of Public Opinion Research, 13(2). 140-158. Delli Carpini, M . X . & Keeter, S. (1989). What americans know about politics and why it matters. New Haven: Yale University Press. Fearon, J. D. (1998). Deliberation as discussion. In J. Elster (Ed.) Deliberative democracy. Cambridge: Cambridge University Press. 82 Fishkin, J. (1995). The voice of the people: Public opinion and democracy. New Haven, CT: Yale University Press. Fournier, P. (2002). The uninformed Canadian voter. In J. Evritt & B. O'Neil (Eds.) Citizen politics: Research and theory in Canadian political behaviour. Don Mills: Oxford UP. Glynn, C. J., Hayes, A . F. & Shanahan, J. (1997). Perceived support for one's opinions and willingness to speak Out: A meta-analysis of survey studies on the 'spiral of silence'. Public opinion quarterly, 61. 452-463. Glover J. A . & Corkill, A . J. (1990) The implications of cognitive psychology for school psychology. In T. B. Gutkin & C. R. Reynolds (Eds.), The handbook of school psychology, (pp. 104-123). New York: John Wiley & Sons. Granovetter, M . S. (1973). The strength of weak ties. American Journal of Sociology. 78(6). 1360-1381. Green, M . C , Krosnick, J. A . , & Holbrook, A . L. (2002). The survey response process in telephone and face-to-face surveys: Differences in respondent satisficing and social desirability response bias, [online] Available: http://www.umich.edu/~ nes/resources/techrpts/techrpts.htm. [2002, September 20.] Hansen, S. B. (1997). Talking about politics: Gender and contextual effects on political proselytizing. The Journal of Politics, 59 (1). 73-103. Huckfeldt, R., Johnson, P. E., & Sprague, J. (2002). Political environments, political dynamics, and the survival of sisagreement.' The Journal of Politics, 64(1). 1-21. Huckfeldt, R., Sprague, J., & Levine, J. (2000). 'The dynamics of collective deliberation in the 1996 election: Campaign effects on accessibility, certainty, and accuracy.' American Political Science Review, 94(3). 641-649. Huckfeldt, R., Beck, P. A. , Dalton, R. J., Levine, J., & Morgan, W. (1998). Ambiguity, distorted messages and nested environmental effects on political communication. The Journal of Politics. 60(4). 996-1030. Huckfeldt, R. & Sprague, J. (1995) Citizens, politics, and social communicatioa Cambridge: Cambridge UP. Huckfeldt, R. Beck, P. A. , Dalton, R.J. & Levine, J. (1995). Political environments, cohesive social groups, and the communication public opinion. American Journal of Political Science. 39(4). 1025-45. Ikeda, K; & Huckfeldt, R. (2001). Political communication and disagreement among citizens in Japan and the United States. Political Behavior. 23 (1). 23-51. 83 Kim, J., & Wyatt, R. O. (1999). News, talk, opinion, participation: The part played by conversation in deliberative democracy.' Political Communication, 16(1). 361-386. Katz, E. & Lazarsfeld, P. F. (1955). Personal influence. Gelcoe: Free Press. LaDuke, R., & Huckfeldt, R. (1998). Social capital, social networks, and political participation. Political Psychology, 19 (3). 567-83. Levine, J. M . & Russo, E. (1995). The impact of anticipated interaction on information acquisition. Social Cognition, 13(3). 293-317. Luskin, R. C. (1990). Explaining political sophistication. Political Behavior, 12(4). 331-361. Luskin, R. C. (1987). Measuring political sophistication. American Journal of Political Science, 31(4). 856-899. MacKuen, M . (1990). Speaking of politics: Individual conversational choice, public opinion, and the prospects for deliberative democracy. In J. A . Ferejohn & J. H Kuklinski (Eds). Information and democratic processes. Urbana: University of Illinois Press. Manin, B. (1987) On legitimacy and political deliberation. Political Theory, 15. 338-368. McLeod, J. M . & Scheufele, D. A . (1999a). Understanding deliberation. Communication Research, 26(6). 743-775. McLeod, J. M . , & Scheufele, D. A . (1999b). Community, communication, and participation: The role of mass media and interpersonal discussion in local political participation. Political Communication, 16(3). 315-337. Mendelsohn, M . & Cutler, F. (2000) "The effects of referendums on democratic citizens: Information, politicization, efficacy and tolerance." British Journal of Political Science, 30, 685-701. Miller, R. L. , Wilford, R., & Donoghue, F. (1999). Personal dynamics as political participation. Political Research Quarterly, 52(2). 269-291. Mondak, Jeffrey, J. (1995). 'Media exposure and political discussion in U.S. elections. The Journal of Politics. 57(1). 62-85. Mutz, D. C. (2002). "Cross-cutting social networks: Testing democratic theory in practice. American Political Science Review, 96(1). 111-126. 84 Mutz, D. C, & Martin, P. S. (2001). Facilitating communication across lines of political difference: The role of the mass media. American Political Science Review, 95(T). 97-114. Mutz, D. C. (1998). Impersonal influence: How perceptions of mass collectives affect political attitudes. Cambirdge: Cambridge UP. Mutz, D. C. (1997). Mechanisms of momentum: Does thinking make it so?. The Journal of Politics. 59(1). 104-125. Nevitte, N . (1996). The Decline of deference: Canadian value change in cross national perspective. Peterborough: Broadview Press. Nevitte, N ; Blais, A. ; Gidengil, E.; & Nadeau, R. (1999). Unsteady state: The 1997 Canadian federal election. Don Mills: Oxford University Press Norris, P & Sanders, D. (2002). Message or medium?: Campaign learning during the 2001 British general election, [online] Available: http://ksghome.harvard.edU/~.pnorris.shorenstein.ksg/ACROBAT/MessageorMedium.pdf [2002, September 13.] Noelle-Neuman, E. (1974). The Spiral of Silence: A Theory of Public Opinion. Journal of Communication, 24. 43-51. Noelle-Neuman, E. (1993). The spiral of silence: Public opinion - our social skin. Chicago: University of Chicago Press. Page, B. I. (1996). Who deliberates: Mass media in modern democracy. Chicago: University of Chicago Press. Page, B. I., & Shapiro, R. Y . (1992). The rational public: Fifty years of trends in Americans' policy preferences. Chicago: University of Chicago Press. Pammett, J. H. (1997) The voters decide. In A . Frizzell & J. Pammett (Eds.) The Canadian general election of 1997. Toronto: Dundurn. Pattie, C. & Johnston, R. (2000). 'People who talk together vote together': An exploration of contextual effects in Great Britain. Annals of the Association of American Geographers, 90(1). 41-66. Pattie, C , & Johnston, R. (2001). Talk as a political context: Conversation and electoral change in British elections. Electoral Studies, 20. 17-40. Popkin, S. L. (1991). The reasoning voter. Chicago: University of Chicago Press. 85 Praktis, A . R. & Turner, M . E. (1996). Persuasion and democracy: Strategies for increasing deliberative participation and enacting social change. Journal of Social Issues, 52(1). 187-205. Pressley, M . with McCormick, C. (1995). Advanced educational psychology for educators, researchers, and policymakers. New York: HaperCollins. Price, V. & Neijens, P. (1997). Opinion quality in public opinion research. International Journal of Public Opinion Research, 9(4). 336-359. Putnam, R. D. (2000). Bowling alone: The collapse and revival of American community. New York: Simon & Schuster. Robinson, J., & Davis, D. (1990) Televison news and the informed public: A n Information Processing Approach', Journal of Communication 40(3): 106-119. Robinson, J. P., & Levy, M . R. (1986). Interpersonal communication and news comprehension. Public Opinion Quarterly, 50. 160-175. Scheufele, D. A . (2002). Examining differential gains from mass media and their implications for participatory behavior. Communication Research, 29(1). 46-65. Scheufele, D. A . (2000). Talk or conversation?: Dimensions of interpersonal discussion and their implications for participatory democracy. Journalism and Mass Communication Quarterly, 77(4). 727-743. Schudson, M . (1997). "Why conversation is not the soul of democracy. Critical Studies in Mass Communications, 14. 297-309. Schudson, M . (1982). The ideal of conversation in the study of mass media' in G. Grumpert & R. Cathcart (Eds.) Inter/Media: Interpersonal communication in a media world. New York: Oxford UP. Simonson, P. (1996). Dreams of democratic togetherness: Communication hope from Cooley to Katz." Critical Studies in Mass Communication, 13. 324-342. Sniderman, P., Brody, R. A. , Tetlock, P. (1991) Reasoning and choice: Explorations in political psychology. New York, NY:Cambridge University Press. Sniderman, Paul., Simon Jackman and Vincent Tiberj. (2001). Democratic Discussion: The Role of Reasons and Political Sophistication in Political Argument, [online] Available: http://tamarama.stanford.eduysimon/papers/discussionapsaQ 1 .pdf [2002, September 24.]. Sotitovic, M . & McLeod, J. M . (2001). Values, communication behavior, and political participation. Political Communication, 18. 273-300. 86 Stouffer, S. (1955). Communism, conformity and civil liberties. New York: Doubleday. Straits, B. C. (1991). Bringing strong ties back in: Interpersonal gateways to political information and influence. Public Opinion Quarterly, 55. 432-48. Stimson, J. (1995). Opinion and Representation American Politicl Science Review, 89(1). 179-183. Tarde, G. (1898/1969). On communication and social influence. (T.Clark, Ed. & Trans.). Chicago:University of Chicago. Ulbig, S. G. & Funk, C. L. (1999). Conflict avoidance and political participation. Political Behavior, 21(3). 265-282. Uslaner, E. (2002). The moral foundations of trust. Cambridge, M A : Cambridge University Press. Verba, S. & Nie, N .H . (1972). Participation in America: Political democracy and social equality. New York: Harper & Row. Weaver, D. H., Zhu, J., & Willnat, L . (1992). The bridging function of interpersonal communication in agenda-setting. Journalism Quarterly, 69(4). 856-867. Wyatt, R., Katz, E., & Kim, J. (2000a). Bridging the spheres: Political and personal conversation in public and private spaces. Journal of Communication, 50(1). 71-92 Wyatt, Robert, Elihu Katz, and Joohan Kim. (2000). How feeling free to talk affects ordinary political conversation, purposeful argumentation, and civic participation. Journalism and Mass Communications Quarterly, 77(1). 99-114. Wyatt, R., Katz, E., Levinson, H., & Al-Haj, M . (1996). The dimensions of expression inhibition: Perceptions of obstacles to free speech in three cultures'. International Journal of Public Opinion Research. 8. 229-447. Zaller, J. R. (1992). The Nature and Origins of Mass Opinion. Cambridge: Cambridge University Press. Zaller, J. R. (2002) The statistical power of election studies to detect media exposure effects in political campaigns. Electoral Studies, 21, 297-329. 87 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items