Beyond Opinion Leaders

Abstract

The number of Americans who report engaging in interpersonal persuasion during elections has drastically increased over the past decade. While past studies have demonstrated the impact of such proselytizing on vote choice, the author finds substantial evidence that it may also have larger democratic benefits, both for those attempting to persuade and for those whom they choose to target. Data from the 2008 National Annenberg Election Survey suggest that (a) attempting to persuade contributes to people’s ability to give reasons in support of both their own preferred candidate and the opposing candidate, and (b) persuasive conversation is a powerful channel for the spread of political information from the more engaged to the less engaged.

Keywords

interpersonal communication persuasion attempts political knowledge

The past decade has seen increasing attention paid to the role of informal political conversation in the democratic process (e.g., Eveland, 2004; Kim, Wyatt, & Katz, 1999; Moy & Gastil, 2006). Although political talk has long been understood as an important catalyst of opinion change (Berelson, Lazarsfeld, & McPhee, 1954), more recent work suggests that it also has beneficial outcomes for democracy as a whole. Even political conversation that does not meet the strict requirements of formal deliberation can lead to increased political tolerance (Mutz, 2006), awareness of oppositional rationales (Price, Cappella, & Nir, 2002), and political knowledge (Eveland, 2004).

However, there is no reason to expect all conversations about politics to be equally effective at producing these results. A casual chat between two like-minded friends, a heated debate between coworkers, and a highly involved partisan’s attempt to mobilize his apathetic brother might each affect its participants differently. Despite the wide variety of political conversation that exists in the real world, many academic studies of political talk group all these forms of talk under the larger umbrella of “political discussion.” The existing literature’s reliance on such broad measures means we still know little about what types of informal talk might have the largest effects on tolerance, knowledge, efficacy, or other democratic outcomes (Mutz, 2008).

This article examines the deliberative potential of interpersonal persuasion attempts: instances in which a person tries to show her friend, neighbor, coworker, or anyone else in her social network why he or she should vote for or against a particular candidate. Attempts to persuade are deliberate efforts to change someone else’s political preferences. Persuasion attempts can occur when a person tries (successfully or unsuccessfully) to persuade an undecided voter or when she tries to convince someone his currently preferred candidate is the wrong choice. Persuasion attempts are not the same as influence: Influence may occur without a conscious effort on the part of the influencer (Katz & Lazarsfeld, 1955). Persuasion attempts differ from regular political talk in that they are undertaken with the intention of changing attitudes rather than to inform, entertain, or deliberate.

Over the past decade, social networks have become more politically homogeneous (McPherson, Smith-Lovin, & Cook, 2001) and the American electorate has (arguably) become more polarized (Abramowitz & Saunders, 1998). Although one might expect this atmosphere of partisan self-selection to correspond with a decrease in persuasive conversation, the data show the opposite trend: Not only does persuasive conversation continue to thrive, it has actually become more common over the past decade.

Figure 1 shows the percentage of people in midterm and Presidential election years who reported that over the course of the campaign they “tried to show someone why they should vote for or against a particular candidate.” Prior to 2000, this number remained steady at about 20% during midterm elections and 35% in Presidential election years. Both of these figures rose by nearly 10 points in 2000 and 2002 and have remained at similar levels in the decade since.¹

Figure 1.

Rates of self-reported attempts to persuade, 1972-2008.

This article takes a first step toward understanding some of the results of those discussions, leveraging two large data sets that ask respondents whether they themselves tried to persuade as well as whether they were targeted for persuasion by someone they know personally. I find evidence that interpersonal persuasion attempts may, depending on the circumstances, help to bridge the divide between partisans and effectively spread political information. A person’s desire to recruit new supporters for his preferred candidate can spur him to engage in potentially uncomfortable discussions that occur across lines of political difference. Such cross-cutting discussions can in turn enhance participants’ awareness of oppositional arguments (Price et al., 1999) or help them clarify their own positions (Mutz, 2006). Attempts to persuade are also opportunities for the politically informed to share relevant information about candidate stances and issue positions with other members of their social network, especially those who are undecided, ambivalent, or politically uninvolved. Persuasive conversation may thus serve as a channel through which political information is conveyed from the more informed to the less informed.

The Deliberative Potential of Persuasive Conversation

A growing empirical literature has sought to connect the theoretical attributes of deliberative talk such as civility, disagreement with others, and face-to-face discussion to specific outcomes like political tolerance, knowledge gain, or civic engagement (see Mutz, 2008, and Thompson, 2008, for reviews). The concept of persuasion attempts as a form of political deliberation remains largely unexplored, as past studies of interpersonal persuasion have centered on how it might affect vote choice (Katz & Lazarsfeld, 1955) rather than its potential normative outcomes. Persuasive conversation does not meet every proposed requirement for deliberation. It is not (necessarily) public, the conversation is not governed by a set of objective rules, and no binding action emerges from a persuasion attempt. That being said, few (if any) real-world conversations fulfill every condition for deliberation, and so this article follows the model proposed by Thompson (2008), Mutz (2008), and Delli Carpini et al. (Delli Carpini, Cook, & Jacobs, 2004) by empirically testing the deliberative potential of a type of talk that does occur (and, in the case of persuasion, is becoming more and more common) rather than formulating conjectures about an ideal form that may never exist on any large scale.² The following section draws on both deliberative theory and empirical work in order to formulate testable hypotheses about the effects of participating in persuasive conversations. I focus on two particular outcomes that are particularly relevant to persuasive conversation: awareness of own and oppositional rationales and political knowledge.

Persuasion as a Catalyst for Cross-Cutting Discussion

Deliberative theory suggests that conversations between those who disagree can confer several benefits on their participants. Such talk may help citizens to sort out conflicting viewpoints on the issues of the day and come to a more enlightened position (Gutmann & Thompson, 2004). In addition, by exchanging opinions with people who have different political views, individuals may gain a more nuanced and empathetic perspective of the views of others (Fishkin, 1991). The evidence suggests that individuals who engage in cross-cutting discussion are better able to give reasons for their own opinions as well as reasons why someone might hold an opposing viewpoint (Mutz, 2006; Price et al., 2002).

Despite the positive consequences of exposure to diverse opinions, the tendency for people to associate and bond with those who are politically similar limits opportunities for cross-cutting discussion (Mutz, 2006). The more an individual is surrounded by others who hold similar opinions, the fewer chances he has to understand why someone might hold an alternative viewpoint. As a type of conversation that by definition requires a lack of consensus (if not outright disagreement) between its participants, persuasion attempts can be one such opportunity. Although many individuals express a preference for avoiding political conflict (Conover, Searing, & Crewe, 2002), the desire to win over new supporters for their candidate of choice may serve as an incentive for these citizens to break out of their usual reinforcing discussion network and seek out conversation partners who do not share their views. In the age of social networking websites that broadcast friends’ political preferences, identifying such targets—even among distant acquaintances—may be easier than ever.

Persuasion attempts are a unique subtype of cross-cutting conversation because they involve a conscious effort by one party to change the other person’s political opinions. A persuasion attempt requires both a “persuader”—someone who initiates the persuasion attempt—and a “persuadee”—the person who is targeted for an attempt. I predict that persuasion attempts will affect each of these parties differently: the persuader by enhancing his ability to argue for his own position, and the persuadee by exposing him to arguments for the opposing position.

The persuader’s argument repertoire in support of his own position may be increased through one of several mechanisms. First, someone who seeks to persuade may engage in “anticipatory elaboration,” purposefully seeking out information he can use to support his position (Eveland, 2004). To the extent that persuasion is premeditated, persuaders may be motivated to prepare arguments and seek out information in support of their preferred candidate (Pingree, 2007). During the persuasion attempt itself, a second mechanism may also contribute to the crystallization of the persuader’s argument: The very act of defending their preferred position may itself help individuals organize their opinions and express them more clearly (Mutz, Sniderman, & Brody, 1996). Given these two potential (nonmutually exclusive) paths through which engaging in persuasion may enhance one’s own ability to argue one’s own position, I expect that individuals who report engaging in persuasion will be better able to articulate reasons for supporting their preferred candidate.

Hypothesis 1A (H1A): Engaging in a persuasion attempt will be positively related to a person’s ability to articulate arguments for his own position, even controlling for other relevant factors such as education, political interest, and political knowledge.

The target of persuasion experiences the attempt differently from the initiator. Anticipatory elaboration is less likely to occur for the “persuadee” given that he or she likely did not have prior knowledge that he or she would be targeted for persuasion. Learning through exposure, in which individuals come into contact with alternative viewpoints during the discussion itself, is a more relevant approach to understanding how the conversation might directly affect his or her argument repertoire. By coming into contact with views that are different from his or her own, a person may gain a more balanced perspective of both the weaknesses of his or her own position (or preferred candidate) and the strengths of the opposition.

Hypothesis 1B predicts that among those who already have a candidate preference, being targeted for persuasion will result in an increased awareness of the opposing side’s point of view.

Hypothesis 1B (H1B): Among individuals with a preexisting candidate preference, being targeted for persuasion (in other words, exposed to an opposing viewpoint) will be positively related to the ability to articulate arguments for the opposition.

Persuasive Conversation as a Channel of Political Information

Another proposed benefit of political talk is the spread of information (Gutmann & Thomson, 2004). Evidence for a strong relationship between political discussion and political knowledge has been found both cross-sectionally (Bennett, Flickinger, & Rhine, 2000; Eveland, 2004; Kennamer, 1990) and longitudinally (Eveland & Thomson, 2006). Social networks serve as shortcuts for less engaged individuals to come in contact with relevant political information, and this information is often conveyed via casual conversation (Huckfeldt, 2001). Persuasion attempts may be a particularly effective conduit for this political information (above and beyond “discussion” more generally) because persuaders have a strategic interest in ensuring that the facts they provide are maximally relevant to their target. As McClurg points out, “Expert discussants are useful to their peers because they add clarity to information exchanges in networks, thereby helping people to connect that information to their predispositions” (2006, p. 740). During persuasion attempts, the desire to influence may encourage persuaders to tailor their pitches to their target’s interests and background as well as provide information that they believe their target does not already possess.

A persuader who is acting strategically should target not only those who disagree with him but also the ambivalent members of his social network—in other words, those most susceptible to persuasion. The recent (and deserved) focus on disagreement in political discussion has meant that conversations between decided and undecided voters are often ignored altogether. Many studies of political conversation divide political discussion into two categories: those that take place between people who agree, and those that take place between people who disagree (e.g., Huckfeldt, 2007; Mutz, 2002; Price et al., 2002). When it comes to hot-button issues like abortion or gay marriage, this sharp distinction is both relevant and meaningful. Most people have strong opinions about these issues and can place themselves firmly on one side or another. However, this binary agree/disagree distinction cannot capture the shades of gray present during an election period, in which large numbers of people stay undecided till late in the campaign and many others are simply disengaged from politics altogether (Chaffee & Rimal, 1996; Delli Carpini & Keeter, 1996).

Understanding and measuring political discussion in the campaign context therefore requires us to recognize that most individuals’ social networks consist not only of people who agree or disagree with their political views but also those who are undecided or indifferent (Huckfeldt & Sprague,1995). From a strategic perspective, interpersonal persuasion will be especially effective on this group: Individuals who are the least politically interested are also the most likely to rely on their social networks for advice on how to vote (Beck, Dalton, Greene, & Huckfeldt, 2002). The desire to effect opinion change may thus serve as an incentive for political experts to share factual information with their less knowledgeable friends and relatives.

Hypothesis 2 (H2): Being targeted for persuasion will be positively related to knowledge of the candidates’ policy positions, even controlling for frequency of political discussion. The effect will be strongest for the least politically interested.

Data and Measures

To investigate the extent to which persuasive conversation increases awareness of rationales and campaign knowledge, I use data from both the telephone and online components of the 2008 National Annenberg Election Survey (NAES). The telephone component is a rolling cross-sectional (RCS) survey conducted daily over the course of the 2008 election (December 17, 2007, to November 3, 2008).³ The online component of the study is a representative sample recruited through random digit dialing by Knowledge Networks and reinterviewed several times over the course of the 2008 election. Although the survey was administered via the web, respondents who did not already have Internet service at home were provided with access via WebTV. This article uses data from a preelection wave (administered during the general election, between September 1 and November 3) and a postelection wave (administered after the election).

While both surveys asks respondents about whether they engaged in and were targeted for persuasion, only the online survey includes a measure of awareness of oppositional rationales, and only the telephone survey includes a measure of campaign-specific political knowledge. Thus, the analyses in this article use data from the online survey to test H1A and H1B (awareness of own and oppositional rationales) and data from the telephone survey to test H2 (political knowledge). The following section outlines the measurement of the key dependent and independent variables in the analyses. In addition, the Appendix includes a table that lists the relevant variables included in both the RCS and online surveys, as well as when they were collected.

Attempted to persuade

In the telephone survey, respondents were asked during the past 3 months of the campaign, “In the past week, have you talked to any people and tried to show them why they should vote for or against one of the presidential candidates?,” with 25% responding that they had. In the online survey, respondents were asked after the election whether they had attempted to persuade over the course of the campaign. Slightly over half of the respondents (52.2%) reported that they tried to show someone how to vote over the course of the campaign. The difference between the two instruments is likely attributable to the differences in question wording: “in the past week” versus “over the course of the campaign”.⁴

Targeted for persuasion

During the last 27 days of the campaign, participants in the telephone survey were asked, “In the past week, has anyone you know personally tried to show you why you should vote for or against a candidate,” with 30.1% responding affirmatively. In the online survey, the question wording and timing was slightly different. After the election, respondents were asked whether over the course of the campaign, anyone they knew personally tried to show them why they should vote for or against a candidate. The majority of respondents, 71.7%, reported that they had been the target of a persuasion attempt.⁵ Again, the difference between the telephone and online components is likely attributable to the question wording of “in the past week” versus “over the course of the campaign.”⁶

Articulation of own and oppositional rationales

The ability to articulate rationales for a given position is frequently examined in the context of specific issues. For example, Mutz (2002) asks respondents to give reasons that people might hold views on both sides of the issue in regards to affirmative action and the welfare system, while Price et al. (2002) elicit opinions about Medicare. However, because this article deals with political talk in the context of an election, I focus on opinions about candidates rather than issues. The measure was constructed using the responses from an open-ended likes and dislikes battery administered in the 2 months prior to Election Day. A series of four open-ended questions asked respondents to list what they liked and disliked about Barack Obama and John McCain. Responses were coded by counting the number of unique likes and dislikes in each category.⁷ The number of rationales for supporting one’s preferred candidate was computed by summing each respondent’s likes of his or her preferred candidate and dislikes of the opposing candidate (respondents who did not indicate a preferred candidate were excluded). Number of rationales in support of the opposing candidate was computed by summing the dislikes of each respondent’s preferred candidate and the likes of the opposition.⁸ Intercoder reliability between two coders was assessed on a subsample of 634 likes and dislikes randomly selected from all nonblank responses. The Cohen’s kappa value for the coders’ agreement above chance was .833 (p < .0001).

Table 1 shows that, unsurprisingly, people were able to provide more reasons to support their own candidate than reasons to support the opposition (p < .0001). In addition, the two measures were strongly correlated with each other, indicating a general tendency toward opinionation and political involvement. Both of these patterns are consistent with similar analyses in past work (Mutz, 2006; Price et al., 2002).

Table 1.

Rationales Provided for Own and Opposing Candidate Preference.

	Mean	Range	Correlation	t Value
Number of rationales for supporting one’s preferred candidate
Likes of own candidate + dislikes of opposition	2.570	0-6	.440*	114.712*
Number of rationales for supporting the opposing candidate
Dislikes of own candidate + likes of opposition	0.884	0-3	—	—

Note: t values are based on paired t tests comparing the number of rationales given for own and oppositional views.

p < .0001.

Knowledge of candidate policy positions

Candidate policy position knowledge is an additive index of 10 questions on the telephone survey about the 2 candidates’ positions on specific issues in the 2008 election (M = 5.82, SD = 2.14). Respondents answered questions about whether particular policies (e.g., closing Guantanamo or increasing federal funding for stem cell research) were supported by Barack Obama, John McCain, neither, or both (see Appendix for the complete list of questions).

Results

Cross-Cutting Persuasion and Awareness of Oppositional Rationales

Table 2 examines the relationship between persuasive conversation and the number of rationales the respondent gave in support of his or her own candidate and in support of the opposing candidate. The first column examines H1A, which predicts that engaging in persuasion will be positively related to the number of reasons a person is able to provide in support of his or her preferred candidate. The second examines H1B, which predicts that being targeted for persuasion will be positively associated with the ability to provide arguments in support of the opposing candidate. Both models include a control for each individual’s underlying tendency to elaborate: Entering “reasons for supporting one’s own candidate” as a covariate in the model predicting reasons for supporting the opposition (and vice versa) makes it possible to isolate the residual variation and determine which factors uniquely affect the number of reasons given.

Table 2.

Persuasive Conversation as a Determinant of Ability to Articulate Rationales for Own and Oppositional Views, With Controls for Vote Choice Certainty and Standardized Coefficients.

	Reasons Given in Support of
	Own Position			Opposing Position
	B	β	SE	B	β	SE
Constant	−1.107***		0.105	0.428***		0.091
Socioeconomic status (SES)/demographics
Age	0.005***	0.036	0.001	−0.005***	−0.065	0.001
Gender	0.321***	0.087	0.03	−0.086***	−0.043	0.024
Race (black)	−0.242***	−0.038	0.055	−0.172***	−0.053	0.012
Race (Other)	0.002	0	0.075	0.019	0.004	0.04
Education	0.152***	0.09	0.015	0.085***	0.091	0.062
Household income	0.047*	0.02	0.055	0.006	0.005	0.016
Political involvement
Republican	−0.096	−0.025	0.071	0.118	0.057	0.052
Democrat	−0.329***	−0.088	0.071	0.124*	0.061	0.043
Strength of party identification	0.055	0.025	0.038	−0.086**	−0.07	0.015
Strength of ideology	0.142***	0.044	0.028	0.013	0.007	0.022
Political interest	0.230***	0.107	0.021	0.118***	0.095	0.016
Internet access	0.113*	0.021	0.047	0.035	0.012	0.037
News exposure	0.083***	0.06	0.012	0.050***	0.066	0.009
Rationales articulated
Reasons given for own position	—		—	0.191***	0.346	0.007
Reasons given for opposing position	0.689***	0.387	0.016	—		—
Strength of vote decision
Candidate affect	0.111***	0.162	0.006	−0.089***	−0.229	0.005
Vote certainty	0.436***	0.106	0.037	−0.270***	−0.088	0.036
Persuasive conversation+
Attempted to persuade	0.447***	0.122	0.033	—		—
Targeted for persuasion	—		—	0.148***	0.065	0.027
R ²	0.326					0.279
N	10,397					5,483

p < .05. **p < .01. ***p < .0001.

The models also include a number of controls that fall into two basic categories: demographic variables associated with political involvement and political knowledge (Delli Carpini & Keeter, 1996; Verba, Schlozman, & Brady, 1995) and antecedent measures of political involvement associated with argument repertoire. The first category includes age, race, gender, education, and income. The second group includes strength of partisanship and ideology, political interest, and news consumption. These variables were selected because they are relatively stable measures of political awareness.⁹ In addition, a control is included for web access, since those with web access may be exposed to a wider range of opinions and information.

Another explanation for an observed association between engaging in persuasion and ability to articulate rationales is that a third variable, strength of candidate support, causes both outcomes. Partisanship does not entirely capture strength of support: A strong partisan could be only lukewarm about his or her party’s candidate or an independent could be a strong supporter of a candidate. Therefore, the models also include two separate controls that assess the strength of an individual’s vote decision: candidate affect and vote certainty. Candidate affect is the absolute value of the difference between the thermometer ratings for each candidate,¹⁰ scaled from 0 to 10. Vote certainty is a dichotomous indicator of how certain the respondent is that he or she will vote for his or her preferred candidate.¹¹

Table 2 shows that several variables are associated with the ability to articulate rationales for both positions. These include education, political interest, and consumption of political news. However, not all controls contribute similarly to both outcomes: Partisans are better at arguing for their own position and worse at seeing the other side. The “strength of vote decision” variables are also differently associated with each dependent variable. Those who are very sure of their vote are both better able to defend their own candidate and less able to see the positive aspects of their opponent. People who rate the candidates extremely far apart on the feeling thermometers state fewer reasons for supporting the opposition but more reasons in support of their own position.

The first column in Table 2 examines the hypothesis that individuals who engage in persuasion attempts will be able to articulate more reasons for supporting their preferred candidate. As predicted by H1A, engaging in persuasion is positively and significantly associated with the number of arguments a person provides in support of his or her preferred candidate, even controlling for other measures of political involvement and vote certainty (B = .447, p < .0001). The relationship is strong: Attempting to persuade has about the same effect on the number of rationales articulated as moving from “not very interested” to “very interested” in politics. Adding “attempted to persuade” to the model increases the variance explained from 31.4% to 32.5%, a significant change (F = 180.87, p < .0001).

The second column examines H1B, which predicts that being targeted for persuasion will be positively associated with the ability to provide arguments in support of the opposing candidate. The results suggest that individuals who report being the subject of persuasion are able to name significantly more rationales for the opposition than those who were not (B = .148, p < .001). The effect is about the same size as 3 additional days of exposure to campaign news per week, and nearly as large as moving from “high school graduate” to “4-year college degree” in years of education. Adding “targeted for persuasion” to the model increases the variance explained from 27.4% to 27.7%, a significant change (F = 29.11, p < .0001). These results provide evidence that the exposure to an alternative point of view that comes with being targeted for persuasion may increase a person’s ability to articulate reasons why someone might support the opposing candidate.

A plausible alternative explanation for an association between being targeted for persuasion and the ability to articulate oppositional rationales is that “persuadees” are targeted by their friends and neighbors precisely because they are more open to the opposition’s arguments and, thus, aware of reasons to support both sides. This analysis addresses this potential reverse causality in several ways. First, the analysis is restricted to individuals who report making their vote decision prior to October, thus reducing the likelihood that they were ambivalent or undecided voters.¹² Of course, since time of vote decision was a retrospective measure asked after the election was over, individuals may be misremembering when they decided. Partly for this reason, the same two measures of strength of decision employed in the previous model are repeated in this analysis: candidate affect and confidence in vote.

Persuasion as a Channel for Campaign Knowledge

This section tests the hypothesis that being targeted for persuasion increases knowledge of the candidates’ policy positions, particularly among those who are less politically engaged. The analyses that follow use data from the telephone survey, as a knowledge battery was not included in the online survey. While the previous section focused on identifying individuals who may have been exposed to oppositional viewpoints, this analysis is concerned with the level of expertise present in a given discussion. To what extent do persuasive conversations provide an opportunity for those who follow politics closely to share information with members of their social network who are less involved?

One obstacle to isolating the effect of persuasive conversation in particular is that there is already a well-established positive relationship between political discussion and political knowledge, and so any observed effect of persuasive conversation on knowledge may be attributable to discussion more generally and not to the persuasion attempt in particular. However, because the survey included a question about the frequency with which the respondent discussed politics, it is possible to isolate the effects of persuasion attempts by controlling for overall levels of discussion.

Table 3 presents an OLS (ordinary least squares) regression testing the hypothesis in a multivariate setting with a number of controls. An interaction term between political interest and being persuaded tests the secondary hypothesis that the posited relationship between persuasive conversation and political knowledge is strongest for those who are least politically involved.

Table 3.

Effect of Being Targeted for Persuasion on Campaign Knowledge: With Standardized Coefficients.

	Model 1: Without Control for Discussion			Model 2: With Control for Discussion
	B	β	SE	B	β	SE
Constant	1.245***	—	0.151	1.402***	—	0.151
Socioeconomic status (SES)/demographics
Age	−0.008***	−0.055	0.002	−0.006***	−0.046	0.002
Female	−0.148**	−0.034	0.047	−0.171***	−0.040	0.047
Education	0.212***	0.122	0.022	0.203***	0.117	0.022
Income	0.048***	0.051	0.011	0.042***	0.044	0.011
Race (black)	−0.015	−0.002	0.090	−0.022	−0.003	0.090
Race (Other)	−0.302***	−0.037	0.089	−0.286***	−0.035	0.089
Political involvement
Political interest	0.717***	0.251	0.039	0.608***	0.213	0.040
Republican	−0.388***	−0.082	0.058	−0.390***	−0.083	0.058
Democrat	0.496***	0.110	0.055	0.492***	0.110	0.055
Strength of ideology	−0.004	0.001	0.031	−0.014	−0.005	0.031
News exposure	0.116***	0.094	0.014	0.088***	0.072	0.015
Civic knowledge	0.552***	0.258	0.026	0.536***	0.250	0.026
Political conversation
Persuadee	0.815***	0.181	0.238	0.657**	0.146	0.237
Persuadee × Political Interest	−0.191**	−0.154	0.067	−0.164*	−0.132	0.067
Political discussion	—		—	0.092***	0.114	0.010
R ²	0.284			0.294
N	6,330			6,330

p < .05. **p < .01. ***p < .0001.

In Model 1 (in the left-hand column), the highly significant positive coefficient for being targeted for persuasion (B = .815, p < .001) supports the hypothesis that persuasive conversations are uniquely effective at transmitting political knowledge. Being targeted for persuasion has a stronger effect on knowledge than moving from 1 to 7 days of exposure to campaign news. Moreover, a significant interaction term with political interest (B = –.191, p < .01) suggests that the effect is strongest for those who are least politically involved. As a persuadee’s political interest increases, the knowledge-transmitting effect of persuasion attempts decreases. This pattern makes intuitive sense—the less closely a person follows politics, the more likely it is that being drawn into a political discussion will provide him or her with new factual information. Adding these two terms significantly increases the variance explained by the model, from 28.2% to 28.4% (F = 8.779, p < .001).

Model 2 introduces a control for political discussion in order to test an alternative explanation for the observed relationship: that it is not persuasive conversation per se that increases knowledge but any political discussion.¹³ Discussion is a significant predictor of issue knowledge (B = .092, p < .001), but even with this control, being targeted for persuasion remains a significant predictor of campaign knowledge: It has the same effect on knowledge as moving from 1 to 7 days of political discussion in the past week. Persuasive conversation has an impact on knowledge above and beyond that which past research suggests is granted by political discussion more generally.

As with all cross-sectional analyses, establishing causal direction is a potential problem: Does being targeted for persuasion increase knowledge, or are more knowledgeable people more likely to be targeted for persuasion? This concern is somewhat mitigated by the fact that being persuaded is something that is done to an individual rather than something that himself or herself initiates. In addition, several other variables also address this alternate explanation by controlling for other factors that might contribute to campaign knowledge. Most importantly, both models include a measure of civic knowledge.¹⁴ As we might expect, civic knowledge is a strong predictor of candidate issue position knowledge (B = .552, p < .001). Other controls include political interest, partisanship, strength of ideology, and exposure to campaign news. Democrats tended to perform better on the candidate policy position battery than did Republicans.¹⁵ Even with these controls, H2 receives substantial support: Individuals who are targeted for persuasion are more knowledgeable about where the candidates stand on the issues of the day, and the effect is magnified for those who are least interested in politics.

Discussion

The results presented in this article suggest that persuasive conversation helps to fulfill three normatively important functions: increasing individuals’ ability to articulate their own position, their awareness of the opposing position, and their ability to clarify the candidates’ positions on the issues. Because the data are cross-sectional, the analyses present evidence of association rather than causation. There are two parallel alternative causal explanations for the observed relationship between persuasive conversation and ability to articulate rationales. First, affinity for a candidate might lead a person both to engage in persuasion and to have a larger repertoire of arguments in support of one’s choice. Second, ambivalence about one’s preferred candidate might lead to a person to be targeted for persuasion by members of his or her social network and to be more aware of reasons to support the opposition. Both analyses include a number of controls to address these alternative explanations, including several measures of political involvement as well as of decisiveness and candidate affinity. Even with these controls, the relationship between persuasive conversation and the dependent variables of interest is strong and significant. In addition, it is important to note that the analyses included only those who expressed a candidate preference at the time of the survey, suggesting that even those who have strong preexisting views are capable of acquiring and retaining information from the other side.

Personality traits, which are not measured in either survey, could also contribute connection between persuasive conversation and ability to articulate rationales. For instance, a personality trait such as “openness to new ideas” could lead someone both to engage in persuasive conversation and to be more aware of the opposition. However, it is likely that such a personality trait would also lead people to consume more news and be more politically interested, and as such, much of the variance produced by such a personality trait may be captured by those variables. Future investigations of interpersonal persuasion should attempt to better understand the interplay between personality and political discussion.

Persuasive conversation can be conceptualized in two different ways: as a single incident or as a pattern of conversational behavior. In the online survey, respondents are asked whether they have persuaded or been targeted for persuasion over the course of the campaign. In the telephone survey, they are asked about the past week. While respondents could be responding based on their memory of a particular incident, it is likely that their answer reflects a larger pattern of social behavior. Certain people are embedded in social contexts in which they have the opportunity to persuade and/or are targeted for persuasion, and so the observed effects of persuasive conversation may not be the result of one individual instance but rather the effect of multiple conversations that take place over the course of the campaign.¹⁶ Future work in this area would benefit from an additional question asking respondents how often they persuade and are persuaded. This information would provide insight into whether the observed outcomes are the cumulative effect of a sustained series of persuasive conversations or whether they occur after just a few interactions. Probing for further details could help to determine whether there is a social desirability bias in responses—to what extent is persuasion overreported, especially those who tend to be confident in their positions?

Issues with establishing causality could be usefully addressed using experimental or panel designs. An experiment would have the added benefit of clarifying causal mechanisms. For example, an experiment designed to test the anticipatory elaboration hypothesis could warn participants that they would soon be asked to engage in a persuasion attempts, then allow a randomly selected group to access relevant campaign information. If this group was better able to argue for its position than a group not allowed access, it would suggest that anticipatory elaboration contributes to persuasive conversation’s role in increasing argument repertoire. A panel design, on the other hand, would offer less insight into mechanisms but more external validity, especially regarding the frequency and context of persuasive conversation.

Although deliberative conversation has been characterized as an essential component of a functioning democracy, such talk may be jeopardized by the balkanization and polarization of American political life. Individuals are more and more likely to self-select into social networks comprised of similar-minded others (McPherson et al., 2001), lessening their potential exposure to diverse viewpoints. At the same time, the proliferation of niche media outlets allows the politically disengaged to avoid political information altogether (Prior, 2008). The increasing ability of individuals to select the information and opinions to which they are exposed highlights the importance of understanding the conditions under which people do come in contact with people who hold views different from their own and information that they might not otherwise seek out.

One such opportunity is interpersonal persuasion attempts, a form of political conversation that has become more and more common over the past decade: Indeed, the 2002 and 2004 midterm elections (relatively low-salience political events) saw more people engaging in persuasion than did the 1992 or 1996 Presidential elections. Campaigns are contributing to this trend by allocating more resources to determining which of their supporters are embedded in networks that contain “persuadable” voters and identifying opinion leaders who are then given special encouragement to persuade (Schaller, 2004). Technology may also help citizens identify potential subjects of persuasion among their friends, family, and neighbors, as social networking sites allow people to keep in contact with (and often learn the political opinions of) both close friends and distant acquaintances.

Persuasion attempts include both cross-cutting discussion between individuals who support different candidates and conversations in which partisans target those who are undecided or indifferent. People who reported engaging in the first type of conversation are better able to both articulate their own position and outline reasons to support the opposition candidate. This finding corresponds with existing evidence about how political talk affects awareness of oppositional views: The benefits accrue not from conversation in general but from conversation with those who have different views. For many citizens, the desire to persuade may push them to disregard the social norm of avoiding controversial political topics. This tendency may be bad for maintaining harmony at Thanksgiving dinners, but good for democracy in general: People who are willing to step outside their comfort zone of political conversation make more informed citizens. In a democracy, the ability to understand and articulate the views of the opposition has implications for legitimacy, especially in an election context. Supporters of the loser who are able to see good points in the victor may be more willing to recognize his legitimacy (Manin, 1987). In addition, evidence suggests that exposure to dissonant political views may have an indirect effect of increasing political tolerance (Mutz, 2008).

Persuasive conversation can also be an effective channel for communicating political information. Individuals who report being targeted for persuasion are significantly more knowledgeable about candidate issue positions than those who are not, and the effect is strongest for the least politically interested. Although the positive relationship between political talk and political knowledge is well demonstrated (Bennett, Flickinger, & Rhine, 2000; Eveland, 2004), the data suggest that persuasion has an additional effect well beyond that conferred by political discussion more generally. One explanation for this difference may be that the information conveyed during a persuasion attempt is chosen specifically for its relevance to the target: When a person attempts to persuade, he or she marshals the information he or she believes will be most convincing given the specific concerns of his or her discussion partner. When talking to someone without insurance, this may be the details of John McCain’s health care tax credit. For someone concerned about the war, it may be a reminder that Barack Obama voted against invading Iraq. Thus, while “discussion” could encompass topics ranging from talk about the candidates’ hairstyles to exchanging opinions about a vice-presidential choice, persuasion attempts may be more likely to contain relevant and pertinent campaign information. The desire to win new supporters provides additional incentive for political experts to contextualize political information for less informed members of their social networks.

This article presents evidence that persuasive conversation can have meaningful effects on both the persuader and his target. Persuasive conversation can be a channel through which political information is transmitted from the more politically active to the less politically active, as well as an opportunity for exposure to opposing viewpoints: both processes that are crucial to an informed and civil democracy.

Footnotes

Appendix 1

Appendix 2

Variables by Survey and Time Collected

Variable	Rolling Cross-Sectional Survey (RCS; Telephone): Time Collected	Panel (Online): Time Collected
Age	Sept 1-Nov 3	Sept 1-Nov 3
Gender	Sept 1-Nov 3	Sept 1-Nov 3
Race	Sept 1-Nov 3	Sept 1-Nov 3
Education	Sept 1-Nov 3	Sept 1-Nov 3
Strength of party identification	Sept 1-Nov 3	Sept 1-Nov 3
Strength of ideology	Sept 1-Nov 3	Sept 1-Nov 3
Political interest	Sept 1-Nov 3	Sept 1-Nov 3
Political participation	Sept 1-Nov 3	Sept 1-Nov 3
Internet access	Sept 1-Nov 3	Sept 1-Nov 3
News exposure	Sept 1-Nov 3	Sept 1-Nov 3
Knowledge of candidate policy positions	Sept 1-Nov 3	Not collected
Civic knowledge	Sept 1-Nov 3	Not collected
Reasons given for own position	Not collected	Sept 1-Nov 3
Reasons given for opposing position	Not collected	Sept 1-Nov 3
Candidate affect	Sept 1-Nov 3	Sept 1-Nov 3
Vote certainty	Sept 1-Nov 3	Sept 1-Nov 3
Engaged in persuasion over the course of the campaign	Not collected	Postelection
Targeted for persuasion over the course of the campaign	Not collected	Postelection
Attempted to persuade in the past week	Sept 1-Nov 3	Not collected
Targeted for persuasion in the past week	Oct 4-Nov 3	Not collected
Time of vote decision	Not collected	Postelection
Days of political discussion	Sept 1-Nov 3	Not collected

Acknowledgements

The author is grateful to Kristi Andersen, Kjerstin Thorson, and to the editor and anonymous reviewers for their many helpful comments, as well as to the Annenberg Public Policy Center for collecting and making available the data used in this study.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Emily Thorson is a doctoral candidate at the Annenberg School for Communication and in the Department of Political Science at the University of Pennsylvania, Philadelphia.

References

Abramowitz

Saunders

(1998). Ideological realignment in the U.S. electorate. The Journal of Politics, 60, 634-652.

Beck

Dalton

Greene

Huckfeldt

(2002). The social calculus of voting: Interpersonal, media, and organizational influences on Presidential choices. American Political Science Review, 96, 57-73.

Bennett

Flickinger

Rhine

(2000). Political talk over here, over there, over time. British Journal of Political Science, 30, 99-119.

Berelson

Lazarsfeld

McPhee

(1954). Voting. Chicago, IL: University of Chicago Press.

Chaffee

Rimal

R. N.

(1996). Time of vote decision and openness to persuasion. In Mutz

Sniderman

Brody

R. A.

(Eds.), Political persuasion and attitude change (pp. 267-291). Ann Arbor: University of Michigan Press.

Conover

P. J.

Searing

D. D.

Crewe

I. M.

(2002). The deliberative potential of political discussion. British Journal of Political Science, 32, 21-62.

Delli Carpini

M. X.

Keeter

(1996). What Americans know about politics and why it matters. New Haven: Yale University Press.

Delli Carpini

M. X.

Cook

Jacobs

(2004). Public deliberation, discursive participation, and citizen engagement: A review of the empirical literature. Annual Review of Political Science, 7, 315-344.

Eveland

(2004). The effect of political discussion in producing informed citizens: The roles of information, motivation, and elaboration. Political Communication, 21, 177-193.

10.

Eveland

Thomson

(2006). Is it talking, thinking, or both? A lagged dependent variable model of discussion effects on political knowledge. Journal of Communication, 56, 523-542.

11.

Feldman

Price

(2008). Confusion or Enlightenment? How exposure to disagreement moderates the effects of political discussion and media use on candidate knowledge. Communication Research, 35, 61-87.

12.

Fishkin

J. S.

(1991). Democracy and deliberation: New directions for democratic reform. New Haven, CT: Yale University Press.

13.

Gutmann

Thompson

(2004). Why deliberative democracy? Princeton, NJ: Princeton University Press

14.

Huckfeldt

(2001). The social communication of political expertise. American Journal of Political Science, 45, 425-438.

15.

Huckfeldt

(2007). Unanimity, discord, and the communication of public opinion. American Journal of Political Science, 51, 978-995.

16.

Huckfeldt

Sprague

(1995). Citizens, politics, and social communication. Cambridge, UK: Cambridge University Press.

17.

Katz

Lazarsfeld

P. F.

(1955). Personal influence: The part played by people in the flow of mass communications. Glencoe, IL: The Free Press.

18.

Kennamer

J. D.

(1990). Political discussion and cognition: A 1988 look. Journalism and Mass Communication Quarterly, 67, 248-352.

19.

Kim

Wyatt

Katz

(1999). News, talk, opinion, participation: The part played by conversation in deliberative democracy. Political Communication, 16, 361-385.

20.

Manin

(1987). On legitimacy and political deliberation. Political Theory, 15, 338-368.

21.

McClurg

S. D.

(2004). Indirect mobilization: The social consequences of party contacts in an election campaign. American Politics Research, 32, 406-443.

22.

McClurg

S. D.

(2006). The electoral relevance of political talk: Examining disagreement and expertise effects in social networks on political participation. American Journal of Political Science, 50(3), 737-754.

23.

McPherson

Smith-Lovin

Cook

(2001). Birds of a feather: Homophily in social networks. Annual Review of Sociology, 27, 414-444.

24.

Moy

Gastil

(2006). Discussion networks, media use, and deliberative conversation. Political Communication, 23, 443-460.

25.

Mutz

(2002). Cross-cutting social networks: Testing democratic theory in practice. American Political Science Review, 96, 111-126.

26.

Mutz

(2006). Hearing the other side. New York, NY: Cambridge University Press.

27.

Mutz

(2008). Is deliberative democracy a falsifiable theory? Annual Review of Political Science, 11, 521-538.

28.

Mutz

Sniderman

Brody

(Eds.). (1996). Political persuasion and attitude change. Ann Arbor: University of Michigan Press.

29.

Panagopoulos

Francia

(2009). Grassroots mobilization in the 2008 Presidential Election. Journal of Political Marketing, 8, 315-333.

30.

Pingree

R. J.

(2007). How messages affect their senders: A more general model of message effects and implications for deliberation. Communication Theory, 17, 439-461.

31.

Price

Cappella

Nir

(2002). Does disagreement contribute to more deliberative opinion? Political Communication, 19, 95-112.

32.

Prior

(2008). Post-broadcast democracy. New York, NY: Cambridge University Press.

33.

Schaller

(2004). New math. The American Prospect. Retrieved from http://www.prospect.org/cs/articles?article=new_math

34.

Thompson

(2008). Deliberative democratic theory and empirical political science. Annual Review of Political Science, 11, 497-520.

35.

Verba

Schlozman

Brady

(1995). Voice and equality. Cambridge, MA: Harvard University Press.