kl800.com省心范文网

Political Talking Partners and Civic Engagement An Application of Egocentric Social Network


Political Talking Partners and Civic Engagement: An Application of Egocentric Social Network Analysis in the Field of Political Science

Casey A. Klofstad Harvard University Department of Government October, 2000

This paper was prepared for presentation at the Political Psychology and Behavior Workshop, Harvard University (Cambridge, MA). A previous version of this work was prepared for presentation at the 2000 American Political Science Association Political Methodology Conference (Los Angeles, CA), and for course credit in Sociology 275 (Professor Peter Marsden, Harvard University). I am indebted to Professor Peter Marsden, Professor Sidney Verba and Tammy Frisby, all of Harvard University, for their valuable comments; however, all errors are my own. This is a work in progress; please do not cite or reference.

Abstract: The existing literature on civic engagement in the United States has focused on face-toface interactions as the key to acquiring both the desire and skills necessary to participate in civil society. However, few works have focused on the relationship between individuals' political talking networks and civic engagement on a nationwide level. One notable exception to this general oversight is Lake and Huckfeldt (1998), who present evidence suggesting a link between the frequency of interaction with political talking partners and civic engagement. Making use of public opinion data from the Cross-National Election Studies' United States study (Beck, et al., 1992), the current study extends the work of Lake and Huckfeldt in two parts. The first assesses the relationship between frequency of interaction in the political talking network and multiple indicators of civic engagement. The second addresses the relationship between attitudinal disagreement in political talking networks and civic engagement. The results suggest a positive relationship between frequency of interaction and civic engagement, and also between the amount of disagreement in the network and civic engagement. The methodological and normative implications of these relationships are discussed.

Introduction
Our results suggest that politically relevant social capital is indeed generated in personal networks; that it is a by-product of the social interactions with a citizen's discussants; and that increasing levels of politically relevant social capital enhance the likelihood that a citizen will be engaged in politics (Lake & Huckfeldt, 1998, p. 581).

During the latter half of the 20th century, civic engagement and social connectedness have become topics of regular quantitative study in the field of political science. While debates continue to occur in this literature as to the processes that underlie engagement, a general consensus has been reached on the role of interpersonal interaction. On the whole, most works conclude that face-to-face interpersonal interactions (e.g. participation in voluntary associations) are causally linked to increased levels of engagement with, and activity in, civil society (Huckfeldt & Sprague, 1991 and 1995; Huckfeldt, et al., 1995; Kenny, 1992; Putnam, 2000; Verba, et al., 1995). However, despite the importance placed on interpersonal interactions, few works have addressed the role of individuals’ personal social networks. More specifically, the literature has failed to address the link between political talking networks and civic engagement on a nationwide scale. One notable exception to this general oversight is the work of Lake and Huckfeldt (1998). The results of this study suggest that interpersonal interaction in political talking
1

networks has a significant impact on the individual's level of civic engagement. Through an extension of this work, the current study seeks to further address the role of political talking networks in civil society in two parts. The first task undertaken is a more comprehensive test of the relationship between interpersonal political interaction and civic engagement. Even under closer scrutiny, the results suggest a strong link between the amount of political talk in an individual's personal network and their level of civic engagement. In the second half of this study, we seek to answer a question posed by Lake and Huckfeldt as to the impact of political disagreement in the social network. The authors suggest that disagreement may "...offset the positive effects on engagement by increasing individual levels of political disagreement and ambivalence..." (p. 582). To the contrary, the results of the current study suggest that as disagreement in the social network increases, so does engagement. The methodological and normative implications of these findings are then discussed.

Civic Engagement and Political Talking Groups Before turning to the quantitative evidence, it is necessary to begin with a more developed explication of why interpersonal interaction bolsters civic engagement, and what possible impact disagreement in a social network could have on civic engagement.

The Role of Resources and Normative Influence Based on our understanding of the political science, social network analysis and social psychology literatures, interpersonal interactions provide two key elements which bolster civic engagement: resources and normative influence. On the resources side, scholars

2

have shown that individuals are not automatically equipped to be civically engaged. Specifically, these authors develop the notion that certain resources, such as information, civic skills and recruitment, act as prerequisites to civic engagement in the United States (Schlozman, et al., 1999; Verba, et al, 1995; Putnam, 2000). As an example, consider the case of an individual making a decision on whether or not to vote, and who to vote for. To make these decisions, the individual needs resources, such as information on the candidates, before any action can be taken. Without such resources, it is unlikely that the individual will know who to vote for, much less have an interest in going to the polls in the first place (Putnam, 2000). While resources such as information can be acquired from a variety of sources, the existing literature suggest that the costs of resource acquisition are not equal between all possible sources (e.g. Huckfeldt & Sprague, 1995). Consider again our example of the undecided voter. This individual could gain the resources needed to make political decisions from the consumption of mass media, attending political speeches, and other such activities which can entail a relatively high outlay of human and physical capital (e.g. time, attention, and in some cases, money). In contrast, our hypothetical voter could obtain such resources with less capital expense through interactions in their existing social networks. As Huckfeldt and Sprague (1995) suggest, face-to-face interactions such as these are a lower cost source of information because they allow the individual to gain information from trusted sources and on issues of specific interest to them. Otherwise stated, the interactive nature of social networks, in contrast to the one-way flow of resources from the mass media and other alternative information sources, allows for more efficient resource acquisition. Thus, while resources could possibly be acquired

3

independent of social interaction, social interaction acts as a low-cost source for resources necessary to be a competent actor in civil society. As cost-minimizing actors, individuals will "seek to obtain political information on the cheap" (Huckfeldt & Sprague, 1995, p. 14) and can therefore be expected to turn to their talking networks as a source for civic engagement resources. In addition to resource acquisition, interpersonal interactions also subject individuals to the power of normative influence (Crandall, 1988; Festinger, et al., 1950; Latane' & Wolf, 1981; Michener & DeLamater, 1999; Schachter, 1959). As Putnam (1996) argues, "...the quality of public life and the performance of social institutions... are indeed powerfully influenced by norms and networks of civic engagement" (p. 291). A method of analysis which captures many of the factors behind such influence is Latane' and Wolf's (1981) Social Impact Theory (SIT). SIT defines three major types of causal agents behind group influence: group strength (i.e. levels of solidary and task cohesion, group resources, and group prestige), group immediacy (both in space and in time) and group size1 . When these three factors are robust, members of a group can be influenced to embrace a variety of modes of behavior and sentiment, from self-destructive behavior (Crandall, 1988) to civic engagement. This group process view gives us a great amount of leverage when attempting to examine the influence of political talking partners on civic engagement. The SIT concept of immediacy has the most relevance in the assessment of this debate. As Festinger, et al. (1950) showed in their seminal examination of the group process in a Massachusetts

1

Latane' and Wolf show the universality of this model through an examination of the factors behind majority and minority influence. The authors show that the Social Impact Theory model is an accurate depiction of normative influence in both conditions. 4

Institute of Technology student housing unit, the greater the frequency of interpersonal interaction, the easier group bonds develop. In assessing the effects of this form of immediacy, the authors take the argument further in contending that "[c]ertainly it is through the small face-to-face groups that many attitudes and ideologies which affect our behavior are transmitted" (p. 163). While Festinger, et al. acknowledge the effect of the other two factors which were eventually integrated into the SIT, they single out the immediacy seen in interpersonal interactions as the key causal agent behind group formation and normative influence. Thus what these findings suggest is that because talking partners are immediate, they can be expected to have some amount of normative influence on their members.

The Impact of Disagreement in Talking Groups Clearly, a wealth of theoretical evidence exists from which to construct a definitive link between interaction through political talking networks and civic engagement. However, the existing literature offers us little in terms of theoretical evidence from which to predict the relationship between engagement and disagreement in the political network. In their conclusions, Lake and Huckfeldt present numerous ideas for further research on the issue of talking networks and civic engagement. One of their postulations centered on the notion that disagreement about politics in the social network may negatively impact the effect of talking partners on civic engagement. Some limited tangential evidence in support of this claim can be taken from social psychological studies of attitude structure and change. Works based on the theories of cognitive consistency suggest that when an individual encounters information that is not commensurate with their own beliefs, a state

5

of negative arousal develops which the individual often seeks to reduce (Fazio, et al., 1977; Fink, et al., 1983; Fiske & Taylor, 1991; Heider, 1958; Michener & DeLamater; Petty, et al., 1997; Wood, 2000). Thus, to put an end to this state of dissonance, individuals may, among other reactions, reduce the salience of the situation and viewpoints that are causing the dissonance (Fiske & Taylor; Heider; Michener & DeLamater). Otherwise stated, it is possible that individuals who interact with alters that adhere to political beliefs unlike their own may withdraw from political participation, in order to reduce their own state of cognitive dissonance. Such withdrawal could manifest itself in decreased levels of civic engagement.

In summation, through the analytic lenses of resources and normative influence, we can easily see why students of civic engagement have focused their energies on the study of interpersonal interactions. Face-to-face interactions with others allow individuals to gain relatively low-cost access to resources necessary for civic engagement, such as information. Interaction with others also makes the individual subject to the forces of normative influence. As Latane' and Wolf's Social Impact Theory suggests, the immediacy of face-to-face interactions makes the political talking network well-equipped to socialize individuals to greater levels of civic engagement. However, in contrast to these definitive conclusions, the existing literature offers us little from which to predict the relationship between engagement and network disagreement. What limited information we have suggests, albeit with a low level of certainty, that disagreement will depress the positive effects of interaction with talking partners on civic engagement.

6

Hypotheses Based on our understanding of the existing literature, and specifically the findings presented by presented by Lake and Huckfeldt, we are led to two hypotheses to be tested in the current study.

H1)

Greater frequency of interaction with political talking partners will relate to greater levels of civic engagement. As the level of disagreement in an individual's political talking network increases, the positive effect of interaction on civic engagement will be depressed.

H2)

The existing literature, along with the few studies that specifically address the relationship between network interaction and civic engagement, suggest that interaction in a social network supplies individuals with resources and normative influences that increase the likelihood of an individual becoming an active member of civil society. Thus, in Hypothesis One, we make the assumption that there will be a positive relationship between interaction and engagement in the data. In contrast, Hypothesis Two is based more on speculation than on a clear theoretical argument. However, based on the proposition of Lake and Huckfeldt, as well as the existing body of work on cognitive consistency, we postulate that there will be a negative relationship between network disagreement and civic engagement.

Data and Method Data Source The data used in testing these hypotheses come from the 1992 United States CrossNational Election Studies (CNES-US) public opinion survey (Beck, et al., 1992). The

7

universe for the CNES-US is a stratified cluster sampling of adults aged eighteen or older in the contiguous United States and the District of Columbia2 . The instrument was administered by telephone between November of 1992 and January of 1993. The total number of valid cases came out to 1318, yielding an adjusted response rate of 48 percent 3 .

Measures and Hypothesis Testing4 Hypothesis One: Assessing Frequency of Interaction In order to test the notion that there is a positive relationship between interaction with political talking partners and civic engagement, an ordinary least squares (OLS) regression model was developed for the purpose of predicting various indicators of civic engagement5 . Included in the set of civic engagement items are measures of campaign activity, campaign interest, news media usage, and frequency of voter participation during the 1988 and 1992 presidential elections. These items were selected because they reflect sentiments and activities related to engagement with, and interest in, civil society. The campaign activity measure is an ordinal six point additive index of five questions, gauging how often the respondent participated in various campaign activities (e.g.

2 3

This cluster sampling procedure was done by county. The adjusted response rate was calculated as the number of completed interviews, divided by the total number of completed interviews, partial interviews and refusals. 4 Please refer to Table A1.1 in Appendix 1 for descriptive statistics of the measures used in these tests. 5 While OLS is appropriate for this study in most cases, some of our dependent variable of interest are three or four category ordinal variables. In these cases, a model that accounts for the ordinal nature of the dependent variable would have been more appropriate. Unfortunately, we were unable to get such maximum likelihood estimates to converge in GAUSS, the program we used to generate our graphs. However, an initial examination of ordered probit results from STATA suggests no significant difference in model structure between the OLS and ordered probit specifications. 8

attending a candidate rally, giving money to a candidate, and the like) during the 1992 presidential campaign. Our campaign interest measure is a five point ordinal measure of how interested the respondent reported to be in the 1992 presidential campaign. News media usage is an ordinal twenty-five point additive index that aggregates how many days during the past week the respondent used various news media outlets during the 1992 presidential campaign. Finally, our vote frequency measure is an ordinal three point index, which measures how often the respondent voted during the 1998 and 1992 presidential elections. Those who did not vote in either election were coded zero, those who voted in one of the two were coded 1, while those who voted in both were coded at point 2. The model used to predict these various indicators of civic engagement includes measures of frequency of political discussion in the respondent's social network, total network size, the number of organizational affiliations the respondent has, income, employment status, education, race, gender, age in years, years lived in the current area of residents, strength of partisan identity, and strength of ideological identity. Our interaction frequency measure is derived from the CNES-US name generator procedure, in which the respondent was asked to name up to four people with which they discussed important matters with over the past six months. For each alter named, the respondent was asked a battery of questions about the nature and structure of the dyadic relationship. One of these questions asks the respondent to rate on a four-point scale how often they and the alter discuss politics. To create our index of political interaction frequency, the value of these scales were summed, and divided by the total number of alters mentioned by the respondent to standardize the measure for all respondents, based

9

on network size. This yields a zero to one proportion scale where greater values indicate that of the alters mentioned, the greater proportion of them discuss politics with the respondent 6 . To further control for network size, a four-point ordinal scale of the number of alters named by the respondent was also included in the model. We have also chosen to control for the number of voluntary associations the respondent affiliates with, under the assumption that such ties could relate to civic engagement and political interaction in the same manner as political talking frequency (Lake & Huckfeldt; Putnam, 2000; Verba, et al.). Therefore, to be certain that our measure of political interaction is accounting for effects independent of traditionally studied affiliation effects, we have added an index of such affiliations to our model. This measure is an ordinal thirteen point additive index, which sums up the various types of affiliations that the respondent claims to belong to (e.g. labor, service, fraternal, political activist, and others). Along with these social network and participation measures, we have also included various demographic traits in order to control for the likely possibility that traditionally marginalized members of society (such as women, the unemployed, and racial minorities) may be disadvantaged in their levels of civic engagement and in the size and structure of their social networks (Burt, 1992; Cohen & Dawson, 1993; Coleman, 1994; Coleman & Hoffer, year; Goss & Sander, 1999; Klein & D'Aunno, 1986; Marks, 1994; Marsden, 1987; Sampson, et al., 1997; Verba, et al.; Wilson, 1991; Wacquant & Wilson, 1990).

6

If the respondent had no talking partners, they were not assigned a value for the frequency proportion. It should also be noted that the CNES-US did include a fifth name generator question that read "Aside from anyone you have already mentioned, who is the person you talked with the most about the events of the recent presidential election campaign". Clearly, this question pertains to a different type of alter than the standard “important matters” name generator. Thus, in contrast to Lake and Huckfeldt, we have excluded this fifth alter from the analysis in order to keep our network measures consistent. 10

Income is coded on a six point ordinal scale, ranging from low to high. Employment status is coded as an indicator variable, where presence indicates that the respondent is employed. Education is coded on a seven point ordinal scale, ranging from "no high school diploma" to "post-graduate degree holder". Race is coded as an indicator variable, where presence indicates that the respondent is non-white. Gender is also coded as an indicator variable, where presence indicates that the respondent is a woman. Age is coded in years. Years living in the area is also coded in years as reported by the respondent. In the case that the respondent reported that they have lived in the area their entire life, their value on the index was coded as their age in years. This control is included in the model under the assumption that newcomers to a community may not yet have fully established political talking networks.

Hypothesis Two: Assessing Disagreement Our test of the notion that disagreement in the political network leads to lower levels of civic engagement was conducted in a similar manner to our test of Hypothesis One. The same model and dependent variables were utilized in this assessment, save the fact that our measure of frequency of political discussion was replaced with a measure of frequency of political disagreement. Our index of political disagreement is coded in a similar fashion to the interaction index. In this case, self-reported four-point scales of level of political disagreement between the respondent and each political discussant mentioned were summed. The measure was standardized, as was the interaction frequency measure, by dividing by the total number of alters mentioned by the

11

respondent 7 .

Results 8 Following the methodological processes outlined by King, Tomz and Wittenberg (2000), our method of assessing and presenting the results seen in these various models centers on graphs which represent the mean effect of political discussion (or disagreement) on civic engagement. Holding all other predictors at their mean value9 , the estimated values of the model coefficients (βs) are simulated 1000 times by drawing randomly from the multivariate normal distribution, given the estimated values of the βs and the variance-covariance matrix. With simulated values in hand, we are then able to calculate 1000 predicted values over the range of the independent variable of interest that we have allowed to vary. This method allows us to calculate and graph not only the mean effect line, but also 95 percent confidence intervals about this line. Thus, while this method of analysis is seemingly over-simplified, it captures the full information of the model, including uncertainty estimates, in a much more parsimonious and readily interpretable manner than a standard table of regression coefficients and significance values. In all cases, the x-axis represents the level of political discussion/disagreement, while the y-axis represents the level of a given indicator of civic engagement.
7

Again, if the respondent did not have any alters, they were not given a score on the disagreement proportion index. Also note that amount of disagreement is standardized by total network size, not the size of the political talking network. Thus, this item is a measure of political disagreement within the entire social network. 8 Please refer to Tables A1.2 and A1.3 in Appendix 1 for the coefficient, uncertainty and goodness of fit estimates for all OLS models presented. It should be noted that in all models, we have solved for missing values through the process of listwise deletion. Recent works (e.g. King, et al., 2000) have criticized this standard procedure. However, on average, we only lose 25.69% of our cases due to deletion. Thus, any concern about model missingness should be minimal.

12

The Effects of Interaction Frequency Our first task is to assess the relationship between frequency of interaction in the political talking network and civic engagement. The results in Figure 1 clearly suggest a relationship between the two. Figure 1: Mean Effect of Political Talking Interactions on Civic Engagement

As the interaction frequency proportion increases, we see significant increases in campaign activity, campaign interest, media usage, and vote frequency. For example, we

9

Otherwise stated, this allows us to ask, “For the average person, what is the mean effect of political interaction/disagreement on a measure of civic engagement?”. 13

see that as the amount of political talk increases from zero to one in an individual’s social network, their interest in the 1992 presidential campaign increases by almost two points on the one to five interest scale. Thus, it appears that we have marginal support for Hypothesis One, which states that as the frequency of political discussion increase, so does the level of civic engagement.

The Effects of Disagreement The results of our investigation of the impact of political disagreement on civic engagement are presented in Figure 2. Figure 2: Mean Effect of Political Disagreement on Civic Engagement

14

Contrary to expectation, the results of our second investigation suggest that as individuals are exposed to increasing levels of political disagreement in their social networks, their level of civic engagement may increase. Figure 2 shows clear increases in campaign interest and media usage, as well as marginal increases in campaign activity and vote frequency. In comparison to the results seen in Figure 1, these relationships are small in magnitude. Nonetheless, Figure 2 clearly shows a positive relationship between disagreement and engagement. Thus, the data suggest rejection of Hypothesis Two, which assumed that disagreement would leave individuals disaffected and lead to withdrawal from civic life. Based on the unexpected nature of these results, a further test of Hypothesis Two was conducted. Based on our understanding of the cognitive dissonance and consistency literature, it could be the case that only individuals with strong political beliefs and convictions are immune to the effects of disagreement. This proposition is based on the notion that individuals with firm beliefs are less likely to be affected by the persuasion attempts of others, and thus less likely to change their beliefs or behaviors in the face of beliefs and behaviors counter to their own (Kenny, 1994; Lavine, et al., 1997; Michener & DeLamater; Petty & Cacioppo, 1986). To test this notion, we ran our disagreementengagement models again. However, instead of setting strength of partisan identity and ideological strength at their means, we ran each engagement model twice: once under a condition of extreme ideological and partisan strength, and once under a condition of extremely weak partisan identity and ideology. The two runs were then graphed together for each civic engagement predictor, allowing us to clearly see whether or not political disagreement has a differing impact on those with stronger beliefs. The results of this

15

investigation are presented in Figure 3. Figure 3: Mean Effect of Political Disagreement on Civic Engagement, Weak and Strong Belief Conditions

Again, contrary to expectation, the strength of an individual's beliefs does not significantly alter the positive impact of disagreement. In each case, we see that for both the weak and strong belief conditions that disagreement has a positive relationship with civic engagement. The mean effect lines are almost parallel in each case, suggesting that not even the magnitude of the effect is altered under either condition. In addition, upon assessing the effect of disagreement on campaign interest and news media consumption, the overlapping confidence intervals suggest that there may not even be a difference

16

between the strong and weak belief conditions. The only difference between the two conditions appears to be the intercept of the mean impact line. What this result suggests is that as political beliefs become weaker, the individual's overall level of civic engagement also decreases, on average. This finding is not unexpected, however, considering that the existing literature has documented that those with stronger political beliefs are more likely to participate in civil society (Campbell, 1960; Campbell, et al, 1960; Campbell, 1987; Miller & Shanks, 1996).

Discussion Hypothesis One: Assessing Frequency of Interaction On the margin, the evidence from the CNES-US offers strong evidence in support of Hypothesis One. In all four indicators of civic engagement assessed, we see a positive relationship between the frequency of political interaction and civic engagement, even after factoring in other antecedents of civic engagement, such as affiliation with voluntary associations, income and education, into our model. Despite the overwhelming strength of these results, however, they are far from being the last word on the relationship between egocentric social network structure and civic engagement. It is instructive to note that the indicators of civic engagement that were assessed in the current study largely focus on voter behavior and engagement with political campaigns. While these are important aspects of engagement, they are not the only sorts of engagement. Future studies will need to focus on assessing other types of engagement that were unmeasured in the data set used in the current study (e.g. voluntarism and philanthropic work, social trust, political efficacy, and the like).

17

In addition to focusing on electoral behavior, the data used in the current study were collected around the time of the 1992 presidential election. Works in political science have establish that on the whole, the public is more engaged with politics and the electoral process during election times, and especially so during presidential elections (Campbell, 1960; Campbell, 1987; Rahn, et al., 1999). These underlying phenomena may serve to magnify not only levels of civic engagement, but also the impact of political interactions, if only for the fact that during elections, talk about politics is bound to be more prevalent than during off years. Thus, further studies should attempt to study and collect data during off-election times, to see if political interactions always have the extremely positive effect on engagement seen here. A final limitation seen in our hypothesis testing rests on a methodological issue. It has been suggested that simultaneous causation may be present in models of political participation (e.g. Kenny, 1992; Franklin, 1992). For example, partisan identity, once though to be the ultimate causal variable of political behavior (Campbell, et al.) is now seen as both a cause and effect of political behavior (Franklin). This same sort of simultaneity could be present in the relationship between political interactions and civic engagement. Greater engagement in civil society may lead individuals to more opportunities to talk about politics, and vice versa. Thus, future studies of the relationship between interaction and engagement will likely benefit from the use of instrumental variable, or two-stage, modeling techniques. Overall, the current study shows strong support for the notion that political talking partners facilitate engagement in civil society. However, further study of this relationship is necessary before comprehensive conclusions about this relationship can be made. The

18

results seen in the current study, at worst, suggest that such study is worthy of our energies, and likely to yield findings of import to the study of civic engagement and social networks alike.

Hypothesis Two: Assessing Disagreement Unlike the results in our study of Hypothesis One, we have no support in the current study for the notion that disagreement about politics in individuals' social networks leads to lower levels of civic engagement. To the contrary, the results strongly suggest that greater levels of disagreement may lead to greater levels of engagement. This relationship also appears to hold for individuals regardless of how strong their political beliefs are. The fact that the data suggest the invalidity of our Hypothesis Two is not surprising. With little in terms of existing literature to draw upon, Hypothesis Two was on uncertain theoretical ground to begin with. However, the grander question before us now is if the existing literature can explain why we see a positive relationship between disagreement and engagement. Theories of cognitive consistency do suggest that the withdrawal response predicted by Hypothesis Two may not be the only behavior individuals adhere to in the face of inconsistent information. Works in this field have also demonstrated that individuals will also redefine issues and situations, often by disparaging or discounting the views of others and by strengthening currently held convictions, to maintain consistency between their current beliefs and new information (Fink, et al.; Fiske & Taylor; Heider). Other works have shown that dissonance and persuasion can lead to attitude change and subsequently strong adherence to the once inconsistent viewpoint, in order to make the attitude change more palatable in the face of one's own previously held,

19

and now dissonant, viewpoints (Fiske & Taylor; Petty & Cacioppo). With the internal processes of human thought and cognition left as an unmeasured black box in the CNES-US data set, we are unable to discern with certainty the exact reason that disagreement relates positively to engagement. The observations we do have, however, offer initial evidence in favor of the notion that upon facing political views counter to their own, individuals may be engaging in one of these two alternative response strategies. The first response that correlates with our findings is where the individual adheres to their own views and are more active in support of them because of exposure to dissonant information. The second would be where the individual is led to make attitude changes, and thus becomes an active supporter of the new point of view. In both of these cases, the theoretical process matches our observable outcomes of increased civic engagement. Further study, and more comprehensive data collection techniques will likely shed further light on this confounding issue than the space and results of the current study provide.

Normative Implications: An Extension of the Inequality Gap? Model limitations and unanswered questions aside, it is necessary to acknowledge that the results of the current study suggest a strong link between the frequency of interaction in political talking networks, regardless of disagreement, and civic engagement. Thus, it is important to briefly examine the implications of such a relationship. One of the more insightful observations to come out of the current civic engagement literature is the concept that not all individuals have equal access to the resources and influences that equip a citizen for activity in civil society (Schlozman, et al.; Verba, et

20

al.). The literature has shows that traditionally marginalized members of society, such as the unemployed, racial minorities and the undereducated, suffer the greatest impact of unequal access to the resources and positive influences of human, physical and social capital (Schlozman, et al; Verba, et al.). Thus, we are led to the question of whether or not these same types of individuals are also disadvantaged in terms of access to political talking partners and social networks in general. While a full analysis of this question is not possible in the space provided here, an initial assessment of the CNES-US data suggests that traditionally marginalized members of society may, on average, have less access to political talking partners than more advantaged members of society. However, our initial assessment also suggests that this difference is not statistically significant10 . These initial findings leave us with two possible implications of the relationship between engagement and social network interaction. First, if traditionally marginalized members of society have lower levels of access to political talking partners, such individuals may also have less access to the resources and normative influences embedded in such networks. Thus, while interaction within political talking networks has a positive impact on the citizenry's engagement in civil society, the power of such interactions may actually widen the engagement gap between the advantaged and disadvantaged members of American society. However, if our models are correct in predicting the inequality between advantaged and disadvantaged members of society to be insignificant, greater use of the resources and influence embedded in such networks may actually serve to equalize the civic engagement social inequality gap.

10

Please refer to Appendix 2 for a full presentation and discussion of these results. 21

Conclusion The study of participation in American civil society has become a fruitful field of quantitative inquiry during the latter half of the 20th century. Conjecture and disagreements aside, the majority of these works support the notion that face-to-face interactions are the key to acquiring both the desire and skills necessary to participate actively in civil society. Despite the importance placed on interpersonal interaction in the existing literature, few works have focused on the relationship between individuals' political talking networks and civic engagement on a nationwide scale. One notable exception is the work of Lake and Huckfeldt, a study that demonstrated a link between the frequency of interaction with political talking partners and civic engagement. The current study has sought to extend the work of Lake and Huckfeldt through a more indepth assessment of the relationship between civic engagement and the frequency of interaction in the political talking network, and by assessing the impact of attitudinal disagreement in political talking networks and civic engagement. The overall results of this investigation are conclusive, but not comprehensive, evidence in favor of the notion that social networks bolster individuals' levels of civic engagement. Examination of the implications of this relationship suggests that the resources and influence embedded in social networks could reduce existing civic engagement inequalities in American society.

REFERENCES Beck, P. et al. (1992). Cross-National Election Studies: United States Study. Interuniversity Consortium for Political and Social Research, Study #6541. Burt, R.S. (1992). Structural Holes: The Social Structure of Competition. Cambridge, MA: Harvard University Press.

22

Campbell, A. (1960). Surge and Decline: A Study of Electoral Change. Public Opinion Quarterly, 24: pp. 397-418. Campbell, A. et al. (1960). The American Voter. New York, NY: John Wiley & Sons. Campbell, J.E. (1987). The Revised Theory of Surge and Decline. American Journal of Political Science, 31, pp. 965-979. Cohen, C.J. & Dawson, M.C. (1993). Neighborhood Poverty and African American Politics. American Political Science Review, 87: pp. 286-302. Coleman, J.S. (1994) Social Capital in the Creation of Human Capital. American Journal of Sociology, Supplement, 94: pp. S95-S120. Coleman, J.S. & Hoffer, T. (1987) Public and Private High Schools: The Impact of Communities. New York, NY: Basic Books. Crandall, C. (1988). Social Contagion of Binge Eating. In Halberstadt, A. & Ellyson, S. (Eds.). Social Psychology Readings: A Century of Research. New York, NY: McGraw-Hill, Inc. Fazio, R.H. et al. (1977). Dissonance and Self-Perception: An Integrative View of Each Theory's Proper Domain of Application. In Halberstadt, A. & Ellyson, S. (Eds.). Social Psychology Readings: A Century of Research. New York, NY: McGraw-Hill, Inc. Festinger, L. et al. (1950). Social Pressures in Informal Groups: A Study of Human Factors in Housing. New York, NY: Haprer & Brothers Publishers. Fink, E.L. et al. (1983). Positional Discrepancy, Psychological Discrepancy, and Attitude Change: Experimental Tests of Some Mathematical Models. Communication Monographs, 50: pp. 413-430. Fisk, S.T. & Taylor, S.E. (1991). Social Cognition. New York, NY: McGraw-Hill, Inc. Franklin, C.H. (1992). Measurement and the Dynamics of Party Identification. Political Behavior, 14(3): pp. 297-309. Goss, K. & Sander, T. (1999). Work and Social Capital: An Overview. Saguaro Seminar Paper. Cambridge, MA: JFK School of Government, Harvard University. Heider, F. (1958). The Psychology of Interpersonal Relations. John Wiley & Sons, Inc. Huckfeldt, R.R. & Sprague, J. (1991). Discussant Effects on Vote Choice: Intimacy, Structure, and Interdependence. The Journal of Politics, 53(1): pp. 122-158.

23

Huckfeldt, R.R. & Sprague, J. (1995). Citizens, Politics, and Social Communication: Information and Influence in an Election Campaign. Cambridge, England: Cambridge University Press. Huckfeldt, R.R., et al. (1995). Political Environments, Cohesive Social Groups, and the Communication of Public Opinion. American Journal of Political Science, 39(4): pp. 1025-1054. Kenny, C.B. (1992). Political Participation and Effects for the Social Environment. American Journal of Political Science, 36(1): pp. 259-267. Kenny, C.B. (1994). The Micorenvironment of Attitude Change. The Journal of Politics, 56(3): pp. 715-728. King, G. & Tomz, M., and Wittenberg, J. (2000). Making the Most of Statistical Analyses: Improving Interpretation and Presentation. American Journal of Political Science, 44(2): pp. 341-355. King, G. et al. (2000). Analyzing Incomplete Political Science Data: An Alternative Algorithm for Multiple Imputation. http://gking.harvard.edu (as on 5/9/00). Klein, K. & D'Aunno, T. (1986). Psychological Sense of Community in the Workplace. Journal of Community Psychology, 14: pp. 365-377. Lake, R. & Huckfeldt, R.R. (1998). Social Capital, Social Networks, and Political Participation. Political Psychology, 19(3): pp. 567-583. Lavine, H., et al. (1997). The Development of Interattitudinal Consistency: The SharedConsequences Model. Journal of Personality and Social Psychology, 72(4): pp. 735-749. Latane', B. & Wolf, S. (1981). The Social Impact of Majorities and Minorities. Psychological Review, 88(5): pp. 438-453. Marks, S.R. (1994). Intimacy in the Public Realm: The Case of Co-Workers. Social Forces, 72(3): pp. 843-858. Marsden, P.V. (1987). Core Discussion Networks of Americans. American Sociological Review, 52(1): pp. 122-131. Michener, H.A. & DeLamater, J. (1999). Social Psychology. New York, NY: Harcourt Brace College Publishers. Miller, W. & Shanks, J.M. (1996). The New American Voter. Cambridge, MA: Harvard University Press.

24

Petty, R.E. & Cacioppo, J.T. (1986). The Elaboration Likelihood Model of Persuasion. In Berkowitz, L. (Ed.), Advances in Experimental Social Psychology, Volume 19. New York, NY: Academic Press, Inc. Petty, R.E. et al (1997). Attitudes and Attitude Change. Annual Review of Psychology, 48: pp. 609-647. Putnam, R. (1996). Bowling Alone: America's Declining Social Capital. In Diamond, L. & Plattner, M. (Eds.), The Global Resurgence of Democracy. Baltimore, MD: The Johns Hopkins University Press. Putnam, R. (2000). Bowling Alone: The Collapse and Revival of American Community. New York, NY: Simon & Schuster. Rahn, W., et al. (1999). National Elections as Institutions for Generating Social Capital. In Skocpol, T. & Fiorina, M. (Eds.), Civic Engagement in American Democracy. Washington, DC: Brookings Institution Press. Sampson, R.J. et al. (1977). Neighborhoods and Violent Crime: A Multilevel Study of Collective Efficacy. Science, 277: pp. 918-924. Schachter, S. (1959). The Psychology of Affiliation: Experimental Studies of the Sources of Gregariousness. Stanford, CA: Stanford University Press. Schlozman, K.L. et al. (1999). Civic Participation and the Equality Problem. In Skocpol, T. & Fiorina, M. (Eds.), Civic Engagement in American Democracy. Washington, DC: Brookings Institution Press. Verba, et al. (1995). Voice and Equality: Civic Voluntarism in American Politics. Cambridge, MA: Harvard University Press. Wacquant, L.J.D. & Wilson, W.J. (1990). The Cost of Racial and Class Exclusion in the Inner City. Annals of the American Academy of Political and Social Science, 501: pp. 8-25. Wilson, W.J. (1991). Studying Inner-City Social Dislocations: The Challenge of Public Agenda Research (1990 Presidential Address). American Sociological Review, 56: pp. 1-14. Wood, W. (2000). Attitude Change: Persuasion and Social Influence. Annual Review of Psychology, 51: pp. 539-570.

25

Appendix 1: Descriptive Statistics and Civic Engagement Model Estimates Table A1.1: Descriptive Statistics
Campaign Activity Campaign Interest News Media Consumption Vote Frequency Political Talk Frequency Proportion Political Disagreement Proportion Total Network Size Number of Organizational Affiliations Income Employment Status (Employed) Education Race (Non-White) Gender (Female) Age in Years Years Lived in Area Strength of Partisan Identity Ideological Strength Valid N 1314 1315 1299 1245 1064 1064 1318 1301 Min. 0 1 0 0 0 0 0 0 Max. 5 5 24 2 1 1 4 12 Mean 0.88 3.94 14.28 1.53 0.62 0.49 2.25 2.45 Std. Dev. 1.10 1.33 4.26 0.69 0.21 0.23 1.48 1.96

1228 1318 1315 1313 1318 1313 1318 1294 1282

1 0 1 0 0 18 0 1 1

6 1 7 1 1 92 87 7 5

3.35 0.59 3.55 0.15 0.56 45.24 24.54 4.41 2.33

1.60 0.49 1.81 0.36 0.50 17.29 19.33 2.14 1.37

26

Table A1.2: Political Talk Frequency Proportion OLS Model Estimates
Campaign Activity -0.59* (0.20) 1.23* (0.17) 0.14* (0.03) 0.12* (0.02) -0.00 (0.02) 0.15* (0.08) 0.04* (0.02) -0.05 (0.10) -0.12* (0.07) -0.01* (0.00) 0.00 (0.00) 0.05* (0.02) 0.07* (0.03) 957 0.18 17.49* (12 , 944) Campaign Interest 2.05* (0.22) 1.86* (0.19) 0.10* (0.03) 0.05* (0.02) 0.04* (0.03) -0.11 (0.09) 0.03 (0.02) -0.09 (0.11) -0.10 (0.08) 0.01* (0.00) -0.00 (0.00) 0.00 (0.02) 0.05* (0.03) 958 0.16 14.78* (12 , 945) News Media Usage 7.82* (0.74) 6.90* (0.62) 0.41* (0.12) 0.22* (0.07) 0.17* (0.09) -0.84* (0.28) -0.02 (0.08) 0.43 (0.38) -0.17 (0.26) 0.02* (0.01) -0.01 (0.01) 0.01 (0.06) -0.01 (0.10) 953 0.18 16.76* (12 , 940) Vote Frequency -0.07 (0.11) 0.44* (0.09) 0.03* (0.02) 0.06* (0.01) 0.04* (0.01) 0.06 (0.04) 0.05* (0.01) -0.08 (0.06) 0.05 (0.04) 0.01* (0.00) 0.00 (0.00) 0.03* (0.01) 0.01 (0.01) 921 0.27 28.37* (12 , 908)

Constant Political Talk Frequency Proportion Total Network Size Number of Organizational Affiliations Income Employment Status (Employed) Education Race (Non-White) Gender (Female) Age in Years Years Lived in Area Strength of Partisan Identity Ideological Strength Valid N R2 F (df)

* Estimate is significant at the p <= 0.1 level.

27

Table A1.3: Political Disagreement Proportion OLS Model Estimates
Campaign Activity -0.30 (0.21) 0.50* (0.16) 0.14* (0.03) 0.12* (0.02) 0.00 (0.02) -0.15* (0.08) 0.05* (0.02) -0.05 (0.10) -0.11 (0.07) -0.01* (0.00) 0.00 (0.00) 0.07* 0.02 0.08* (0.03) 957 0.14 13.27* (12 , 944) Campaign Interest 2.57* (0.23) 0.60* (0.18) 0.10* (0.04) 0.06* (0.02) 0.06* (0.03) -0.12 (0.09) 0.05* (0.03) -0.10 (0.12) -0.09 (0.08) 0.01* (0.00) -0.00 (0.00) 0.02 (0.02) 0.07* (0.03) 958 0.08 6.90* (12 , 945) News Media Usage 9.01* (0.78) 3.69* (0.59) 0.41* (0.12) 0.27* (0.07) 0.20* (0.09) -0.91* (0.30) 0.02 (0.09) 0.41 (0.39) -0.12 (0.27) 0.03* (0.01) -0.00 (0.01) 0.08 (0.06) 0.08 (0.10) 953 0.11 9.23* (12 , 940) Vote Frequency 0.05 (0.12) 0.16* (0.09) 0.03* (0.02) 0.06* (0.01) 0.04* (0.01) 0.06 (0.04) 0.05* (0.01) -0.08 (0.06) 0.05 (0.04) 0.01* (0.00) 0.00 (0.00) 0.03* (0.01) 0.02 (0.01) 921 0.26 26.28* (12 , 908)

Constant Political Disagreement Proportion Total Network Size Number of Organizational Affiliations Income Employment Status (Employed) Education Race (Non-White) Gender (Female) Age in Years Years Lived in Area Strength of Partisan Identity Ideological Strength Valid N R2 F (df)

* Estimate is significant at the p <= 0.1 level.

28

Appendix 2: Initial Social Inequality Analyses For this analysis, two OLS models consisting of income, employment status, education, race, gender, age and years lived in the area, were used to predict total network size and frequency of political interaction11 . Using the simulation method that was used to generate our mean effect graphs, expected values of network size and political interaction were calculated for two scenarios: extreme social advantage and extreme social disadvantage. In the extreme advantage scenario, income was set at the maximum, employment status was set to employed, education was set to its maximum, race was set to white, gender was set to male, and age and years lived in the area were set at their means. The extreme disadvantage scenario was set to the exact opposite condition. From these fitted values, first differences between the advantage and disadvantage scenarios were calculated. The results of this analysis are presented in Tables A2.1 and A2.2. Table A2.1: Expected Group Size First Differences, “Extreme Advantage” vs. “Extreme Disadvantage” Social Inequality Scenarios
Extreme Advantage Scenario Expected Value (95% confidence interval) Extreme Disadvantage Scenario Expected Value (95% confidence interval) First Difference (Adv. – Disadv.) (95% confidence interval) 3.020 (2.933 , 3.108) 2.990 (2.990 , 2.990) 0.030 (-0.057 , 0.118)

Table A2.2: Expected Political Talk Proportion Score First Differences, “Extreme Advantage” vs. “Extreme Disadvantage” Social Inequality Scenarios
Extreme Advantage Scenario Expected Value (95% confidence interval) Extreme Disadvantage Scenario Expected Value (95% confidence interval) First Difference (Adv. – Disadv.) (95% confidence interval) 0.693 (0.680 , 0.706) 0.686 (0.686 , 0.686) 0.007 (-0.006 , 0.019)

11

The coefficient, uncertainty and goodness of fit estimates for these models appear in Table A2.3 of Appendix 2. 29

The results seen here suggest that there is a disparity in terms of network size and frequency of political talk between the advantaged and disadvantaged scenarios. However, the calculated confidence intervals about these first differences estimates suggest that the difference may not be statistically significant.

Table A2.3: Social Inequality and Network Structure OLS Model Estimates
Network Size Constant Income Employment Status (Employed) Education Race (Non-White) Gender (Female) Age in Years Years Lived in Area Valid N R2 F (df) * Estimate is significant at the p <= 0.1 level. 1.64* (0.19) 0.09* (0.03) 0.08 (0.09) 0.17* (0.03) -0.31* (0.11) 0.32* (0.08) -0.01* (0.00) -0.00 (0.00) 1217 0.11 21.695* (7 , 1209) Political Talk Proportion Score 0.49* (0.03) 0.01* (0.00) -0.00 (0.01) 0.01* (0.00) 0.00 (0.02) -0.00 (0.01) 0.00 (0.00) 0.00 (0.00) 999 0.03 4.846* (7 , 991)

30


赞助商链接