Does Civility Matter in the Blogosphere? Examining the

[admin]

Contingent condition: counter-attitudinal communication

As we discussed before, detrimental effects of uncivil attack can be moderated by the target of attacking message. In this sense, message congruence might be an important contingent factor for determining characteristics and level of incivility effects. That is, message dissonance from one’s own position may moderate effects of incivility on receivers’ negative reactions. In fact, prior research shows that exposure to dissonant views may produce detrimental effects on citizens’ socio-relational attitudes. For example, research on resistance to persuasion shows that people tend to regard counter-attitudinal information as a threat to their personal identity, and thus, tend to show negative reactions to the information, especially when they have strong commitment on their attitudes (Eagly & Chaiken, 1993). Research on thought-induced polarization also shows that merely thinking about an attitude causes that attitude to polarize or become more extreme (see Tesser, 1978; Tesser, Martin, & Mendolia, 1995). These studies suggest that exposure to conflicting views may produce negative attitudes toward those who have conflicting views and intolerance toward the oppositional views because counter attitudinal communication may enhance feeling of threat and induce more extreme attitudes.

Given that source incredibility and low message quality generally decrease message persuasion (see Eagly & Chaiken, 1993) and that uncivil expression tend to increase source incredibility and negative evaluations of message (Ng & Detenber, 2005), uncivil attacking in counter-attitudinal messages might be a optimal setting producing resistance to persuasion including negative reactions to persuasive communication. In fact, some studies on campaign ads show that negative campaign advertising may also produce unintended effects, such as backlash against its sponsoring candidate. Garramone (1984) dubbed this unintended effect “boomerang effect.” According to him, “negative political advertising may achieve its intended effects, but it may also produce boomerang effects. A strong attack on a candidate, if perceived by the audience as untruthful, undocumented, or in any way unjustified, may create more negative feelings toward the sponsor, rather than the target (p. 251).”

[admin]

Hypotheses

As we discussed before, detrimental effects of uncivil attack may be confined to when the attack is directed to position the receiver advocates. Along this line, research on the role of attitude strength in counter-attitudinal communication has shown that individuals with strong commitment on an attitude tend to react negatively to persuasion regarding the attitude (Eagly & Chaiken, 1993). Thus, we expected that incivility effects would differ between ideologically congruent and incongruent blogger’s commentary. As research on resistance to persuasion suggest, we expected that counter-attitudinal argumentations with uncivil tone may be seen as a threat to the self because one’s opinions have been incorporated in one’s self concept, resulting in greater negative emotional reactions, less open-minded, less complex reasoning and competitive or even retaliatory behavior (De Dreu & van Knippenberg, 2005). Thus, we set the following four hypotheses to interaction effects between incivility and ideological congruence on readers’ emotions and attitudes:

H1: Exposure to uncivil commentary of blogger would significantly increase negative emotions than exposure to civil commentary among participants in ideological incongruence condition, while this difference would not be found among participants exposed to ideologically congruent commentary.

H2: Exposure to uncivil commentary of blogger would significantly decrease participants’ open-mindedness than exposure to civil commentary among those in ideological incongruence condition, while this difference would not be found among those exposed to ideologically congruent commentary.

H3: Exposure to uncivil commentary of blogger would significantly increase participants’ attitude certainty than exposure to civil commentary among those in ideological incongruence condition, while this difference would not be found among those exposed to ideologically congruent commentary.

H4. Exposure to uncivil commentary of blogger would significantly decrease participants’ willingness to talk with the other side than exposure to civil commentary among those in ideological incongruence condition, while this difference would not be found among those exposed to ideologically congruent commentary.

A crucial indicator of whether these negative reactions to persuasive message have occurred is the immediate effect of such exposure on one’s affective reactions to the message. Strong negative emotions resulting from exposure to uncivil counter-attitudinal communication indicate an ego-defensive response, instead of a deliberative or open-minded response. In this sense, we expected immediate negative emotional reactions to uncivil and ideologically incongruent blogger’s commentary is a key mediator to influence more general ego-defensive reactions such as open-mindedness, attitude polarization, and avoidance of talking with the other side. Thus, three hypotheses were established to test meditative role of negative emotions in interaction effects between incivility and ideological incongruence:

H5: Negative emotions would mediate the interaction effects between incivility and ideological incongruence on open-mindedness.

H6: Negative emotions would mediate the interaction effects between incivility and ideological incongruence on attitude certainty.

H7: Negative emotions would mediate the interaction effects between incivility and ideological incongruence on willingness to talk with the other side.

[admin]

Methods

Our hypotheses were tested in an experiment embedded in a web survey in which participants viewed a fictitious news story about global climate change policy accompanied by commentary from a fictitious political blogger. Participants were undergraduate students at a large Midwestern University. Students received extra credit for their participation. The study (N=877) was fielded in April and May 2007.

[admin]

Design and procedure

The study used a 2 (civil vs. uncivil tone) x 2 (ideological congruence) between-subjects design, where civility of tone was manipulated and ideological congruence was measured based on whether or not participants viewed messages that “matched” their political ideology. Respondents read a news story about global climate change policy, which was written to emulate journalistic practice by providing a balanced summary of two positions on climate change policy. The article described both a liberal policy position for addressing climate change, endorsing mandatory caps on emissions, as well as a contrasting conservative position, supporting voluntary emissions reductions and technological innovation by industry (see Appendix 1 for full text of the news story). The news story was attributed to the Associated Press. The content of the news story remained consistent across all experimental conditions.

In order to replicate conditions under which mainstream news content would be reproduced in full, but also co-located with opinionated information content, we embedded the policy news story in a post on a fictitious web blog authored by a blogger named “Curt.” The blog commentary was written to critique either the conservative or liberal policy positions as represented in the news story. Therefore, the ideological incongruence condition was created by comparing participants’ reported political party identifications with the ideology of the blogger’s critique. Participants who reported Democrat affiliation and viewed blogger commentary critiquing the conservative position on global warming policy were in the ideologically congruent condition, as were Republican participants who read blog critiques of liberal policy. Republicans who read critiques of the conservative position and Democrats who viewed critiques of the liberal position were considered to be in the ideologically incongruent condition. Participants who reported either “Independent” affiliation or affiliation with a third party were excluded from the analysis.

To produce the tone conditions, we manipulated whether the blogger’s commentary included civil or uncivil references to the other side of the debate (Brooks & Geer, 2007). In the civil condition, the critique of the news story maintained a respectful tone (i.e., “Democrats often criticize Bush for renouncing the Kyoto Protocol, but they fail to recognize that most of the signing nations have failed to live up to the promises of the accord.”). In the uncivil condition, while the argument remained the same, the blogger used derogatory terms and insulting language when referring to opponents of the blogger’s position (i.e., “Democrat whiners often criticize Bush for renouncing the Kyoto Protocol, but they can't seem to get it through their thick skulls that most of the signing nations have failed to live up to the misguided promises of the accord.”).

[admin]

Measurement

Negative emotions. Negative emotions were measured with respondents’ indication of whether the blogger’s commentary made them feel the following five negative emotions: anger, disgust, contempt, frustration, and irritation (KR-20 =.59, M=1.77, SD=1.35).

Open-mindedness. Respondents were asked to indicate on an 11-point scale (0=strongly disagree, 10=strongly agree) their level of agreement with each of the three statements, asking how reading of blogger’s commentary affected their thoughts: (1) “I felt more open to the arguments on both sides of the issue;” (2) “I learned new things regarding this issue,” and (3) “I changed or modified my opinions about this issue.” These items were averaged into an index (M=3.87, SD=1.90, α =.72).

Attitude certainty. Attitude polarization was constructed by asking respondents’ level of agreement on the following two statements: after reading the blogger’s commentary, (1) “I felt my opinions on this issue became stronger” and “I felt more confident in my own opinion.” Both items were measured on a zero (strongly disagree) to ten (strongly agree) scale (M=5.43, SD=2.04, Inter-item correlation =.56).

Willingness to talk with the other side.. Respondents were asked to indicate their feeling about discussion with people who have opposing views on the issue by asking their level of agreement with the following two statements: (1) “I would enjoy interaction with these people” and (2) “I would find it difficult to talk with these people on the issue” (reverse coded). Both items were measured on a zero (strongly disagree) to ten (strongly agree) scale (M=4.96, SD=2.09, Inter-item correlation =.43).

Control variables. The experimental factor of blog message structure (global vs. interspersed structure) was included as a control variable in order to exclude its potential effects on the dependent variables of this study. In addition to controlling for this experimental factor, we controlled respondents’ gender (68.7% respondents were female), year in college (Median=sophomore), and ideological extremity (M=1.31, SD=.08). ). To construct ideological extremity, we first created ideological conservatism by taking the mean scores of two items (M=2.23, SD=1.33, inter-item correlation=.66) asking about respondents’ ideological orientation on a seven-point scale (0=very liberal, 6=very conservative) with regard to (1) economic issues and (2) social issues. This composite scale was recoded into a 0-3 point scale of ideological extremity by folding the scale at the midpoint (M= 1.31,

[admin]

Results

To examine interaction effects of incivility and ideological incongruence in the blogger’s commentary on emotions and attitudes, we employed dummy variable regression analysis instead of Analysis of Covariance because dummy variable regression analysis provides an easy way to determine which group mean differ from the others with the t-tests of individual dummy coefficients. In addition, the regression analysis provides an efficient way to examine mediation effects.

[admin]

Interaction effects between incivility and ideological incongruence

For testing interaction effects between incivility of blogger’s commentary and ideological incongruence, regression analyses were conducted after controlling for the experimental factor of blog message structure, gender, year in college, and ideological extremity. The results for interaction effects on the four dependent variables (i.e., negative emotions, openmindedness, attitude certainty, and willingness to talk with the other side) were reported in Model 1 of Table 1.

As Figure 1 illustrates, the regression analysis revealed that there was a significant ordinal interaction effect on negative emotions toward the blogger’s commentary (β = .15, p < .05) by showing that there was no significant difference in negative emotions between uncivil and civil blogger’s commentary condition among those who were exposed to ideologically congruent blogger’s commentary (M=1.49, SE=.09 for the incivility condition; M=1.33, SE=.09 for the civility condition), while the difference was statistically significant among those who were exposed to ideologically incongruent blogger’s commentary (M=2.42, SE=.90 for the incivility condition ; M=1.81, SE=.93 for the civility condition). The results also revealed that those who were exposed to the ideologically incongruent commentary reported significantly greater negative emotions than those who were exposed to the congruent commentary even among those who were in civil blogger’s commentary condition (β = .18, p < .001). Thus, H1 was supported.

H2 predicted that incivility and ideological incongruence would have significant interaction effect on open-mindedness. As Figure 2 illustrates, the results confirmed a significant ordinal interaction effect on open-mindedness by showing that ideological incongruence increased negative effect of incivility on open-mindedness (β = -.12, p < .05). The difference in open-mindedness between uncivil and civil commentary condition among participants who were exposed to ideologically incongruent blogger’s commentary (M=3.15, SE=.12 for the incivility condition; M=4.13, SE=.13 for the civility condition) was significantly greater than among those who were exposed to ideologically congruent blogger’s commentary (M=4.36, SE=.13 for the civility condition; M=3.90, SE=.13 for the incivility condition). The results also showed that negative effect of incivility on open-mindedness was still significant even when participants were exposed to ideologically congruent blogger’s commentary (β = -.12, p < .05).

The results supported H3, which predicted an interaction effect between incivility and ideological incongruence on attitude certainty. As Figure 3 illustrates, the results revealed a significant disordinal (crossover) interaction effect (β =.16, p < .05) such that those in the incivility condition reported greater increased attitude certainty (M=5.74, SE=.14) than did those in the civility condition (M=5.40, SE=.15) among those who were exposed to ideologically incongruent blogger’s commentary, while those who were exposed to ideologically congruent blogger’s commentary showed the opposite pattern (M=5.09, SE=.15 for the incivility condition; M=5.48, SE=.14 for the civility condition).

H4 predicted an interaction effect between incivility and ideological incongruence on willingness to talk with the other side. Again, the results confirmed our prediction by showing that the interaction term of incivility and ideological incongruence was statistically significant (β =-.14, p < .05). As Figure 4 shows, the results revealed a disordinal (crossover) interaction pattern such that among those who were exposed to ideologically incongruent blogger’s commentary, uncivil tone of commentary produced lower level of willingness to talk with the other side than did civil tone (M=4.84, SE=.15 for the incivility condition; M=5.26, SE=.15 for the civility condition), while among those who were exposed to ideologically congruent blogger’s commentary uncivil tone produced higher level of willingness to talk with the other side than did civil tone (M=5.01, SE=.15 for the incivility condition; M=4.75, SE=.15 for the civility condition). In addition, the regression analysis showed that among participants in civil tone condition, those who were exposed to ideologically incongruent blogger’s commentary did show significantly higher level of willingness to talk with the other side than those exposed to ideologically congruent blogger’s commentary (β = .12, p < .05).

[admin]

Testing the role of negative emotions in mediating interaction effects

H5 to H7 were established to test how negative emotions mediate the interaction effects of incivility and ideological incongruence on the three dependent variables of respondents’ attitudes (i.e., open-mindedness, attitude certainty, and willingness to talk with the other side). To test mediation, we used Baron and Kenny’s (1986) procedure. Baron and Kenny (1986) propose a series of necessary steps in establishing mediation: 1) the IV predicts the DV; 2) the IV predicts the mediator; and 3) when the IV and the mediator are both used to predict the DV, the mediator should be a significant predictor, and the IV decreases in its influence on the DV (partial mediation), or is removed (full mediation). Following this steps, we conducted additional analyses by entering negative emotion variable as a predictor in the prior regression analyses for testing Model 1. The results were presented in Model 2 in Table 1.

The results showed that negative emotions did fully mediate interaction effect of incivility and ideological incongruence on open-mindedness (H5) by showing that negative emotions had significant and negative effects on open-mindedness (β = -.24, p < .001) and interaction effect was reduced to no significant level when negative emotion variable was enter into the equation (β =-.12, p < .001 for Model 1; β =-.08, ns for Model 2). The results also revealed that negative emotions partially mediated interaction effects of incivility and ideological incongruence both on attitude certainty (H6) and willingness to talk with the other side (H7) such that negative emotions were significant related to attitude certainty (β =.21, p < .001) and interaction effect on dependent variable in Model 2 was smaller (β =.16, p < .05 for attitude certainty; β =-.14, , p < .05 for willingness to talk with the other side)than that in Model 1 (β =.13, p < .05 for attitude certainty; β =-.13, , p < .05 for willingness to talk with the other side). Thus, both H6 and H7 were supported.

[admin]

Discussion

Responding to recent concern about incivility in the blogosphere, the current study attempts to find connections between blogger tone, reader ideology and attitudes toward political disagreement. Our findings generally support scholars’ concerns about detrimental effects of incivility especially when bloggers attacked the views consistent with the participants’. Among individuals who were exposed to like-minded blogger’s commentary, uncivil attack did not affect negative emotional reactions to the message and their open-mindedness. However, among individuals who were exposed to the unlike-minded blogger’s commentary, there were no significant differences in negative emotions and open-mindedness. In addition, the findings show that uncivil attack produced a backlash or boomerang effects such that it reinforced certainty of unlike-minded participants’ prior issue attitude and at the same time weakened certainty of likeminded participations’ prior attitude.

The findings also showed that unlike-minded blogger’s uncivil attack decreased willingness to talk with the other side, while like-minded blogger’s uncivil attack increased willingness to talk with the other side. These findings seem to be consistent with findings from prior research on disconfirmation bias (Edwards & Smith, 1996) and the counter-attitudinal message effects on defensive reactions (Lerner & Tetlock, 1999). Further, the findings from the study provide some empirical evidence supporting the role of negative emotions in mediating joint effects of incivility and incongruence on reader’s attitudes. Negative emotions were found to fully mediate the interaction effects on open-mindedness and partially mediate the interaction effects on attitude certainty and willingness to talk with the other side.

Our findings suggest that effects of exposure to disagreement may be contingent upon the way of communicators’ expressing disagreement, as Mendelberg and Oleske (2000) pointed out, “the positive effects of deliberation rest on the ability of the opposing sides to communicate about their disagreement” (p. 186). That is, the beneficial effects of exposure to disagreement in political discourse are not solely dependent on the content of disagreement, but also on the way of communicating disagreement. In this sense, incivility in expression of disagreement in political blogs may have detrimental effects on readers’ mind such as negative evaluations of arguments (Holtgraves, 1997), hostile perception (Jessmer & Anderson, 2001), and negative expectation of deliberative process (Mutz & Reeves, 2005).

Two particularly noteworthy aspects about these findings are the role of like-minded communication and incivility in the polarization process. Our findings suggest the possibility that polarization comes not just from people congregating in like-minded groups, but also from the lack of civility. Nasty language and name-calling tend to discourage people from sorting themselves into heterogeneous groups, partly because nasty language is off-putting and partly because people who expect to receive nasty attacks want a group of like-minded folks who will back them up when that happens. It appears that counter-attitudinal messages create a negative evaluation of the other side especially when message tone is uncivil. Our findings seem to be consistent with findings from prior research on disconfirmation bias (Edwards & Smith, 1996) and the counter-attitudinal message effects on defensive reactions (Lerner & Tetlock, 1999).

Second, the manner of communicating disagreement appears to produce significant effects on message receivers’ attitudes toward political disagreement, especially when individuals were exposed to a counter-attitudinal message. Given that uncivil manners are perceived to violate norms of courtesy and reciprocity in resolving social confliction (Funk, 2001), uncivil expression from an opposing speaker could eventually initiate acrimonious public debate and worsen deeper confrontation (Mansbridge, 1983), and thus have detrimental effects on the democratic potential of such communication.

These results also have implications for broader theories of the role of discussion in democracy. Hostile attitudes toward the other side of an issue may lead people with strong attitudes to avoid future discussions with the other side. This is precisely the opposite of the “virtuous circle” of discussion, mutual understanding, and engagement discussed above. Future research should explore other aspects of message characteristics that might reduce or eliminate this hostile response, and should also directly test willingness to discuss politics with people on the other side in the future as a dependent variable.

[admin]

Limitations

This study operates under several important limitations. First, our measurement of negative emotional reaction, constructed with a series of dichotomous variables, does not allow us to pinpoint exactly where effects on emotion are occurring. Although incongruence has an impact on negative emotion generally, each distinct emotional response may occur somewhat differently than the others, which a more precise measure might be able to reveal. Additionally, there may be some emotional response that is affected by exposure to a congruent uncivil message, which is not to be found in the emotions we chose to measure. Future research that examines reactions to congruent messages specifically should use emotional constructs more suited to the congruent context.

Second, our model does not address the potential beneficial effects of exposure to uncivil, unlike-minded messages. Some research has shown that the tendency to be open to other points of view may be in conflict with the tendency to engage and participate in the political process (Mutz, 2006). Although we note some of the negative impacts of incivility in a heterogeneous political discussion, it is possible and perhaps likely that some democratically desirable outcomes occur as well.

Finally, our study makes use of an undergraduate student sample and has all the limitations that come with it. Our subjects live in a liberal, politically engaged city, and many of them are taking courses in journalism, mass communication and political science. Because of this, they may approach our stimulus more critically and with greater availability of relevant knowledge than would the general public. Our results should be seen not as necessarily generalizable to the entire population, but as a first step toward examining one of the defining phenomena of the blogosphere.