Follow the Leader: Partisanship in the Processing of Political Speech

Introduction

During election season, voters in the United States are overwhelmed with political speech through debates, campaign speeches, advertisements, and interviews, the prevalence of which raises the question of how voters evaluate political speech. Do people think objectively, basing their evaluation solely on the content of a message, or are they subjective, introducing partisan bias into their judgment of politicians’ ideas? A perfectly rational and fair person would judge each candidate by the merit of their arguments and the content of their speech, but in reality, most people cannot set aside their prior judgments about candidates and political parties when evaluating a new argument. I contend that party members will express higher agreement with political statements when made by a politician from their own party than when made by one of the opposing party, regardless of the statements’ contents. I will extend prior research on partisan bias from policy proposals to the more general area of political speech, and will also eliminate confounding features of prior studies that could skew results.

In the United States, people tend to think of themselves as independent thinkers who develop their own opinions rather than blindly following the pronouncements of party leaders. Americans tend to see themselves as more objective than the average U.S. citizen, seeing bias at work in others but less so in themselves (Pronin, Lin, & Ross, 2002). A “third-person effect” has also been observed in people’s response to mass media. This means that media consumers tend to think more about a persuasive message’s effect on others, dismissing the idea that they themselves might fall prey to the media’s persuasion (Davison, 1983). These studies suggest that American voters consider themselves objective in their evaluation of political speech, but studies that test for evidence of subjective influences show that this perceived objectivity does not hold up in practice.

Cohen exposed the subjective analysis used in political thinking in his study of the effect of partisan cues on policy attitudes. In this study, party members were shown to support a policy that matched their own ideology when no reference group information was provided, but when a proposal was presented as the policy of either the Democratic or Republican Party, participants almost always supported the policy of their own party, regardless of its consistency with their own ideology (Cohen, 2003). When policies include “source cues,” which are indications of the partisan source of a message, subjectivity enters the process. These results support my hypothesis, providing evidence that voters will take the word of their party leaders over their own ideologies, suggesting that party leadership may influence their constituents’ policy views rather than the other way around. Naturally, people are predisposed not only to trust their party’s leaders, but also to mistrust and disapprove of leaders of the opposing party. Aaroe’s studies show that partisan source cues decrease support for a policy among political opponents (Aaroe, 2012). Aaroe and Cohen’s studies are especially useful in the defense of my argument because they show the same trend in policy approval that I predict will be observed in political speech processing.

Psychologists have suggested various reasons to explain why people allow subjective judgments to enter their evaluation process. The Elaboration Likelihood Model proposed by Petty and Cacioppo assumes that people are not always motivated or able to think about messages carefully, so they may employ two different “routes” to evaluate persuasive messages. When people are motivated and able to think carefully, they follow the “central route”, which involves an effortful process of comparing relevant information to previous knowledge and developing a judgment of the message’s merit. However, when people lack the motivation or time required for the central route, they take the “peripheral route”, relying on indicators external to the message itself, including source attractiveness (Petty, Cacioppo, Strathman, & Priester, 2005). In the political realm, the political party associated with a message’s source may be the principle element of source attractiveness, especially for decidedly partisan listeners. The Elaboration Likelihood Model can be applied to political speech, as voters who are inundated with it may be more receptive to the peripheral route of persuasion, basing their judgments on the speaker’s party identification rather than their message’s content. Mondak argues that when people lack understanding of or interest in an issue, they have a high need for cognitive efficiency (Mondak, 1993). When this happens, they use heuristic processing with source cues, in which they take a prior judgment about the source and associate it with the political message or policy, thus judging speech by their opinion of its source rather than its content.

Method

My study extends Cohen’s research to apply the effect of partisan source cues to the evaluation of the more general ideas presented in political speech, including ideological assertions and claims about the state of American politics, rather than having subjects choose between specific policies. Cohen makes changes to the articles used to describe the policies in his experiment so that each policy is framed in a way that makes it more acceptable to the party of the subject, but my study presents the same statements to each subject with no alteration to the statements. Only the source information changes between different experimental groups. This eliminates framing influences that may have affected Cohen’s results.

In order to gauge the effect of partisan source cues on people’s evaluation of a political statement, I employed a survey method with a control group and two experimental groups. The survey was conducted online and distributed through social media, so it does not constitute a representative sample of U.S. voters, but random assignment to test groups nullifies the representation problem by controlling for any confounding variables. 450 people participated in the study, and about half of the subjects identified with one of the two major U.S. political parties, with 87 Republicans and 136 Democrats taking the survey. 52% of respondents were male and 48% were female.

Subjects were randomly assigned to one of three groups, and each group took a slightly modified survey. Each subject was asked to indicate their level of agreement with ten political statements, choosing from the options “strongly agree,” “agree,” “neither agree nor disagree,” “disagree,” and “strongly disagree”. Various statements represented either typical conservative or liberal opinions, and some were ideologically neutral. Statements ranged from ideological claims, such as “Personal freedom is the defining value our country was founded upon,” to claims about the current state of affairs in the U.S., such as “The economy of the United States is strong and robust”. Results indicate that the statement set was balanced and did not favor one side over the other, as Republicans and Democrats had similar overall agreement levels for the survey statements on average.

All three versions of the survey contained the same ten political statements, but the survey given to each treatment group differed in its source information. The survey for the control group listed the statements plainly with no source information. The first test group presented each statement as a quote from a U.S. congressman or state governor. Extremely high-profile political figures such as Barack Obama, Mitt Romney, Chris Christie, or Nancy Pelosi were not used as sources. Half of the statements were presented as being said by Democrats and the other half were presented as quotes by Republicans. The second test group’s survey also listed statements as quotes, but for this group, the source given was a politician from the opposite party of the politician who was quoted for that statement on the first test group’s survey. This way, we can look at the difference between evaluations when two candidates present the exact same message. This means responses can be analyzed to see if Democrats more often agree with statements that were thought to be stated by Democratic candidates than statements perceived to be from Republican candidates.

Results

Results were analyzed by assigning an “agreement score” to each response, and then composite agreement scores could be compiled for a group by averaging scores from individual responses. A “strongly disagree” response correlates to an agreement score of zero, while a “strongly agree” is a 100. When the control group’s agreement scores were calculated for all statements, the two parties showed very similar agreement levels (Republicans 63.5, Democrats 63.3), indicating that the statement set was balanced.

Analysis of responses by subjects from both parties yielded interesting results that support my hypothesis. With scores averaged from all statements in the survey, Republicans expressed higher agreement levels for statements they believed to be said by Republican politicians (score: 63.6) than those they believed to be from Democrats (score: 58.9). Democrats showed the same trend of preferring statements from their own party, having an agreement score of 67.5 for statements with Democratic sources and a score of 61.0 for those with Republican sources. The results show statistically significant differences between agreement levels for subjects from both parties (p < .03), so the disparities between the answers given by different treatment groups can be attributed to the source cues rather than random chance.

The positive and negative agreement effects of partisan source cues that Cohen and Aaroe observed were seen in different measures in this study’s results. Republicans who took the survey were more susceptible to contrast effects, tending to agree with the Democrat statements significantly less than in the control group. However, positive agreement effects were not as strong for Republicans, as the average agreement score for statements from Republican politicians was about the same as the control group score. Democratic subjects showed the opposite trend, as they agreed with statements quoted from their own party at a significantly higher level than in the control group, while the composite agreement score for statements by Republicans was not much lower than that of the control group.

Analysis & Discussion

The results indicate that people do take partisan source cues into consideration when evaluating political speech. Knowledge of a speaker’s party identification was shown to make people more receptive or skeptical of the general ideas presented, which has important implications for the way people form their political opinions in a democratic society. People tend to think that they form opinions on their own and then choose to ally with politicians that support those opinions, but this study’s results show that party leaders can influence their constituents’ responses to political ideas. The interaction between politicians and public opinion is a two-way street, with each influencing the other’s positions and ideologies.

This is not to say that voters blindly follow party platforms and use only partisan source cues to make judgments. If this were so, the agreement score for party members would fall between 75 and 100 for statements quoted from their own party’s politicians, because they would choose “agree” or “strongly agree” every time. Accordingly, the agreement score would fall between 25 and 100 for statements said by opposing party candidates, because partisans would choose “disagree” or “strongly disagree” for every statement from an opponent. Instead, the agreement scores fall between 25 and 75 for both parties, indicating that other factors are at play. People still consider their own ideology when processing political speech, but disparities in agreement when the source’s party changes suggest that people frame statements differently in their minds based on the political party of the source, perhaps assuming different intentions or different meanings behind a statement from what they know about each party’s values. Petty & Cacioppo’s Elaboration Likelihood Model can explain the actions of some respondents, as some people take the peripheral route, using party indicators to help them answer questions for which they are unsure of their own opinion or are not motivated enough to think about.

The results of the study presented here have important implications for Americans. As voters, it is important that we think carefully about the political speech we hear and to dissociate the content of candidates’ messages from their party affiliation. By using the peripheral route to evaluate political speech, people allow campaigns to rely less on substantive speech in their advertisements and other discourse. It is easy to assume that politicians we have agreed with in the past will continue to support the agreeable policies and hold true to their professed ideology, but this is often not the case. Everyone is subject to the effects of source cues, even though most people believe they think objectively. It takes conscious effort to overcome one’s own biases, but the effort is worthwhile if voters in the United States want to cultivate the country’s tradition of individual thinking.