Role Of Twitter During The Elections Essay

In a media climate with constantly changing technology, how citizens receive their election coverage greatly impacts their perception of American democracy. The modern media increasingly covers elections using strategic frames rather than focusing on the issues. Also, the news media tends to cover elections in a way more favorable to Democratic candidates than Republican ones. In addition, the rise of new media significantly effects election coverage and how candidates reach voters. In particular, the medium of exposure to political candidates running for office has a tremendous effect on how viewers perceive these candidates. This effect was captured in James Druckman’s experimental study of the differences between analysis of radio listeners and television viewers of the 1960 presidential debate between John F Kennedy and Richard Nixon. To capture the effect of how new 21st century mediums of media affect political behavior, a randomized controlled experiment evaluating the effect of Twitter while watching televised presidential debates on participants’ perceptions of candidates and their political behavior is needed.

The news media’s increased use of strategic frames represents a change from the mid-20th century that defines modern election coverage. Over the last half century, news coverage of elections has shifted away from focusing on policy and issues to focusing on the strategic elements of the campaign and horserace coverage. According to Thomas Patterson, the media views campaigns as a game and covers policy as a strategy to win the election rather than as substantive stances on the issues (Patterson 96-100). Because the modern media is most concerned with framing election coverage in a strategic way, an increase in interpretative frameworks is associated with the modern media. Moreover, Majorie Hershey undertook a content analysis to conclude that strategic frames especially dominate the last week before the election. In her study, she found three quarters of broadcast media coverage in the last week before the 2000 was framed in a strategic horserace manner (Hershey 66). In addition to strategic frames, negativity is a major characteristic of modern election coverage.

In general, the modern media covers elections in a negative light and in a way much more favorable to Democratic candidates than Republican candidates. According to Graber and Dunaway, the modern media is biased towards covering politics with negativity, because cynical coverage of the political game sells newspapers (Graber/Dunaway 346-348). Moreover, the modern news media covers elections with a partisan bias favorable towards Democrats. According to Stephen Farnsworth and Robert Lichter, Democrats received more favorable coverage from the news media in every election except for the 1988 race between George H.W. Bush and Michael Dukakis (Farnsworth/Lichter 104). In addition, they believe unfairness overshadows election coverage. For example, Democrat Barack Obama received 68% favorable coverage in 2008 while his opponent Republican John McCain received favorable coverage only 33% of the time (Farnsworth/Lichter 99). Clearly, modern media is plagued by negativity and biases that generally favor Democratic candidates.

The rise of new forms of media significantly shapes modern coverage of elections. With the rise of Twitter, Facebook, and other social media websites that give a voice to millions of people, the American public is increasingly looking for alternative sources of news. Specifically, Twitter’s 140-character limit reflects the tendency of the shrinking soundbite. From 1968 to 2008, the average length of time a candidate’s actual words were aired by television news dropped from 42 seconds to 8 seconds (Lecture 4/4). Moreover, Twitter and these new forms of media facilitated what Marvin Kalb refers to as a lack of sourcing and a rush to judgment (Kalb 11-13). These tendencies have led the public to portray the news media in a much more negative light than decades ago. As a result, candidates have found ways of bypassing traditional media sources to reach voters directly through Twitter and new media sources. This change in medium of media consumption can have huge effects on political participation by the average American.

James Druckman studied how the medium of media affects political behavior by creating a randomized controlled experiment surrounding the differences between radio listeners and television viewers of the first televised debate in 1960 between Richard Nixon and John F Kennedy. Druckman’s research design consisted of 210 younger people of voting age in 2003 taking both pretest and posttest surveys to determine eligibility for the experiment and how the participants reacted to the debate. Because he wanted to mitigate confounding variables and reduce bias, Druckman omitted any participant that had any prior exposure or knowledge about the Kennedy-Nixon debate and was left with 171 participants. To provide context to the debate, Druckman gave a brief background of the election to the remaining participants and showed both research groups basic images of the two candidates. Then, he used random assignment to assign 85 participants to watch the debate on television and 86 other participants to listen to it on the radio (Druckman 564-565). Druckman’s use of random assignment successfully combats the problem of self-selection into a certain treatment group, because there are no apparent pre-existing differences between the two groups due to the nature of random assignment. Therefore, the results of the experiment can be attributed to the content of the study rather than due to overwhelming differences between the socioeconomic compositions of the two treatment groups. To analyze the results, Druckman compared the survey responses and highlighted aggregate differences between the two isolated groups. Certainly, Druckman’s study has high internal validity due to the research design’s nature as a randomized controlled experiment.

While Druckman’s experiment’s internal validity appears satisfactory, several issues remain with regards to causal interference, omitted variable bias, self-selection into the study, and external validity. First, Druckman’s study falls victim to the fundamental problem of causal interference in the manner that he can’t observe the same subject in both experiments. While he can’t fix this problem because participants subjected to both experimental group will be influenced by their first experience, random assignment does not completely eliminate any margin of error resulting from this issue. Also, Druckman’s study may be influenced by confounding variables not accounted for when designing the study. For example, one’s level of exposure to radio and television may affect their perception of the candidates in a way not quantifiable to the researcher. Moreover, the self-selection of participants into this political study naturally captures a group of participants who are more politically knowledgeable than the average American. For similar reasons, Druckman’s study suffers from low external validity due to a non-representative sample. Because Druckman’s sample consists of University of Minnesota college students, his participants are significantly younger than an average American, more educated and disproportionately from the Minnesota area. To be able to generalize Druckman’s study to the general public, the experimental participants must reflect the demographics of the population intended to study. Lastly, Druckman’s study comparing television viewers and radio listeners is out of date, because the television consumption of media has soared since 1960 while the radio media ratings decline yearly.

While Druckman’s study succeeds in determining that television viewers rely more on personality when judging political candidates and radio listeners rely more on issue agreement, the rise of new news makes his findings obsolete for today’s media culture. Specifically, today’s consumption of media is defined by Twitter’s power to influence the political message being transmitted to the viewer. For example, the Brian Williams scandal was unearthed through a series of Twitter and Facebook posts by soldiers present at the time of the incident (Somaiya). Moreover, police brutality in Ferguson, Missouri in 2014 flew under the national radar until Twitter lifted it into national consciousness through the use of #Ferguson. According to New York Times writer David Carr, Twitter allows information to be disseminated throughout the internet much more quickly through the use of condensed 140 character sound bites (Carr). Candidates have increasingly been using Twitter to clearly and distinctly transmit their political message directly to potential voters in a way that reflects both the trend toward shrinking soundbites and declining trust in the media (Lecture 4/4). Therefore, a study will be conducted measuring Twitter’s effects on political behavior.

This study will test the research question of how the use of Twitter effects which issues American citizens find most important, the criteria for how Americans judge candidates, and citizens’ attitudes toward American democracy. To answer these questions, this study will be conducted based on the independent variable of method of watching the upcoming fall 2016 general election debates between the Republican and Democrat nominees for president. Research participants will be randomly assigned to two groups in order to eliminate any pre-existing demographic and socioeconomic differences between the research groups. The first group will watch the full debate on television. In the other group, participants will watch the debate on television while simultaneously following the Twitter accounts of both the Republican and Democratic candidates on tablet devices. In addition, this group will have access to the trending topics on Twitter in order to expose them to popular mass reactions to the debate. This duel-consumption of the debate between both the television coverage and Twitter feed simulates the popular 21st century two-screen living room experience. After the debate, participants will fill out a survey asking participants to rank their most important issues and their criteria for their evaluation of the candidates. The last section of the survey will ask participants to rate their trust in the American democratic system on a scale on 1-10. After this assemblage experiment, I will compare the results of the surveys between the two groups in order to draw any conclusions about the effect of Twitter on political behavior.

Going into this study, I have three hypotheses for the effects of Twitter on political behavior. First, I hypothesize that the Twitter group will identify social issues such as education reform and healthcare as more important than the television-only group due to their exposure to the voices of millions of people directly affected by social programs via Twitter. My second hypothesis is that the television-only group will focus more on personality and image for evaluating candidates while the Twitter group will focus more on the issues due to their print exposure to the candidates’ rhetoric. Third, I hypothesize that Twitter users will have a more negative view toward the candidates and American democracy due to their exposure to the cynical masses. My underlying assumption behind these hypotheses revolves around the overly cynical view toward the state of American democracy influencing these participants through the mass medium of Twitter.

In this experiment, I will use random assignment to combat challenges to internal validity. Random assignment among a large pool of participants attempts to eliminate the effect of omitted variable bias and will ensure that there are no preexisting differences in group composition, because participants will have an equal chance of being assigned to either treatment group. Also, random assignment eliminates problems having to do with self-selection into one group or the other. Moreover, the experiment will be held in real time which eliminates the problem of bias from prior knowledge of the debate that Druckman dealt with in his Kennedy-Nixon debate experiment. Also, the experiment will take place in two identical rooms to mitigate any bias surrounding viewing conditions. In order to overcome issues with small sample size, this experiment is expected to be conducted with 1,000 participants total. Therefore, this experimental design allows the study to have high internal validity.

This study attempts to significantly improves upon Druckman’s original work in terms of external validity. In an ideal world, research participants would come to our lab from all corners of the country and perfectly represent the demographics of the United States. However, because resources are scarce, we’ve decided to set up our lab in fictional Purpletown, Pennsylvania which boasts a population with a wide range of demographics and is consistently a swing county in elections. Because this location closely reflects the demographics of American citizens in terms of race, education, income, and political preference, the results of the experiments can be generalized to the American public. This aspect of the study significantly improves Druckman’s population of educated University of Minnesota college students. Also, the representative proportion of self-identified Republicans and Democrats will mitigate any partisan and ideological bias from the results of the study. In addition, the experimental setting where participants will be held for the duration of the debate will be staged with couches and coffee tables to closely resemble the average American living room. Clearly, this study attempts to capture a representative sample that results in high external validity and produces results that can be generalized to the American public at large.

Overall, the rise of new media has significantly changed American election coverage. Twitter and other forms of social media encourages candidates to talk in easily understood sound bites that can be condensed to 140 characters. Moreover, these forms of communication that instantly transmit information to millions of people across the globe lead to a rush to judgment among reporters to be the first one to break a story. While James Druckman conducted an experiment in 2003 studying the effects of television on political behavior, his study needs to be updated to capture changes in technology. This proposed randomized controlled experiment will capture the effect of Twitter and new media on political behavior in a way that is both internally valid and generalizable to the entire American population.

How to cite this essay: