7 Congress 2.0—Controlling the Flow of Information

The importance of social media in the political sphere is based in part on how we consume and understand information. In previous chapters we have set forth a theoretical foundation for our premise that the use of SNSs will have a substantial and significant effect on the nature and structure of our political discourse. We base our proposition on the essential structure of how people obtain and understand political information. Peoples’ attitudes are made up of an average of the range of relevant considerations they may have (Zaller 1992). Since the world and the information environment are not static, people are constantly updating their assessments of politics, policy, and political figures based on the information they receive. This ongoing progression, referred to as online processing, can be influenced by the nature, magnitude, and scope of the information as they sample it from external sources (Lodge, Steenbergen, and Brau 1995; Lodge 1995). As a result, what people know and understand about politics is based on the types, frequency, and point of view of the news and analysis they consume and how they reconcile it with their worldview and predispositions.

Substantial studies on this cognitive process have been the focus of major research (Lodge 1995; Zaller 1992). A substantial piece of the impact of a new media format such as social media is placing it in our current understanding of the cognitive process. We need to consider how social media can alter the process people engage in when deciding which information they will consume and how they will view it. This process is based on an understanding of the psychology of forming beliefs and worldviews. It is also based in part on comprehending the media environment in which people are learning and consuming information. The latter is fertile ground for research. Media politics is its own subdiscipline and has been frequently studied (Bennett 2011; Graber 2010; Prior 2007). Tendencies toward misinformation, infotainment, profit-driven content, commercial influences, and structural biases are prominent outcomes of the mass media model (Bennett 2011).

Social media supplies a paradigm-changing element to the system. Using SNSs to obtain information presents a very different dynamic. The consumption of information is interactive and occurs inside networks of friends and acquaintances. As we discussed in previous chapters, this type of information distribution has some increasingly measurable effects on participation, polarization, group formation, and levels of knowledge. The age of the mass media and its significant pathologies are not over, but rather a new distribution channel is available, which brings its own implications, both positive and negative. Many of the influences we have discussed in earlier chapters are important measures of this new and perhaps novel public sphere. However, they are only a portion of the larger influence of SNS use. All speak, to some degree, about the needs and wants of the information consumer and putting that consumer in some control of their content. As we observed, this can cause important changes to what is consumed and how it is understood.

However, the rise of SNSs and their growing use has a larger importance beyond the consumer. It is an opportunity structure for political actors. As we have argued, the SNSs are an information distribution channel that are open to anyone and allow interactive communication outside the purview of the modern media. Consider that for much of our history, political communication was based on utilizing an often-expensive method of mass information distribution such as a printing press or later a radio or subsequently television broadcast. One of the key limiting factors for political actors was the limited ability they had to use these channels and the mediating effect of the owners of them. With SNSs, the channel is unmediated, the messages are unrestrained in number or frequency, and the distribution is limited only by the rapidly increasing scope and penetration of the Internet and social media.

Earlier we examined how politicians were using this new medium, focusing in particular on Twitter. In this chapter we are going to consider more specifically the nature of the information flow and the means by which politicians can use Twitter to manipulate the scope and direction of the information being given to followers and different networks of people. We will look at why Twitter creates such a strong environment for message control and how political actors can take advantage of that structure to reinforce preexisting views as well as to signal to their followers how they should understand and consider information. Too much of the focus of the work on social media assumes that the individual consumer is the new content editor. We suggest that the consumers’ power is far more illusory. Increasingly clever campaigns help the consumer come to conclusions, though effective online campaigns worked increasingly through ascertainable networks of like-minded friends and acquaintances.

This chapter will make an argument that is two-fold. First, a theory will be developed asserting that use of the Twitter tools described above may serve candidates who are trying to control the flow of information better than simply using simple 140 character or less tweets that fit into the four-category typology developed in Chapter 4. Second, we further develop the argument that Republicans and challengers did a better job of controlling the flow of information in this election cycle but suggest an alternative argument to those presented in Chapter 5. We contend that it is possible that Republicans and challengers by circumstance happened to be on the popular side of the information flow, and Twitter provided the perfect opportunity for them to capitalize on these information control tools as a means to further crystallize their support.

This theoretical development is followed by some descriptive analysis exploring the distribution of use across partisanship, incumbency, chamber, and race identifying that Republicans, challengers, senators, and white candidates tended to use these tools more frequently. We then present qualitative data that provide more examples of how these Twitter tools were used (links, @Twitter names, and hashtags) in particular by Republican challengers. Finally, we conclude with a multivariate model demonstrating that Republicans, challengers, and Senators tended to use these tools of information control more frequently, ceteris paribus.

One of the great media misconceptions held by many Americans is the idea that there is an absolute objective means to report information. Ironically, even attempting to generate news that is divorced from politics or somehow objective creates its own type of bias (Bennett 2011). In every system, there are advantaged parties and disadvantaged ones. When print media was dominant, the most advantaged were the ones that owned the press or could afford to purchase space. In the television age, resources were still an advantage, but visual appeal became important as well. In the Internet age, we would like to believe that infinitely expanding the channels of information distribution created an open and perhaps even democratic media environment. The inherent openness of the Internet leads to a somewhat optimistic view of the future (Allison 2002). However, the removal of the media owners from a dominant position in the distribution network is not necessarily a panacea. The new system, like all previous systems, creates opportunity structures that can be advantageous for some (Riker 1986). It is never a question of whether a system can be exploited, it is a question of who will be in position to do it.

Perhaps the biggest influence of the social media may well be in the shifting of power in the political communication sphere, not from the media to the individual, but from the media to various politicians and political interests. It is in the elevation of the political actors to the role of unmediated information provider that the most lasting and substantial effect may be occurring. Twitter is a new medium that candidates can use to provide and influence the considerations being evaluated by users/potential voters. In doing so, they can circumvent the traditional media—the normal gatekeepers. People are choosing, but it is the political actors that are often doing the speaking. As a result, social media appears to provide open communication with limitless choices, but in escaping the media’s filtering, the SNSs have provided a fertile ground for the political actors to drive and influence content.

The early adopters are biggest beneficiaries so far. The narrative that became evident in Chapters 4 and 5 persists in the data here. Challengers and Republicans did a better job of controlling the flow of information with their use of Twitter in the 2010 election cycle. Chapters 4 and 5 laid the foundation for understanding the ways candidates may use Twitter to campaign. It was there that we started to build the empirical foundation for how candidates may have used Twitter in the 2010 election cycle in an attempt to control the flow of information. We did so by highlighting how they may have used links, @Twitter names, and hashtags in their tweets to secure support. Some of the same strategies that we have traditionally seen members of Congress use, such as advertising, credit claiming, and position-taking are evident (Mayhew 1974). However, they are done with no mediation and with increased frequency, targeted to specific audiences with the advantage of requiring few resources and no substantial costs. Twitter maximizes all of the traditional strategies with unprecedented efficiency, speed, and responsiveness.

Since Twitter makes it relatively easy to reach networks of supporters without having to meet journalistic standards of accuracy or even relevance, candidates have unprecedented control of their message and messaging. With such levers on information distribution, candidates are able to control the flow of information by providing links to types of news, endorsements, and blogs that present them in a favorable light. For some users who obtain most or even all of their information from social media, the influence of political actors can be determined. The idea of living in a bubble of one type of information is no longer theoretical, it is an obtainable reality. It is also a reality that people are creating around them, without necessarily understanding its existence. It is in this bubble that opinions can be reinforced and contrary information excluded or eliminated.

While these networks are problematic in the area of accuracy, there are other concerns as well. Significantly, SNSs are a media reality readily exploitable by political actors who will integrate into the network and cater to the predispositions of its users. Yet it is not solely message control, which would alone be paradigm shifting. It is messaging in a targeted way toward populations and networks with readily ascertained predispositions. Followers of particular people or groups are cognitively receptive to information tailored to their predispositions. For example, a network of liberal activists is fertile ground for messaging on income inequality or alleged threats to the social safety net. As people are uncomfortable when confronted with information that is not consistent to their predispositions, they order themselves online into groups and networks with very real and consistent patterns or beliefs and understandings. These patterns of networks online are the inadvertent effect of human psychology and the desire to avoid cognitive dissonance or any general discomfort. Challenging information causes discomfort. Agreeable information sources are preferred, because they prevent the user from experiencing discomfort by helping them avoid exposure to any contrary information which could cause confusion or doubt.

A prime example of this trend was in the 2012 US presidential election. Conservatives and conservative political actors discarded the weight of the polling information that suggested a narrow but persistent lead favoring President Obama and redirected their followers using social media to blogs, media, and other postings that were said to report polling without media bias (Morris 2012). Many tweets directed conservatives to Unskewed.com, a website that purported to fix polls by weighting them with greater numbers of Republican voters. Other tweets simply confirmed that the polls in general were wrong with a reference to supporting websites. The election results, which were largely consistent with the disparaged media and public polling, surprised many conservatives, including the authors of some of the websites championing their incorrectly adjusted polling (Benson 2012). This type of cueing through the social media is possible because it plays on the inherent distrust in the media by conservatives. The organization of people ready to believe that the media intentionally favors one political party already exists. The exploitation of that network is relatively simple as a result. The social media can play a significant role in creating or reinforcing predispositions like this as long as the politicians and political interests continue to self-interestedly drive content consistent with such a narrative into targeted elements of the SNSs.

Not only is Twitter ideal for this type of messaging, the tools inside Twitter are useful in fine-tuning the message and its audience. A tweet can be personalized inside these networks, making it more likely to be read and more carefully directed. Candidates can respond directly to one person, but with a public message. This message would have the @Twitter name of the individual in it. A retweet would also have a @Twitter name in it. This is a way for the candidate to appear concerned about an individual with very little resource cost. Politicians, or their staff can respond directly to constituents or can pass on positive information that was generated from somewhere else. An @Twitter name can also be included in a tweet to link into the followers of another Twitter user that tends to be favorable toward the candidate. Taken altogether, including an @Twitter name is an effective method to help define and control the flow of information.

The name is important, but a more significant organizing tool inside Twitter is the hashtag. Political actors may use hashtags to direct content and emphasize particular facts or points of view. This is a particularly important means of controlling the flow of information. As described earlier, hashtags are when a user includes the # symbol to mark keywords or topics in a tweet. People use the hashtag symbol before relevant keywords in their tweet to categorize those tweets, making them identifiable in a Twitter search. Hashtagged words that become very popular are often categorized as “trending topics.” These are topics that are immediately popular rather than topics that have been popular for a while or on a daily basis. They can cue people to the most important breaking news stories from around the world. They can also suggest or signal to a person to how they should understand the information and its relative importance. A hashtag has many uses but is a particularly convenient or even shorthand way to insert an opinion into an otherwise neutral statement. Political actors can do this to the stories relevant to congressional candidates, or to generally trending topics that can be made relevant to a particular point of view. More directly, candidates can try and marshal or influence the flow of information by using hashtags that connect their tweets into popular groups and/or topics.

As a campaign and political tool, Twitter has much to recommend it. It is unmediated, immediate, and able to be targeted at important groups with little cost. The Twitter conventions including links, hashtags, and user names are useful for directing information and reinforcing messaging. In the Twitter universe, the sophisticated political actors are the advantaged users.

The quantitative analysis in this chapter is based on counts of the number of times candidates had linked information in their tweets (counted by checking to see if the tweet contained “http”), included @Twitter names (counted by checking to see if the tweet included an @ symbol), and hashtags (counted by checking to see if the tweet included a # symbol). After doing counts for each candidate, we also constructed separate measures to gauge the frequency with which candidates were using these means of controlling the flow of information by calculating the percentage of their total tweets comprised of each respective means (http, @, #). We simply divided the number of times each candidate used one of these tools, respectively, by the total number of tweets. This means that it was possible for the percentage of use of these tools to be higher than 100%. This outcome is possible because many candidates are using links, @Twitter names, or hashtags, more than once in a given tweet. We chose to measure it this way because including more of these tools in a single tweet should have a greater effect on the control of the flow of information and connect followers to a wider audience. This practice was quite common as we saw from the qualitative data presented in Chapters 4 and 5. For the analysis in this section we, first, plot the frequency of this percentage for each outcome. Next, we explore the number of times candidates used each of these tools across party, incumbency, chamber, and race.

The first observation that can be made from the distributions displayed in Figure 7.1 is that generally speaking these tools seem to be used quite frequently. Notice that the proportion of the X axis often exceeds 1.¹Close The results illustrate that most candidates have a very high proportion of their total tweets containing links to other information. In fact, the modal outcome was 60% and the mean is around 65%, suggesting that generally most tweets included a link. This is a significant result because providing a link to information that paints the candidate in a favorable light may be the most effective way of using Twitter to control the information that followers have cognitively accessible to form an opinion about the candidate. While the distribution of candidates having tweets with a @Twitter name in them was tilted toward 0, this was nonetheless still a quite frequently used tool. On average around 29% of the total tweets by candidates contained an @Twitter name. Again, remember that this number is higher than the actual number of tweets containing an @Twitter name because we divided the total number of tweets by the total number of @Twitter names in those tweets, but nonetheless it reflects that this is a fairly common practice. This means that there are over a quarter of the instances of the use of an @Twitter name relative to the total number of tweets. Finally, the distribution of total tweets with a hashtag was also titled toward zero, but nonetheless, hashtags were not used infrequently. On average, the calculation of number of uses of a hashtag divided by the total number of tweets was around 52%. Again, many of them had multiple hashtags, suggesting that candidates were trying to connect constituents to multiple streams of information simultaneously.

Next, we examine potential differences in the types of candidates who use these different tools for controlling the flow of information. As described above candidates did so quite frequently. In fact, Tim Griffin, a successful Republican challenger for a House seat in Arkansas included 1,311 links in his tweets and he also included an astonishing 4,399 hashtags. While Sean Bielat, an unsuccessful Republican challenger for a Massachusetts seat in the House included an @Twitter name in his tweets an astonishing 2,136 times, Griffin also had a sizable number of @Twitter names at 378. These two candidates were certainly tweeting more than most but as demonstrated in earlier chapters and in the above results, they were not alone and there does appear to be a pattern emerging of who is most likely to both tweet and use certain tools.

The results in Table 7.1 clearly indicate that Republicans and challengers are more likely to seek to control the flow of information via Twitter by including links, replies, and hashtags than their respective counterparts. In fact, the mean number of tweets including a link or @Twitter name is around twice as much for Republicans relative to Democrats (approximately 118 to 61 and 72 to 31, respectively) and over three times as many hashtags are included in Republican tweets relative to Democrat tweets (178 to 55). The pattern is much the same across incumbency. Challengers were also around twice as likely as incumbents to include a link or @Twitter name (approximately 115 to 58 and 67 to 34, respectively) and over four times as many hashtags are included in challenger tweets relative to incumbent tweets (179 to 41). Finally, Senators and white candidates are generally more likely to do all of these things with one exception. The data suggest that Latino candidates were more likely to use @Twitter names.

Perhaps one of the most subtle ways that candidates can use Twitter to control the flow of information is to include @Twitter names. As described earlier, this provides candidates the opportunity to reach a wider audience by potentially connecting them with the followers of the user with the @Twitter name they included. And by including their own @Twitter name, it makes it easy to build new followers if this tweet is retweeted. People can simply click on the candidate’s @Twitter name and then choose to become a follower. We provided many examples of such in Chapters 4 and 5 but provide another here to be clear about what we are measuring in this chapter and why we are measuring it. Because we will continue to build on our argument that Republican challengers more effectively capitalized on the use of Twitter in the 2010 election cycle, we provide an example here of how such a candidate used this tool, and presumably increased his control over the flow of information.

We use Tim Griffin again as our exemplar here. He retweeted the following tweet: Today I voted for: @griffincongress @markdarr @Keet4Arkansas @Boozman4AR jeremy hutchinson and the best of all @David_J_Sanders. He really got the maximum use of the inclusion of @Twitter names. First, the tweet includes his @Twitter name (@griffincongress). Next, it includes the @Twitter names of several prominent Republican candidates from Arkansas including Mark Darr, who had a successful bid to be the lieutenant governor, Jim Keet, an unsuccessful gubernatorial candidate, John Boozman, a successful candidate for the U.S. Senate, and David J. Sanders, who is in the Arkansas House of Representatives but led an unsuccessful bid for state senate. The inclusion of all these candidates is clearly intended to put Griffin in with a group that he believed would be viewed favorably by the audience he is trying to reach. Finally, this retweet originated from a voter who had taken advantage of early voting in Arkansas. By including this fact, Griffin could potentially stimulate early turnout among others. Our data cannot tell us if this in fact worked, but we do know that he won the election as a challenger.

Hashtags are another way that candidates can guide or control the flow of information. We highlighted the use of hashtags in Chapters 4 and 5 but will dig deeper, empirically, here. Before moving on to a quantitative analysis of the use of hashtags, we provide examples below of some of what we consider to be the effective use of hashtags by Republican challengers in this election cycle. Again, we focus on Republican challengers because as the quantitative analysis in previous chapters has suggested concerning other uses of Twitter and in this chapter concerning the use of hashtags, Republicans and challengers tend to use this tool more often. Note that we include three broad types of hashtags in this list: (1) those that tap into a broad, perhaps national or at least statewide, audience, (2) those that reach a race/election specific audience, and (3) those that are centered on an idea or policy. We think that those who try to do all of these, not necessarily in a single tweet, are the most effective. Here is a list of particularly pointed and effective hashtags (some were already highlighted in Chapters 4 and 5):

#tcot: The creators of this hashtag characterize themselves as “the twitter hashtag for following top conservatives on twitter.” This hashtag came up quite frequently. For this reason, it is a great strategy for candidates to include it in their tweets. If it is coming up frequently, there will be great numbers of Twitter users both checking it and including it in their tweets. This provides tremendous opportunity to reach a wider audience. It also attaches Republican candidates to an idea (conservatism) that is popular among their potential voters.

#TeaParty: The Tea Party was quite a popular movement among conservatives during the 2010 congressional elections. Thus, any connection to the Tea Party was likely beneficial for many conservative candidates. This also provides opportunity to reach a wider audience.

#GOP: This is a general hashtag that will connect the candidate to broader discussion going on related to the Republican Party.

#ILSen: This is a hashtag that refers to the Illinois Senate. We cited examples of similar hashtags in Chapters 4 and 5. The idea here is that people doing searches about the Senate (for Illinois in this case) under this hashtag will find tweets discussing issues related to the state’s Senate delegation, election, or campaign. The interesting thing here is that candidates from both sides of the aisle may be using this hashtag, so actual debate and competition for voters can happen in the discussion under these type hashtags. This makes it important for candidates to use these hashtags so that they do not lose this battle for voters.

#azgop and #azright: These are two examples of hashtags referring to the Arizona Republican Party and right-leaning or conservative ideas/people from Arizona. While the previous hashtag for the Illinois Senate is open to the left and right, this hashtag seeks to create a more homogenous discussion. Like the Tea Party and Republican Party hashtags described above, these hashtags connect the candidate to discussions/groups that the candidate is targeting, but unlike those, these are even more targeted, specifically aimed at potential voters from the candidates’ home state.

#AZ01: This is another hashtag from Arizona. It refers to the First Congressional District of Arizona. This is clearly specific to that district and race. Again, this provides opportunity for the candidate to compete for followers, voters, and attention. It is key to include hashtags such as this because it directs one’s supporters to the discussion falling under this hashtag. The more followers pointed in this direction the more likely they are to post in the discussion. This can help a candidate to dominate the discussion, or at least the distribution or supporters/opponents, under this hashtag.

#socialsecurity: This is clearly a hashtag aimed at discussions centered on social security, an extremely salient issue. While this issue is of particular importance to seniors who are not as likely to be tweeting, it is also relevant to people in their 40s and 50s, who are likely voters and also represent a large share of users who gather political information via Twitter (see Table 3.2). Using a hashtag about a policy issue allows the candidate to feed its followers into the discussion. This can help win over other tweeters who are already in the discussion. A candidate can also use an issue hashtag when they believe they are on the winning side, or more popular side, of the issue. Finally, we think it is good practice to create hashtag discussions centered on issues where the candidate knows they are on the winning side. Altogether, candidates did not seem to use issue hashtags frequently, and most candidates who were tweeting did not use such hashtags. We think this is a missed opportunity to directly control the flow of information. They could use the hashtag discussions to post more information through links that painted the candidate or candidate’s position in a favorable light. Perhaps we will see more of this as candidates become more Twitter-savvy.

The final Twitter tool we discuss in this chapter that candidates have at their disposal is the inclusion of links. We think this is probably the most effective way that candidates can control the flow of information and circumvent the gatekeepers in the traditional media. Remember that our primary theoretical assertion is that Twitter provides an opportunity structure that is well-suited for both candidates on the supply side and voters or consumers on the demand side. Consumers tend to want information that reinforces rather than challenges their predispositions avoiding the potential for cognitive dissonance or any resulting discomfort. Candidates want their votes so they are inclined to want to provide them with supportive information. Consumers choose to follow those candidates for whom they are supportive and those candidates then provide them with information that reinforces their predispositions. Twitter facilitates this exchange. Providing links to news stories that are supportive of the candidate allow consumers to gather information without discomfort and builds support for the candidate.

We provide an example of a linked news story in Figure 7.2. This story was linked in a tweet that came from David McKinley, a successful House challenger from West Virginia. The tweet read: So why should I vote for David McKinley for the House of Representatives when he has promised to vote for tax cuts . . . http://fb.me/KYIxFSJY. The link was shared from Facebook and came from the website of the local newspaper in McKinley’s district, The Parkersburg News and Sentinel. This story is an opposite of the editorial page story (op-ed) and is clearly using a biting sarcastic tone. Its intention is to come from the conservative take on a liberal voice. The story offers a series of traditional conservative critiques of liberal policy (e.g., calls it socialist, big spending, sending jobs overseas, big government). Then it asks the question, “Why should I vote for David McKinley for the House of Representatives?” and replies with standard conservative talking points (e.g., he’ll cut taxes, support small business, oppose the national health care bill, and oppose Democrat leadership). The link to this story is a perfect example of how a candidate can provide his followers with information that reinforces their predispositions. The supportive consumer could read this and experience no discomfort.

Finally, we provide the results of the multivariate analysis. We modeled the number of times each candidate included, respectively, a link, an @Twitter name, or a hashtag as a function of party, incumbency, chamber, race (white dummy), district competitiveness, and spending differential.²Close The story here does not significantly change from that being told in Table 7.1 and earlier chapters. Challengers and Republicans appear more likely to use each of these methods of controlling the flow of information (see Table 7.2). This relationship endures even when controlling for chamber and race, and the introduction of controls for district competitiveness (the absolute value of the difference between the loser and winner in that district in the previous election divided by 500 thousand) and campaign spending differential (the absolute value of the difference between the candidates divided by 10 million). Chamber is the only other significant variable in any of the models, suggesting that use of each of these means of controlling the flow of information increases is more predominant for senators.

Because these models are negative binomial regressions, we include the incident rate ratios. Republicans’ number of links is, on average 1.91 times higher than Democrats’ number of links. Their number of @Twitter names is 2.25 times higher than Democrats and their number of hashtags is 2.72 times higher. These are relatively large differences. While not as large, the incident rate ratios associated with the estimates for the effects of incumbency are also meaningful. Incumbents’ number of links is, on average 0.60 times lower than challengers’ number of links. Their number of @Twitter names is 0.59 times lower than challengers and their number of hashtags is 0.31 times lower. The incident rates for chamber are also quite large. The model estimates that senators have 2 times more links in their tweets than House members, 1.85 times more @Twitter names, and 2.66 times more hashtags.

This chapter further developed our theory on how social media can be used as a tool to help candidates control the flow of information. Further, the empirical groundwork demonstrating that both Republicans and challengers seem to be employing such strategies with more frequency was extended in this chapter. Clearly both are more likely to include links to external information, to include @Twitter names, and to include hashtags in their tweets. This finding holds up ceteris paribus. Altogether this chapter is building toward our primary theoretical and empirical result in Chapter 9 that indicates the types of information control tools described in this chapter are the most effective at garnering votes and that Republicans and challengers benefitted more from the use of these tools than their respective counterparts.

Before moving to the analysis in Chapter 9 of the effectiveness of the tools described in this chapter, we revisit the demand side of the information flow by exploring the implications of heightened social media consumption on public opinion. If we are correct that citizens prefer gathering political information via SNSs because it allows them to avoid information that challenges their predispositions and that politicians are providing them with such information, essentially circumventing traditional information gatekeepers, then we should expect to see some attitudinal effects. In Chapter 8, we, first, develop our theory that explains why heightened SNS use should lead to attitude extremity, and second, empirically evidence this relationship across a series of varied political attitudes.