SOCIAL MEDIA IN POLITICAL COMMUNICATION

We rely on a developed literature in congressional political communication and media to explore why some members of Congress chose to adopt Twitter in the 111th Congress (2009–2010). It is well established that the national media focuses on congressional leaders and committee chairs when looking for coverage of the legislative process (Cook Reference Cook1989; Hess Reference Hess1986). Despite this challenge, press coverage remains a key strategy for promoting the policy goals, future electoral success, and political ambitions of all members of Congress. Members devote considerable resources to such goals. Recent findings suggest that during a time of budget cuts, the fastest growing staff position in the House of Representatives is “communications director” (Drutman Reference Drutman2012).

Despite most members' efforts, only a few senators and representatives receive national coverage for their congressional activities (Kuklinski and Sigelman Reference Kuklinski and Sigelman1992). Research suggests that senators who want to increase the volume of their media coverage can make their events less scripted and choose to share the limelight with fellow senators (Sellers and Schaffner Reference Sellers and Schaffner2007). House members may have to rely on hometown coverage, but the push for local media can be considerably affected by whether a member's district overlaps with a newspaper's market or if the paper is owned by a chain or conglomerate (Schaffner and Sellers Reference Schaffner and Sellers2003). Furthermore, members of Congress from nonwhite racial groups may also see their coverage filtered through a stereotypical lens that focuses disproportionately on race rather than the substance of potential legislative accomplishments (Schaffner and Gadson Reference Schaffner and Gadson2004).

Relying on conventional media outlets is not an optimal mechanism for most members of Congress to advocate their ideas or positions to a wide audience. Even the frequency of local coverage can depend on variables completely outside the control of a member. Additionally, news coverage can fall on deaf ears and may not always provide information to a member's most crucial audience, the “attentive public” concerned with a specific issue area (Arnold Reference Arnold1990). Social media provides a new avenue for members of Congress who want to control their public message, reach a national audience, or develop routine contact with an attentive public.

Reaching a national audience might not translate into the same goal for all members. Some members might truly want to generate the largest possible group of followers, perhaps in an effort to raise money or generate support for higher office. Other members might have more modest goals, such as developing a targeted national following based on political ideology or a specific policy interest. Others may want to focus on establishing surrogate representation, perhaps to speak for those looking for a national representative willing to lead on race, gender, or sexual orientation (Mansbridge Reference Mansbridge2003).

Choosing to adopt Twitter, or other social media platforms, has many inherent advantages to members of Congress. These include the lack of financial commitment (currently, there is no user fee to “tweet,” for example) and the relatively small start-up costs (all members and staff have computers, smartphones, and Internet access). Of course, all of these stated benefits are amplified if members believe that those people they are trying to reach are likely to use Twitter.

The decision for a member to engage in a form of social media, such as Twitter, however, does not come without costs. First, social media is relatively new technology, which can introduce uncertainty. Many members have established protocols in their offices for communication strategies, and introducing a new technology into the mix can cause confusion about priorities. Without careful coordination, a member's social-media message could diverge from the image projected in more traditional media. Second, staff resources are scarce and are diminishing as the legislative branch faces budget cuts. Although Twitter requires only 140 characters per entry, it takes time to strategize how to communicate a member's message in such a short, restricted format. Staff members also must monitor the account and respond to inquiries from followers (who may or may not be constituents). Failing to respond to such inquiries, such as replies to tweets, exposes members to the risk of seeming “out of touch” or “unresponsive.” Third, while many members may believe that the possibility of creating a national following is a benefit of social media, the spotlight likewise has the potential to invite unwanted trouble. Any comment that is tweeted can be easily “retweeted” and, therefore, be scrutinized by millions, including the national media, which often closely follow member Twitter accounts. Although considerable precautions can be taken, accounts can be hacked, and unwanted messages could be sent in the member's name. Finally, few specific institutional rules exist concerning the use of social media in both houses of Congress, which may cause members to think twice before engaging in these outlets.

Consequently, members must carefully weigh both the costs and benefits before adopting social-media platforms. Members possessing different district and demographic characteristics, as well as varying political ideologies, may look at the decision calculus in divergent ways. Depending on these characteristics, we hypothesize that some members will decide using Twitter is worth the potential costs; others will come to a different strategic conclusion. We examine member adoption of Twitter in the 111th Congress, a critical time for social-media adoption since the technology remained a relatively novel approach for political communication. In our model, we examine a variety of variables that may influence a member's decision about whether to pursue a social-media presence.

DATA AND METHODOLOGY

In December 2010, data was compiled on the adoption of Twitter by members of the 111th Congress. The data collected includes information on the 547 individuals who served during the 111th Congress. Of these 547 total members, 313 were registered with Twitter, including 62 senators and 251 representatives. To compile the data, official member webpages were consulted and a search was conducted for member Twitter accounts through the Twitter search engine. Using Google, the results were then crosschecked to verify that the correct and most current Twitter feed was captured.

The demographic characteristics of members' districts may influence their decision concerning Twitter adoption. Again, we theorize that members should be more likely to believe the benefits of social media outweigh the costs if they perceive that their target audience will also participate. We include measures such as district wealth (median household income), district education (the proportion of district residents who are high school and college graduates), and urbanization (the proportion of the district considered to be “urban”).Footnote ¹ To test the influence of reelection pressures, we include the proportion of the total vote a member received in his or her last election. Finally, to test whether Twitter may serve as a way to keep in touch with constituents during long work periods in which travel back to the state or district might be difficult or impossible, we also include a measure of the log distance from Washington, DC, to the member's home state.Footnote ²

As stated previously, social media may serve as an outlet for members who want to appeal to a national constituency, or a specific subpopulation that transcends district or state boundaries. To test whether members may adopt Twitter to create a national constituency based on certain demographic characteristics, we include dummy variables for the member's race and gender. Likewise, given that a minority party member or a member with an extreme ideological position may also adopt Twitter to engage with a broader political constituency that matches his or her beliefs, we include a dummy variable for party membership in model 1.Footnote ³ Although the member's party may provide some insight into the relationship between a member's ideology and Twitter registration, it is a blunt measure at best. To better measure ideological extremeness, we include a direct measure of ideology in model 2. Ideology is measured as the member's common space DW-NOMINATE score for the 111th Congress; ideological extremeness is calculated as the square of ideology.Footnote ⁴

Table 1 presents the chamber and party affiliation of the 313 members registered with Twitter. Eighty percent of those registered were representatives, and registration rates in the Senate (59.1%) were slightly higher than in the House (56.8%). As illustrated in table 1, a much higher percentage of total Republican members created a Twitter account (68.1%) than total Democratic members (49.5%).

Table 1

Percentage of Members of the 111th Congress Registered with Twitter, by Chamber and Party

Numbers do not add to 535 because of members who served only part of the 111th Congress.

Given that Democrats were the majority party in the House during the 111th Congress, one might expect a disproportionate number of Democrats comprising the total number of members with Twitter accounts. However, this is not the case. Of those registered, 50.1% were Democrats and 49.2% were Republicans. This difference suggests that Republicans, as members of the minority party, are more likely to create Twitter accounts than Democrats, providing some initial support for the theory that members adopt Twitter to reach a broader constituency.

The relationship between these variables and a member's decision to register a Twitter accountFootnote ⁵ was estimated using two logit models, using robust standard errors clustered on state.Footnote ⁶ Summary statistics for the data used to estimate these models is in table 2.

Table 2

Summary Statistics of Data Used to Estimate Models 1 and 2 (N = 540)

ANALYSIS

In model 1 and model 2 (table 3), we find support for the theory that members adopt Twitter to represent a broader constituency, with some limited support for the theory that adoption is a reflection of district characteristics. Additionally, we found no evidence that Twitter adoption was associated with previous reelection percentages.

Table 3

Estimating Member Adoption of Twitter as of December 31, 2010

Significance in a two-tailed test:

*** p < 0.01,

** p < 0.05,

* p < 0.1;

robust standard errors, clustered on state, in parentheses. N = 540.

In both models, we find that the more urban a member's district or state, the more likely that member is to adopt Twitter. This suggests that members adopt Twitter when they believe that Twitter usage is widespread among their constituency, providing some limited support for the theory that members adopt Twitter to gauge district opinions and policy preferences. All else being equal,Footnote ⁷ a member who represents a district or state where 35% of the land area is considered to be urban has a 0.09 lower probabilityFootnote ⁸ of having a Twitter account than a member who represents a district or state in which 55% of the land area is considered to be urban and a 0.32 lower probability than a member representing an entirely urban district (100%).Footnote ⁹ This theory is also supported by the significance of log distance to district, although this variable is only statistically significant in model 1.Footnote ¹⁰

Stronger support for an alternate theory of Twitter adoption, the desire to represent a broader constituency, is found in both models. Although a member's race and gender are not significant predictors of Twitter adoption, party affiliation, political ideology, and ideological extremeness are all significant. In model 1, the party of the member was the most influential factor in predicting whether that member would register an account with Twitter. As predicted by the “broader constituency” theory, those in the minority party (Republicans) were much more likely to have adopted Twitter than their Democratic counterparts. All else being equal, Republican members had a 0.21 higher probability of having a Twitter account than Democratic members.

In model 2, we find that ideology and ideological extremeness have a statistically and substantively significant positive effect on the probability that a member will adopt Twitter.Footnote ¹¹ Given that one variable (ideological extremeness) is the square of the other (ideology), the two cannot be interpreted separately. Figure 1 illustrates changes in the predicted probability of having a Twitter account over the range of ideology scores found in the model. The solid black line indicates the predicted probability of having a registered Twitter account as of December 31, 2010. The dashed black lines indicate the 95% confidence intervals for the predicted probability across the range of ideology scores.

Figure 1

Predicted Probability of Twitter Registration across Ideology

Specifically, figure 1 demonstrates that members with extremely conservative ideology scores are the most likely to have registered a Twitter account. For example, the model predicted that, holding all else equal, the average Republican member (ideology = 0.447) has a 0.30 probability of adopting Twitter. The most conservative member (ideology = 0.99) has a 0.78 probability of adopting Twitter, a difference of 0.48. This same extremely conservative member has a 0.62 higher probability of adopting Twitter than the most moderate member (ideology = 0).

Figure 1 additionally indicates that members with extremely liberal scores are more likely to use Twitter than members with less extreme liberal scores. For example, holding all else equal, the average Democratic member (ideology = −0.378) has a 0.14 lower probability of adopting Twitter as a communication tool than a member with the most extreme liberal score in the data (ideology = −0.743). This same extremely liberal member has a 0.15 lower probability of adopting Twitter than a member with the most moderate ideology score (ideology = 0).

As expected, the age of the member was also found to be significant in both models. Unsurprisingly, the older the member or senator, the less likely he or she is to have a Twitter account. For example, a 65-year-old member has a 0.04 lower probability of having a Twitter account than a member 10 years his or her junior and a 0.09 lower probability of having a Twitter account than a member 20 years his or her junior. Lastly, seniority is not a good predictor of Twitter adoption. This is in keeping with previous studies on congressional political communication strategies, which show that more senior members of Congress have greater access to traditional press coverage than those with less tenure.

ACKNOWLEDGMENTS

This paper reflects the views of the authors and does not necessarily reflect the views of the Congressional Research Service or the Library of Congress. A previous version of this paper was presented at the 2010 annual meeting of the Midwest Political Science Association. The authors would like to thank Jessica C. Gerrity and the anonymous reviewers for their helpful comments.