Constituent connections: senators’ reputation building in the age of social media

ABSTRACT

Twitter is changing strategic messaging in the U.S. Senate. Senators are using Twitter to frame their political brand for constituents, fostering a new digital dialog with constituents. I propose a constituent-driven theory of strategic messaging where senators curate a reputation on Twitter that matches their perceived expectations of their primary constituency. Representation on social media challenges what we know about senators’ institutionally and politically constrained behavior by analyzing them in a new media climate where individual discretion is high and the costs are low. Using a unique dataset of more than 180,000 hand-coded tweets by senators, I show that senators develop two types of digital constituent relationships – an issue-oriented, national reputation versus traditional outreach to geographic constituents. Senators with issue-based constituencies prioritize policy, conveying an issue-driven style of representation; however, senators with tepid electoral futures pair their policy rhetoric with state-based issues or local concerns. These findings expand the scope of existing theories on congressional communication and link the technological shifts in Congress to information senators use to build relationships with voters.

KEYWORDS:

• Congress
• Twitter
• agendas
• representation

Disclosure statement

The author has no conflicts of interest to disclose.

Notes

1. Murray in the News. https://www.murray.senate.gov/public/index.cfm/mobile/pattyinnews?ID=8BEED324-8CAD-4515-B9DA-A1E67FAB32E3

2. Table of 2013 and 2015 senators is included in the Appendix.

3. All policy tweets were hand-coded by a graduate student coder, and a 6,000-observation sample was double-coded by experienced student coders for reliability measures. Student double-coding yielded the following inter-coder reliability statistics for policy issues: percentage agreement = 87.4%, Cohen’s kappa = 85.6%, Krippendorff’s alpha = 85.6%.

4. Coding guidelines for policy coding is included in Appendix.

5. Because priorities are not mutually exclusive, percent totals do not equal 100%.

6. This model specification is appropriate because, like a beta regression, it can account for the skewness of the dependent variable, but it does not assume the dependent variable follows a beta-distribution (Petrocik, 1996). Variations on the logit model are common practice with explanatory variables that are attributes of individuals, in this case, U.S. senators.

7. A model including partisan extremity includes no significant effects but is included in the Appendix.

8. Each regression is also run with Ideology as a proxy for party and the results remain consistent.

Additional information

Notes on contributors

Annelise Russell

Annelise Russell is an Assistant Professor of Public Policy at the University of Kentucky. She is also a faculty associate of the U.S. Policy Agendas Project. Dr. Russell’s research interests include questions about how policymakers communicate their agendas and the role of the media, particularly social media, in the political process. Much of her research is on congressional decision-making and communication, including an active research agenda in the intersection of social media and political institutions.