The Chapel Hill American Media Project (CHAMP), which I helped found and now coordinate, is an effort by graduate students and faculty at UNC to collect and analyze data about political media in the United States, including who provides political coverage, what topics receive attention, and how political issues are framed. Currently, CHAMP consists of television transcripts from ABC, CBS, NBC, CNN, Fox, and MSNBC from 2001 through 2014, which amounts to approximately 150 shows and tens of thousands of episodes. With this data set, I and my collaborators explore how what is said in the media influence policy, how power dynamic influence whose voice is heard, and how differential coverage by network and show influences opinions and attitudes. 

Below  I summarize some of this research agenda and related efforts. Please contact me with any questions or requests for working papers at:

Working Papers

Pamela J. Conover, Kelsey Shoub, and Leah Christiani. “Ladies on Television News: How Gender Influences the Use of Powerful Language on Television News.” Paper presented at the Midwest Political Science Association Annual Meeting. Chicago, IL, April 2017.

Although women comprise over half of the United States population, politically they resemble a minority group. But when women are seen in leadership roles, they can affect society in important ways. Women on national television news are both substantively and symbolically important. However, if women do not speak powerfully, they may further diminish their presence by consistently reinforcing stereotypes that women are not strong leaders. Many researchers have argued that men and women differ in their language style: women are more likely than men to use tentative or less powerful language, such as expressions of uncertainty or hedges. However, recent research highlights the importance of the context as a possible determinant of whether and how women speak. To test for differences in the use of tentative and assertive language between women and men, we will use data from the Chapel Hill American Media Project—a collection of television transcripts and associated metadata from 2000-2014 that includes daytime and evening news and talk shows.  Using a matching analysis, we test whether men and women use hedge language at differential rates in different contexts. We find that women on average talk less than men, women in positions of authority hedge less than other women, and politicians use hedge language at a higher rate than journalists.

Pamela J. Conover, Kelsey Shoub, Amy Sentementes, and Leah Christiani. “How Powerful a Presence? Woman as Sources in Television News” Paper presented at the American Political Science Association Annual Meeting. Philadelphia, PA, September 2016.

It is both symbolically and substantively important that women are equally represented as sources in the national news on television. When women appear as sources in the national news they serve as visible women leaders for other women, as well as substantively effecting the agenda and framing of the news. Using data from the Chapel Hill America Media Project—a collection of text transcripts and associated metadata from 2000-2014 that includes daytime and evening news and talk shows on ABC, CBS, CNN, FOX News, NBC, and MSNBC—we provide a more comprehensive analysis of women’s presence as sources in the national news. Then by focusing on “speaking time”, we assess the (potential) influence that women sources have compared to men. Overall, we find the women are underrepresented as sources. And when they do appear on the news, women have less influence than men, as measured by speaking time.

Anthony Chergosky, John Lovett, Kelsey Shoub, and Ryan Williams. “Supreme Court Headline News: How Television Media Affects Court Public Opinion.” Paper presented at the American Political Science Association Annual Conference. San Francisco, CA, August 31-September 2, 2017.

Non-elected and lifetime-tenured, the justices of the United States Supreme Court are insulated from direct public pressure and free from navigating regular election cycles. Despite this, decades of scholarship indicate that the Supreme Court heeds the prevailing winds of public opinion, issuing rulings that move in ideological concert with macro opinion. In this paper, we seek to address whether the ideological leans of television news programs broadly defined (i.e. nightly news shows, pundit shows, and magazine programming) differ from one another in their coverage of the Supreme Court and its decisions. To evaluate our theory, we divide the empirical test into two steps. First, we demonstrate that outlets with different ideological leanings cover the Court differently. To do so, we use television news transcripts compiled by the Chapel Hill American Media Project (CHAMP). Then, we predict opinion of the Court using survey data from Pew Research Center that includes questions both on opinion of the Court and where respondents get their news. We show that different outlets cover the Court differently, and the difference in coverage in turn influences the opinions of their viewers. Demonstrating that different types of media cover the Court differently is important for understanding how public opinion connects to Court outcomes.

Under Review

Kelsey Shoub and John Cluverius. “Interest Niches and Policy Bandwagons in Online Outside Lobbying” Paper presented at the American Political Science Association Annual Meeting Political Communication Pre-Conference. Philadelphia, PA, August 2016. 

While participation and trust in government has decreased over time, the number of citizens contacting Members of Congress has increased over time. Additionally, the number of registered lobbyists have decreased overtime. Does this tension indicate that there is a closer link between the people and their representatives; or does it indicate that we are observing a shift in tactics by lobbyists from direct contact to grassroots lobbyists? To inform these questions, we ask: why do citizens choose to comment on bills and which bills do they comment on? We propose and demonstrate that citizens are not randomly voicing their opinions nor are they responding to increased coverage by the media. Rather, interest groups are activating members of the public who share their views and feeding them language. To test this theory, we use a unique data set of commenting patterns by constituents and interest groups to test this relationship. We find that those opposing bills seem to be better organized and are more likely to be activated by groups than those supporting legislation.


“Population Dynamics and Representation.” In David Lowery, Virginia Gray, and Darren Halpin, eds., The Organization Ecology of Interest Communities: Assessment and Agenda. London: Palgrave, 2015, pp 203–224. (Frank R. Baumgartner and Kelsey Shoub)