Campaign Mapping Project

People analyzing a map

Campaign Mapping Project

The Campaign Mapping Project is a multi-year attempt to assemble campaign materials produced during presidential elections between 1948 and 2020, as a guide to understanding the direction of American politics. The CMP's assumption has been that past research has too often lunged from campaign to campaign with few consistent questions being asked and with few broad findings emerging as a result.

The Campaign Mapping Project is a multi-year attempt to assemble campaign materials produced during presidential elections between 1948 and 2016, as a guide to understanding the direction of American politics. The CMP's assumption has been that past research has too often lunged from campaign to campaign with few consistent questions being asked and with few broad findings emerging as a result. 

To gain greater perspective the Project has four goals:

1) To develop a comprehensive view of campaigns: By bringing together many different voices of politics -- the people, the candidates, and the press -- the Project seeks a broad understanding of the nation's quadrennial dialogue.

2) To develop an objective view of campaigns: Because previous campaign commentary has been based on hunch and invective, quantitative analysis has been used here to add rigor to an area of study in need of it.

3) To develop a normative view of campaigns: No campaign can be understood in isolation. By reaching across sixteen elections for insights, the CMP sought out genuinely cultural understandings of American politics.

4) To develop a communicative view of campaigns: Although much previous campaign research has prized attitudinal and economic data, the CMP took this position: if democracy is a dialogue, someone must speak, someone must listen, and someone must evaluate what was said. With political communications now accounting for 60% of all campaign expenditures, and with more and more persons now being joined together via the "old" and "new" media, there has never been a greater need to discover who said what to whom and how. The Campaign Mapping Project helps meet that need.

The Project began under the direction of Professor Roderick Hart of the University of Texas and Professor Kathleen Hall Jamieson of the University of Pennsylvania in June of 1995. The Project received initial funding from the Ford Foundation and the Carnegie Foundation of New York and now continues under the direction of Professor Hart, with support from the Shivers Centennial Chair, the Denius Entertainment Studies Fund, and the Annette Strauss Institute.

The first several years of the Project were devoted to data collection. The latter includes speeches, advertisements, debates, broadcast news stories, newspaper accounts, and letters-to-the-editor generated during seventy years worth of general election seasons (late August through election day).

Speeches were obtained from presidential libraries when possible, but also from state and regional archives, and, in some cases, from groups and individuals.

Political advertising is typically fugitive stuff, but the Project benefited considerably from the fact that Kathleen Jamieson had been collecting these materials for more than twenty years.  More recently, ads and their transcripts have been easily accessible on the Web.

Presidential debates were more easily obtained from Internet sources as was newspaper coverage, since it existed on microfilm even for the early years.

Broadcast news coverage was hard to obtain prior to 1968 (when Vanderbilt University's archives first opened) but considerably easier thereafter.

But hardest of all were the letters-to-the-editor serving as the People's Voice in the Project. Written by ordinary people and printed in small newspapers, the letter existed in city libraries and newspaper morgues in twelve small cities scattered throughout the United States.

The Project's database is not exhaustive but it does contain a fairly representative sampling of campaign discourse as follows.

Speech segments (n=9,665): formal remarks given by the major candidates between late July and early November of the campaign year, including nationally televised address as well as local stump speeches.

Debate segments (n=1,659): all presidential debates between 1960 and the present, with each debate being segmented by speaker prior to analysis.

Advertising (n=988): a sampling of major party and independent party ads broadcast over television. Years include 1976-2016.

Print Coverage (n=22,295): feature and non-feature stories from the New York Times, Washington Post, Christian Science Monitor, Atlanta Constitution, Chicago Tribune, Los Angeles Times, as well as AP, UPI syndicate stories.

Broadcast Coverage (n = 4.423): campaign stories from the nightly news from NBC, CBS, ABC, PBS and, more recently, CNN.

Letters-to the Editor (n=11,276): letters written to the editors of twelve small-city newspapers between 1948 and 2008. The cities represented (all of which had populations of approximately 100,000 persons at the mid-point of this study) include a cross-section of political attitudes and demographic characteristics. The papers are regionally distributed throughout the United States. The are: Billings Gazette, Duluth News Tribune, Lake Charles American Press, Springfield (OH) News Sun, St. Joseph News-Press, The Fall River Herald News, The Provo Daily Herald, The Roanoke Times, The Salinas Californian, The Trenton Times, Utica Observer Dispatch, Wichita Falls Times Record News.

All of these materials were keyboarded and introduced into a computerized database. The Project employed some twenty graduate assistants at the two institutions and also depended on the good will of colleagues and archivists throughout the United States. From here on the Project proceeded in two different directions. While the team at the University of Texas studied language patterns, the Annenberg Public Policy Center team looked at the arguments put forth during campaigns.

Here at the University of Texas, the data was analyzed for language patterns using the DICTION program. Because DICTION breaks a text into smaller segments when analyzing it, the Texas version of the CMP consists of a series of 500 word segments.


2020 Roderick P. Hart: Trump and Us: What He Says and Why People Listen (Communication, Society and Politics)

2018 Sharon E. Jarvis and Soo-Hye Han: Votes That Count and Voters Who Don’t

2018 Roderick P. Hart: Civic Hope: How Ordinary Americans Keep Democracy Alive 

2013 Roderick P. Hart, Jay P. Childers, and Colene J. Lind: Political Tone: How Leaders Talk and Why

2005 Sharon Jarvis: The Talk of the Party: Political Labels, Symbolic Capital & American Life

2004 Roderick P. Hart and co-authors: Political Keywords: Using Language that Uses Us 

2000 Kathleen Hall Jamieson: Everything You Think You Know About Politics (And Why You're Wrong) 

2000 Roderick P. Hart: Campaign Talk: Why Elections Are Good for Us. 

Additionally, ten doctoral dissertations, fifty published articles and 100 convention papers have emerged from the Project.

The Campaign Mapping Project began with the assumption that some campaigns fail and some succeed. The 1980 campaign succeeded when Ronald Regan and Jimmy Carter gave the people a clear, sharp choice, and the 1964 campaign succeeded when Lyndon Johnson and Barry Goldwater did the same thing.

But other campaigns fail. Campaigns fail when they do not properly forecast for the people what the successful candidate will do once in office. Campaigns fail when the do not give the winner a real national mandate, an emotionally committed citizenry. In 1976 Jimmy Carter focused on issues of personal morality during his run for office only to have the cold facts of inflation, unemployment, and energy shortages doom his presidency four years later. And even the election of 1992 was problematic with only 43% of the American people backing Mr. Clinton, in part because of Ross Perot's candidacy but also because "talk show democracy" obviated the sharp exchange with the working press a polity builds to understand the issues and then to build consensus.

So history teaches us that some campaigns serve the nation and others do not. Which is which and why? Obviously no campaign can predict the future completely. Each president will be confronted by the unexpected from his first day in office, but that is not to say that campaigns cannot be improved. To do so requires evidence. Providing some of that evidence has been the job of the Campaign Mapping Project.