A Brief History of the Comparative National Elections Project

Origins

The seeds for what became the Comparative National Elections Project were first sown in conversations among political science colleagues Paul Beck, Russ Dalton, and Scott Flanagan at Florida State University in the mid-1980s. They lamented the absence of cross-national election studies with comparable survey questions and theoretical frames and aspired to design truly cross-national surveys of voting behavior.

The First Wave (1990–1993)

These seeds germinated in the ensuing years into study designs that drew upon the early Columbia University community voting studies of the 1940s–1950s to focus on the flow of information to voters through personal discussion networks; newspaper, television, radio (and later internet) media; and organizational intermediaries (including political parties) beyond the well-established variables from University of Michigan American national election studies.

The first wave of surveys involved Germany, the United States, Great Britain, and Japan. Leading voting behavior scholars of each of the four original countries (Hiroshi Akuto, Ken'ichi Ikeda, and Brad Richardson for Japan; Max Kaase, Hans-Dieter Klingemann, and Manfred Kuechler for Germany; Bob Huckfeldt for the U.S.; and John Curtice and Anthony Heath for Great Britain) joined the Florida State quartet in the late 1980s for group meetings in Wakulla Springs, Florida (sponsored by Florida State University), Columbus, Ohio (sponsored by Ohio State University), and Bloomington, Indiana (sponsored by Indiana University) to plan the design and questionnaire of the initial four comparative national surveys. They covered the 1990 elections in a newly merged East and West Germany, the 1992 U.S. and British elections, and the 1993 Japanese election.

The German and U.S. election studies adopted ambitious designs involving interviews with snowball samples of interpersonal discussion partners, analyses of newspaper and television campaign content, and self-reported contacts by political parties and candidates. The British survey involved a leave-behind questionnaire to measure core CNEP items. Although funding and fieldwork limitations of later surveys proved that these designs were too ambitious, they provided the full model for CNEP conceptualizations as described by Beck, Dalton, and Huckfeldt in a 1988 essay outlining the theoretical bases for the project.

Expansion

In the ensuing decade, the by-then titled Comparative National Elections Project (CNEP) spread to other country settings. Dick Gunther and José Ramón Montero oversaw a 1993 Spanish survey and recruited surveys in other Spanish-speaking countries (Chile, Uruguay) and beyond (Bulgaria, Greece, Hong Kong, Hungary, Indonesia, and Italy) into the project. By then, Dick Gunther had joined Paul Beck (now at Ohio State University) as co-coordinators of CNEP, with Gunther undertaking the homogenization of the comparable items in each new survey and the combining of all existing surveys into a merged file.

By the mid-2000s, repeat elections in some of the earlier countries and new countries in Europe, Africa, and Asia brought another 13 surveys into the project. With each new phase, research themes were expanded in the survey questionnaires to address the most recent developments in elections, especially the rise of new democracies, and to allow comparisons with the more traditional ones.

Distinguishing Features

Several features have distinguished CNEP through its four decades:

  • A common core questionnaire. The surveys have been built around a common core questionnaire of cross-nationally comparable items, standard in all surveys, involving demographic characteristics and traditional voting behavior as well as emerging dimensions of electoral competition such as populism, polarization, democratic attitudes, threats to democracy, and party system fragmentation. As new items have had to be accommodated, some of the older items have given way to keep the common core questionnaires of manageable length while maintaining continuity with earlier surveys.
     
  • A comparative data archive. CNEP produces both cross-nationally comparable voter surveys in single elections and a growing archive of such surveys since 1990. The national surveys have been harmonized around the standard common core items, then combined into a single data set of voters in 75 national elections that has merged all of the CNEP items. The merged file facilitates cross-national analysis of voting behavior across almost four decades. It also supports analysis within the 17 countries that contain surveys for more than a single election: the U.S. (7 elections); Indonesia and Spain (5); Chile, Colombia, Germany, Greece, Italy, Mexico, and South Africa (4); Hong Kong, Portugal, Taiwan (3); Argentina, Britain, Hungary, and Uruguay (2).
     
  • A collaborative network. CNEP is the product of collaboration among leading voting behavior scholars in their home countries whose surveys have reported on the election studied using comparable CNEP core questions. They typically brought their own funding to the project, including grants from their national science foundations (e.g., in the U.S., Britain, France, Germany, Japan, Italy, Spain, and Taiwan), private foundations, and universities.
     
  • An international scholarly community. A CNEP community of voting behavior scholars has developed through a series of conferences over the course of the project. From the beginning, the conferences have focused on development and maintenance of the common core questionnaire, the integration of new themes to reflect new electoral developments and an expanded group of countries, and the sharing of research results. Conferences mostly have been funded by their host universities in, e.g., Chile, Colombia, Germany, Hungary, Indonesia, Italy, Japan, Mexico, Portugal, South Africa, Spain, and the U.S.
     
  • Evolving data-collection methods. As election polling methods have evolved from face-to-face to telephone to the internet, CNEP data-collection methods too have changed following this same trajectory. Internet surveys are now most common, though they are not feasible in some of the less developed countries.
     
  • A growing body of research. Numerous scholars have used CNEP data in comparing voting behavior across elections and in analyzing the survey results in their own countries. These results are published in Democracy, Intermediation, and Voting on Four Continents (2007), Voting in Old and New Democracies (2016), and various refereed journals.
     

CNEP Today

The aspirations of its founders in the mid-1980s have been realized. CNEP has been a leader in comparative studies of voting behavior across the world. Through April 2026, it includes 76 cases covering 75 elections in 30 countries and Hong Kong that are available for analysis, with more to come soon; most recently, 75 of these surveys have been merged into one data file of over 125,000 respondents. The study of voting behavior has become a truly comparative enterprise.
 

A New Chapter: CNEP at Universitat Pompeu Fabra

2025–2026 is a time of transition for CNEP. After almost 40 years as co-coordinators of the project, Paul Beck and Dick Gunther are retiring from their leadership role. Hosting of the project has relocated from the Mershon Center at Ohio State, which has generously supported CNEP in so many significant ways almost from its beginning, to Universitat Pompeu Fabra (UPF) in Barcelona. Mariano Torcal (at UPF) and Bob Mattes (at the University of Strathclyde) have assumed the roles of Principal Investigators. This transition is complete. A new website is now available through UPF ([email protected]) and the harmonization and merging of recent surveys has been undertaken by Mariano and his expert staff there. A total of 76 CNEP-harmonized surveys are now archived and more soon will be in the queue to be processed. Among the recent accomplishments:

  • Merge 72 incorporated surveys from the USA 2024, Indonesia 2024, Germany 2025, and Portugal 2025 elections.
     
  • Merge 73 restored the original Japan 1993 survey to the archive, with full documentation and the original dataset recovered.
     
  • Merge 75 restored the 1990 German data from the original CNEP–GESIS collaboration, split into two cases — GE90W (West Germany 1990) and GE90E (East Germany 1990) — with full documentation.
     
  • Merge 76 incorporated a new country to our data archive: The Netherlands 2025.
     
  • A comprehensive dataset of standardized affective and ideological polarization indexes, based on the formulas of Markus Wagner and Russell Dalton, has been released alongside the archive, covering up to 66 cases.

The current research agenda is organized around three priority areas: political polarization, political intermediation, and democratic support. The project is also exploring methodological innovations, including the transition toward three-wave panel surveys and the incorporation of conjoint experiments to strengthen causal inference.

Mariano and Bob are developing grant proposals to secure more central funding for surveys and data processing in the CNEP tradition. They have made impressive progress in making the transition and improving CNEP going forward.

17/04/2026

Paul A. Beck