True OECD Voter


  1. General information
  2. Sources
  3. Harmonization
  4. Outputs

General information

The data has been assembled primarily between 2014 and 2019, with the purpose of using it for my doctoral dissertation at the Central European University, Budapest. The original inspiration for the data came from the True European Voter (TEV) project. I adopt here the same coding scheme as the TEV, with the goal of ultimately merging the two data sources.


Two general types of data sources have been used:

  1. For the longest time series, original election surveys were downloaded from country statistical offices or data archiving organizations, and manually merged. This was the case for 18 countries: Australia (1967-2019), Belgium (1991-2019), Canada (1965-2019), Denmark (1971-2019), Finland (1973-2019), France (1958-2017), Germany (1953-2017), Iceland (1983-2017), Israel (1969-2020), Italy (1968-2018), Netherlands (1971-2017), New Zealand (1981-2020), Norway (1957-2017), Spain (1989-2019), Sweden (1956-2018), Switzerland (1971-2019), United Kingdom (1966-2019), and the United States (1948-2016).
  2. For the remaining cases, where the original surveys were not publicly available at the time, data was sourced from the Comparative Study of Electoral Systems or the World Values Survey and harmonized as much as possible. This was the case for 19 countries: Austria (1990-2017), Brazil (2002-2018), Chile (1999-2017), Czechia (1996-2013), Greece (2009-2015), Hungary (1998-2018), Ireland (2002-2016), Japan (1981-2017), Latvia (2010-2014), Lithuania (1997 and 2016), Mexico (1997-2015), Peru (2000-2016), Poland (1997-2011), Portugal (2002-2019), Romania (1996-2014), Slovakia (2010-2020), Slovenia (1996-2011), South Korea (2000-2016), and Turkey (2011-2018).

Collectively, these 37 countries comprise the True OECD Voter (TOV) project. In the interest of maintaining regional coverage, a few countries who were not technically part of the OECD in 2019 were also included in the data.


The primary focus of the data has been to explain the turnout decision over time and across countries. Because of this, information on reported or intended turnout is a key indicator in the data.

In addition to it, the data contains information on a basic set of socio-demographics (age, gender, education, income, marital status), as well as political interest (a marker of political engagement), and union membership.

Many decisions, large and small, had to be taken to allow for the harmonization of the data. If you’re interested in what these were, please check the technical notes in Sections 1 and 2 of the Supplementary Information file of this working paper.


Though active development on the data has stopped, it is still being used for research projects:

  1. A recent co-authored article on the impact of the electoral breakthrough of populist parties on the political interest of key population subgroups
    1. Supplementary information
    2. Replication materials
  2. An unpublished manuscript offering a systematic view on the relative importance of institutional and macro-economic factors on turnout gaps over time
  3. An unpublished manuscript on the welfare state characteristics which shape turnout gaps in OECD countries

If you want to hear more about the project, I’d be happy to share more information, along with the code files to generate the data!