Democratic Electoral Systems Around the World, 1946-2000

Matt Golder


Electoral institutions are often considered an important independent or dependent variable in many areas of political science. However, problems with data availability have often artificially constrained much of our empirical research to particular geographical regions or time periods. Although there have been several recent attempts to remedy this situation, there is still no single database addressing electoral institutions across the world in an entirely consistent and comparative manner. The database provided here seeks to build on this recent work by describing some of the more important electoral institutions used in all legislative and presidential elections during democratic periods in 199 countries between 1946 (or independence) and 2000. This amounts to a total of 867 legislative elections to national lower houses and 294 presidential elections. The dataset covers a wide range of institutional features including regime type, the electoral formula, the average and median district magnitude, the number of constituencies and upper tier seats, assembly size etc.. It also includes information on the effective number of electoral and legislative parties. More detailed information concerning the variables included in the dataset can be found in the codebook listed below. This dataset is a work in progress and I hope to fill in missing information and include additional variables as they become available. Those scholars who wish to use this dataset should cite the following article:

"Democratic Electoral Systems Around the World, 1946-2000" Forthcoming in Electoral Studies.


Abstract:

Electoral institutions are often considered an important independent or dependent variable in many areas of political science. However, problems with data availability have often artificially constrained much of our empirical research to particular geographical regions or time periods. This article describes a new data set that covers the electoral institutions used in all of the democratic legislative and presidential elections in 199 countries between 1946 (or independence) and 2000. This amounts to 867 legislative elections and 294 presidential elections. A clear and consistent classification of the electoral institutions used in these elections is followed by a concise geographical and temporal analysis. The worldwide focus of the data set reveals several striking patterns. For example, there have been almost as many elections under dictatorship as there have been under democracy. Other patterns include the fact that presidential regimes nearly always employ proportional electoral formulas, absolute majority rule has become the worldwide norm for electing presidents, majoritarian electoral systems account for the same percentage of legislative elections as they did in the 1950s, and non-majoritarian systems have become more complex due to the increasing use of multiple tiers and mixed electoral formulas.

For a slightly longer version of this paper with more detailed references and citations, click here.


Data

The codebook is a useful place to start. I provide a detailed description of the variables in the dataset and how they were constructed. Summary tables provide additional information. First, I explicitly indicate those cases where seats or electoral districts have been omitted from the dataset (typically, seats have been omitted if they were not elected). Second, I list the number of legislative and presidential elections that have occurred under dictatorship in each country. Third, I provide a list of the countries that appear in each geographical region. Fourth, I provide a table that lists the years in which countries had parliamentary, presidential, or mixed regimes. For presidential and mixed regimes, I indicate how the president was elected (plurality, qualified majority rule, absolute majority rule, electoral college, or STV). Fifth, I provide descriptive statistics on each electoral system type, starting with majoritarian systems, proportional systems, multi-tier systems and finally mixed systems. The codebook finishes with an exhaustive list of the sources consulted as I collected the data. Special thanks go to all those individuals and organizations that provided me with information.

Matt Golder's Dataverse

Codebook

The data can be found in EXCEL format. The EXCEL data typically provides information regarding the source for each observation along with any additional information as a comment (the red triangles in the spreadsheet).

There are two main datasets.

Legislative Elections

This provides information on all 867 legislative elections that have occurred in democratic periods in 199 countries between 1946
(or independence) and 2000.
[EXCEL]

Presidential Elections

This provides information on all 294 presidential elections that have occurred in democratic periods in 199 countries between 1946
(or independence) and 2000.
[EXCEL]

Several people have asked me to put together a dataset in country-year format. The dataset below is in country-year format. Note that I recommend using the two datasets above rather than the country-year formatted dataset. This is for several reasons. First, there are several countries that had two elections (legislative or presidential) in the same year (these are listed in the codebook). As a result, it is not possible to provide data for both elections that occurred in the same year in the country-year dataset. In those cases where there were two elections, I present the figures from the second election. Information on the first election can be found in the 'comments' if it differs from what is already shown. Second, my primary research goal in collecting this dataset did not require me to collect this information in country-year format. As a result, I have put the country-year dataset together at the last moment. There should be no discrepancies between the two datasets listed above and the dataset that is in country-year format. However, if there is a discrepancy you should believe the two datasets shown above since I have checked these more thoroughly. Please also contact me if you find any discrepancies so that I can resolve them. Third, it is important as always to look at the spreadsheet and check that it is coded as you think from reading the codebook. For example, electoral institutions in the country-year dataset are coded one for the election year, zero otherwise. This may or may not be the format you desire. You can of course use the two datasets shown above and merge them into a country-year dataset of your own formating - just remember that there are several countries that had two elections in the same year.

Country-Year Format


Missing Information

In order to construct this dataset, numerous sources were used (all listed in the codebook). I have gone to great lengths to ensure that the information in the dataset is as complete as possible. Nonetheless, I am still missing information on certain variables for particular countries and elections. This missing information is listed in a separate document below in the hope that someone may be able to provide me with this information. If you have any of this information, please email it to me at mrg217@nyu.edu with a citation for the source so that I can verify its authenticity. I will then add you to the growing list of people that have provided data. This list of people can be seen in the codebook.

Missing Information


Incorrect Data

Despite my best efforts, it may be the case that I have made a mistake in coding particular observations. This may be because I have found multiple sources with contradictory data for a given observation. However, it may also be because I have made a human error in inputting the data. In either case, I would greatly appreciate knowing if anyone finds an error. Please email me at mrg217@nyu.edu with a citation for the source of your information so that I can verify its authenticity.


Elections and Electoral Systems

Home Page - Personal Page