Historical V-Dem

About Historical V-Dem

The Historical V-Dem project codes numerous indicators of democracy and other institutional features, all the way from the French Revolution to the early twentieth century. More specifically, Historical V-Dem covers about 80 polities and extends almost 200 V-Dem indicators as far back as 1789, including about 130 of V-Dem’s expert-coded indicators.

The integration of Historical V-Dem with V-Dem means that many countries have data coverage on numerous V-Dem indicators from 1789 to the present. These include all major countries with continuous lineages as independent states from the 18th or 19th centuries, but also some major colonies such as India and Indonesia. Historical V-Dem also contains several extra, historical polities that have ceased to exist, such as Bavaria and the Two Sicilies.

Further, Historical V-Dem introduces several new indicators – including 20 expert-coded and many more coded by Research Assistants – on various institutional features, some of which were particularly relevant for 19th century polities. To name two examples, Historical V-Dem includes extensive sets of new indicators on the state and features of the bureaucracy as well as on political regimes and their supporting groups.

Methodology

Historical V-Dem includes two types of indicators – more objective indicators coded by Research Assistants (termed “A indicators”, following the convention in V-Dem) and more evaluative indicators (“C indicators”) coded by historical country experts.

The methodology of Historical V-Dem closely follows that of V-Dem, more generally, but with some exceptions: The most important pertains to the type and number of expert coders. Due to the lower number of true experts on the politics of a country in the 18th and 19th century, compared to in the 20th and 21st, there are fewer “Historical Country Experts” per country.

Given this, Historical V-Dem also followed a different recruitment and compensation strategy, carefully selecting the best available historical experts, and remunerating them for working closely with the coding over an expanded time period. Nonetheless, the lower number of historical coders typically means that the 18th and 19th century scores are associated with somewhat larger uncertainty estimates (as provided by the V-Dem Measurement Model, see below) than the 20th and 21st century scores.

Yet, several measures have been taken to ensure the comparability of V-Dem and Historical V-Dem scores, and that the data from 1789 until today constitute consistent time series. First, Historical V-Dem experts and V-Dem experts code an overlapping period of 20 years (typically 1900-1920) for “their” countries or that country’s successor state (e.g. Germany for Bavaria). Second, all historical experts coded a set of identical anchoring vignettes, in the form of brief descriptions of hypothetical country contexts, for all relevant indicators. Third, many historical experts coded three additional countries for the first year after 1900 with an election. All these measures provide valuable information, which is leveraged by the V-Dem measurement model to make scores comparable across countries and time, and provide uncertainty estimates.

Funding and Organization

Historical V-Dem has mainly been funded through three larger research grants from the Norwegian (Young Research Talent Grant pnr 240505; PI Carl Henrik Knutsen) and Swedish (grant no 2014-1283; PI Jan Teorell) Research Councils, and Demscore (grant no 2021-00162).

The Historical V-Dem data collection has been managed from Lund University, the University of Oslo, and Stockholm University, in collaboration with Aarhus University, Boston University, Harvard University, University of Gothenburg, and the University of Texas at Austin.