Varieties of Democracy (V-Dem)
Title: Varieties of Democracy (V-Dem)
DNr: SNIC 2016/1-382
Project Type: SNIC Medium Compute
Principal Investigator: Staffan Lindberg <sil@v-dem.net>
Affiliation: Göteborgs universitet
Duration: 2016-10-01 – 2017-10-01
Classification: 50601 50603 50999
Homepage: https://v-dem.net/en/
Keywords:

Abstract

In the wake of the Cold War democracy has gained the status of a mantra. However, no consensus has emerged about how to conceptualize and measure this key concept. Yet, if we cannot measure democracy in some fashion we cannot mark its progress and setbacks, explain the process of transition, reveal the consequences of those transitions, and affect its future course. This is where the Varieties of Democracy (V-Dem) project comes in. V-Dem project is focused on the construction of a wide-ranging database consisting of several hundreds of indicators and providing a variety of democracy indices – one for each major tradition in thinking about democracy. The project is based at the V-Dem Institute, Department of political science, University of Gothenburg, Sweden. The V-Dem dataset on democracy contains some 350+ indicators and more than 30 democracy indices covering 173 countries from 1900-2015. The database is the most comprehensive database ever compiled on democracy, with currently some 15 million data points It is freely available for download on the project's website together with online tools for analysis where the data can be analyzed and displayed in multiple fashions. To process the data we use series of statistical models such as Bayesian ordinal response theory models and Bayesian factor analysis, which are computationally intensive. Therefore we need to process our data using high performance computing infrastructure Four specific features, considered together, distinguish V-Dem from other measurement projects. A) We approach democracy as a multi-dimensional concept with many components. B) We collect information on indicators relevant to democracy at a highly dis-aggregated level and make all data freely available. C) We employ multiple coders for each (non-factual) question, allowing for inter-coder reliability tests, and use state of the art-measurement modeling to correct for biases and provide estimates of the levels of uncertainty associated with the data. D) We extend indicators of democracy back through modern history to 1900. The data collection builds on collaboration with 2,857 'Country Experts' typically local and resident scholars, and other types of experts across the world. The V-Dem data will lead to entirely new findings on the development of different aspects of democracy never measured before in a systematic fashion. It enables the ability to open up the “boxes” of the overall democracy measures and analyze component-indices, and drill into these components down to the specific indicators constituting them. It opens up unprecedented possibilities to analyze and reveal sequences of democratization and autocratization. The data also opens up for sophisticated approaches to understanding the causes of democratization among other things. The V-Dem data is being updated every year to ensure that up-to-date data always is available. The data is collected in January every year, and is being processed in February and March. This means that the usage of SNIC resources will vary across the year, and that an extended capacity will be needed especially during February and March every year.