Fact-checked by Grok 2 weeks ago

V-Dem Institute

The V-Dem Institute is an independent research institute affiliated with the Department of Political Science at the University of Gothenburg in Sweden, serving as the headquarters for the Varieties of Democracy (V-Dem) project, which produces extensive expert-coded datasets evaluating diverse aspects of democratic institutions and practices worldwide from 1789 to the present.^[1]^[2] Led by Professor Staffan I. Lindberg, the institute coordinates a global network of over 4,000 country experts who code more than 600 indicators, aggregated through Bayesian item-response theory models to generate multidimensional indices encompassing electoral, liberal, participatory, deliberative, and egalitarian principles of democracy.^[1]^[2] The V-Dem project distinguishes itself by disaggregating democracy into specific components rather than relying on aggregate scores, enabling nuanced analysis of causal mechanisms behind democratization and autocratization trends, and its data has become a foundational resource for scholars studying governance variations across time and regimes.^[2] Funded primarily by foundations such as the Knut and Alice Wallenberg Foundation alongside university support, the institute publishes annual Democracy Reports documenting global declines in democratic quality, including rising autocratization in over 40 countries since the early 2000s.^[3]^[4] While praised for its scale and granularity, V-Dem's reliance on subjective expert assessments has prompted methodological critiques regarding inter-coder reliability and potential ideological influences in classifications, particularly evident in contentious labels like "electoral autocracy" applied to populous democracies with competitive elections but policy divergences from liberal orthodoxies.^[5]^[6] Such evaluations, drawn from scholarly networks, may reflect prevailing academic priors that systematically undervalue majoritarian or egalitarian emphases in favor of liberal components, leading to debates over the project's empirical robustness in politically polarized contexts.^[7]

History and Founding

Origins of the V-Dem Project

The Varieties of Democracy (V-Dem) project originated from the recognition among political scientists that existing democracy indices, such as Polity and Freedom House scores, suffered from unidimensional aggregation that obscured variations in democratic attributes across electoral processes, civil liberties, participation, and other components.^[8] This approach stemmed from empirical needs in comparative politics research, where scholars sought granular, disaggregated indicators to better analyze causal mechanisms in democratization, regime change, and institutional performance, rather than relying on composite scores prone to oversimplification.^[2] The project's intellectual foundation emphasized measuring multiple conceptions of democracy—initially encompassing up to seven principles including electoral, liberal, participatory, deliberative, and egalitarian dimensions—to enable more precise hypothesis testing and cross-national comparisons.^[8] Initial collaborations formed around 2010, led by principal investigators Michael Coppedge at the University of Notre Dame's Kellogg Institute for International Studies, Staffan I. Lindberg at the University of Gothenburg, John Gerring at the University of Texas at Austin, Jan Teorell at Lund University (formerly Stockholm), and Carl Henrik Knutsen at the University of Oslo.^[8] Seed funding from the Kellogg Institute and the Swedish Ministry of Foreign Affairs supported early planning and infrastructure development, with Notre Dame hosting key software tools for data management starting in 2011.^[8] These partnerships drew on expertise from North American and European universities to address data scarcity in historical and contemporary regimes, prioritizing expert-coded variables for over 400 attributes traceable to 1900 or earlier in select cases.^[2] By 2011–2013, pilot datasets emerged through working papers and preliminary releases, testing aggregation techniques for multidimensional indices and validating indicators against historical benchmarks, which laid the empirical groundwork for broader data collection without yet formalizing an independent institute.^[9] These efforts highlighted the project's commitment to scalability, involving early country coordinators to refine coding protocols for diverse global contexts.^[2]

Establishment of the Institute

The V-Dem Institute was formally established in 2014 at the University of Gothenburg in Sweden under the leadership of Staffan I. Lindberg, who served as its founding director from 2014 to 2025.^[10] This creation marked the institutionalization of the V-Dem project, shifting it from a decentralized academic collaboration to a centralized operational hub hosted by the Department of Political Science.^[1] The institute's formation addressed the need for coordinated management of expanding data collection and analysis, enabling systematic oversight of a growing international network of country experts.^[11] Initial resources for the institute's setup were provided through internal co-funding from the University of Gothenburg, specifically the College of Social Sciences and the Department of Political Science, which supported data infrastructure and early administrative functions.^[11] This foundational backing facilitated the institute's role in curating and validating expert-coded datasets, laying the groundwork for independent research outputs. By centralizing operations in Sweden, the entity gained autonomy to pursue long-term projects on democracy measurement without reliance on ad hoc partnerships.^[1]

Key Milestones and Expansion

The V-Dem project initiated annual dataset releases starting in the mid-2010s, progressively expanding temporal and spatial coverage to encompass over 200 polities worldwide. By 2018, version 8 integrated the Historical V-Dem extension, pushing data for nearly 200 indicators back to 1789 across approximately 80 polities, including major European states and others from the era of the French Revolution onward.^[12]^[13] Subsequent versions maintained this historical depth while updating contemporary data through 2024, with version 15 released on March 13, 2025, compiling over 31 million data points for 202 countries using refined Bayesian item response theory models for aggregation.^[14]^[15]^[16] These updates incorporated methodological enhancements, such as improved handling of coder disagreement and new evaluative indicators, to address evolving global democratic trends.^[16] Parallel to dataset evolution, the expert coder network scaled to more than 4,000 contributors born across 185 countries by 2025, with 76% holding PhDs and averaging 30 years of in-country experience per coded unit, enabling consistent annual revisions and finer disaggregation of institutional variations.^[2]^[10] This growth supported extensions into subnational-level coding in select regions via affiliated centers by the mid-2020s, broadening applicability beyond national aggregates.^[17]

Organizational Structure and Funding

Leadership and Personnel

The V-Dem Institute is directed by Staffan I. Lindberg, a Swedish political scientist and professor at the University of Gothenburg, who founded the organization in 2014 and serves as one of its principal investigators with a focus on autocratization processes and democratic backsliding.^[1]^[18]^[19] Lindberg's research emphasizes empirical analysis of regime changes, including publications on how electoral manipulations contribute to authoritarian consolidation, shaping the institute's emphasis on disaggregated democracy indicators.^[20] The institute maintains a core staff of approximately 14 personnel at its Gothenburg headquarters, including program managers responsible for coordinating expert quotas, training protocols, and data oversight to ensure consistency in contributions.^[2] These managers facilitate the aggregation of inputs from a global network exceeding 3,000 country experts, primarily recruited from academia and professional fields with documented regional expertise.^[21]^[22] Expert selection prioritizes individuals with specialized knowledge in specific countries and domains, such as elections or civil society, to code targeted indicators, with a structural goal of securing at least five coders per metric to average out individual variances through statistical weighting.^[2]^[23] This approach aims to mitigate subjective biases via volume and cross-verification, though reliance on academic-heavy recruitment—predominantly from institutions prone to ideological homogeneity—raises causal questions about systemic influences on interpretive judgments, as evidenced by director Lindberg's public assessments of figures like Donald Trump as existential threats to U.S. democracy.^[24]^[23]

Funding Sources and Governance

The V-Dem Institute's funding primarily derives from competitive academic grants and philanthropic contributions, with key supporters including the Swedish Research Council (via grants such as 2021-00162 for DEMSCORE infrastructure), the European Research Council (e.g., Consolidator Grant 724191 awarded to principal investigator Staffan I. Lindberg), and the U.S. National Science Foundation (e.g., grant SES-1423944).^[3]^[25]^[26] Additional funding comes from foundations such as the Knut and Alice Wallenberg Foundation (grant 2018.0144) and Riksbankens Jubileumsfond (grant M13-0559:1), as well as international organizations like International IDEA and Denmark's Ministry of Foreign Affairs.^[3]^[27] These sources have enabled the project's expansion, supporting data collection across over 200 countries and involving thousands of experts, with total grant values reaching multimillion euros over project phases since 2010.^[28]^[29] Governance of the V-Dem Institute is anchored in its affiliation with the University of Gothenburg's Department of Political Science, where it operates as a research unit under university oversight, including ethical review boards and academic standards.^[17] The institute maintains an internal organizational structure with a director (Staffan I. Lindberg since inception) and program-specific management, but decision-making on research priorities and data methodology remains insulated from direct donor influence through contractual independence clauses in grants.^[3]^[9] No formal governmental control extends to content production, as funding agencies like the Swedish Research Council emphasize arm's-length support for basic research.^[3] Transparency in funding is facilitated by the institute's public disclosure of donors on its website, allowing scrutiny of potential alignments, particularly given the concentration of support from Western public and private entities that may share liberal democratic priors.^[3] This reliance on such sources—predominantly European and North American—supports operational scale but underscores the need for users to evaluate independence in assessments of non-Western regimes.^[29] The absence of diversified global funding streams, such as from non-Western governments, reflects the project's academic ecosystem but limits broader financial pluralism.^[3]

Institutional Affiliations

The V-Dem Institute maintains its primary institutional affiliation with the University of Gothenburg in Sweden, where it operates as an independent research entity and serves as the central headquarters for coordinating the V-Dem project's global activities, including data aggregation and methodological oversight.^[1] This hosting arrangement, established alongside the project's formalization in the early 2010s, provides academic infrastructure and facilitates integration with university-based research networks.^[30] Key collaborations include a longstanding partnership with the Kellogg Institute for International Studies at the University of Notre Dame, which hosts V-Dem's North American regional center and has contributed to expanding data coverage through expert recruitment and validation since the project's inception around 2010.^[8]^[31] Additional ties involve the Peace Research Institute Oslo (PRIO) for aspects of historical data extension, broadening temporal scope beyond contemporary metrics.^[17] Internationally, V-Dem engages with entities such as the World Bank, including contributions to the 2017 World Development Report that incorporated V-Dem indicators into governance analysis, and International IDEA, supporting cross-institutional data harmonization for policy applications.^[32] These networks, active by the mid-2010s, enable expanded access to diverse expertise and resources, though reliance on aligned academic and multilateral partners may shape interpretive frameworks within shared outputs. Data-sharing initiatives, such as integration with the Demscore e-infrastructure launched in collaboration with other social science data providers, enhance dissemination by linking V-Dem metrics to broader datasets on democracy and governance.^[33] By the 2020s, V-Dem data appeared in platforms like Our World in Data, amplifying global reach through open-access visualizations and analyses without formal exclusivity restrictions.^[34]

Methodology

Core Measurement Principles

The V-Dem framework conceptualizes democracy as encompassing five high-level principles: electoral, which focuses on free and fair elections as the mechanism for popular control; liberal, emphasizing protections for individual and minority rights against state abuses; participatory, highlighting broad citizen inclusion in decision-making beyond voting; deliberative, prioritizing reasoned public debate and discourse in policy formation; and egalitarian, addressing power distribution to mitigate socioeconomic inequalities in political influence.^[35] These principles disaggregate democracy into distinct components, moving beyond binary classifications such as democracy versus autocracy to capture variations within regimes and avoid conflating unrelated attributes into a single metric.^[36] This approach shifts from traditional additive indices, which sum disparate elements into a composite score potentially masking trade-offs or causal complexities, to multidimensional indices that separately track each principle's realization. By treating democracy's varieties as analytically independent yet interrelated, V-Dem enables assessment of regime hybridity and trajectories driven by specific causal mechanisms, such as electoral manipulations eroding liberal safeguards without immediately dismantling participatory elements.^[37] Such granularity reveals assumptions in competing emphases, like electoral-centric views prioritizing contestation over egalitarian redistribution, fostering empirical scrutiny of how these dimensions interact under real-world constraints rather than normative hierarchies.^[38] Measurement employs continuous scales normalized to a 0-1 interval for latent variables representing these principles, derived from probabilistic models that estimate unobserved democratic qualities from observable indicators. This scaling privileges fine-grained empirical variation—allowing, for instance, a score of 0.73 to reflect partial institutionalization—over dichotomous or ordinal categorizations that impose artificial thresholds and obscure incremental changes or measurement uncertainty.^[36] The framework thus underscores causal realism by modeling democracy as a spectrum of attributes subject to contextual contingencies, rather than an idealized endpoint.^[39]

Expert Coding and Data Collection

The V-Dem Institute relies on expert-coded surveys to collect data on latent political traits, such as the quality of deliberation in legislatures or judicial independence, which are challenging to observe empirically and thus prone to subjective interpretation. Country experts, typically academics or practitioners with specialized knowledge, provide ordinal assessments (e.g., on scales from "not at all" to "to a very high extent") via an online platform, enabling detailed scrutiny of interpersonal variations in judgments that could reflect differing priors or contextual understandings.^[36] This human-input approach contrasts with automated or archival data collection, highlighting the interpretive element in quantifying abstract democratic components.^[40] Invitations to code are managed by program managers at the V-Dem Institute, who consult regional managers and country coordinators to nominate candidates based on expertise criteria, prioritizing those ranked highest (1-3 scale) until a quota of at least five experts per country-indicator-year is met, with typically seven recruited to buffer dropouts. Experts receive standardized instructions through the survey interface, available in English, French, Portuguese, Russian, and Spanish, to promote consistent application of coding guidelines, though no separate formal training modules are detailed. The process covers over 500 indicators grouped into clusters (e.g., elections, civil society) across approximately 182 countries from 1900 to 2024, with historical extensions to 1789 for select cases, and involves 1,200–1,700 active coders annually from a cumulative pool exceeding 3,700.^[36] Annual updates incorporate new recruitments to offset attrition (e.g., ~1,700 added since 2013) and require select coders to revise prior entries for temporal consistency, while systematic screening flags implausible patterns like flat-line or erratic codings in under 1% of cases.^[36] Disagreements among experts are addressed through cross-calibration mechanisms prior to aggregation, employing Bayesian item-response theory (IRT) models to estimate individual reliability and adjust for differential item functioning—such as varying thresholds for scale interpretation—via techniques like anchoring vignettes and overlapping "bridge" codings between experts. This weights more reliable contributions higher, mitigating the influence of outliers while preserving the signal from substantive discord, which may indicate genuine uncertainty in hard-to-measure traits.^[36]^[40] Overlapping assignments and posterior-prediction checks further validate coder performance, ensuring the dataset's robustness despite inherent subjectivity.^[36]

Aggregation and Validation Techniques

The V-Dem Institute employs Bayesian Item Response Theory (IRT) models to aggregate ordinal assessments from multiple country experts into continuous latent trait estimates for governance indicators, accounting for coder reliability, item difficulty, and measurement uncertainty through posterior distributions.^[41] This approach scales expert codes—typically on a 0-4 Likert scale—into probabilistic estimates that incorporate disagreement among coders as a measure of epistemic uncertainty, rather than resolving it via simple averaging.^[42] The Bayesian framework allows for hierarchical modeling, where parameters for individual coders (e.g., accuracy and bias) are drawn from hyperpriors informed by cross-coder correlations and over-time stability within countries.^[38] Higher-level indices, such as the Liberal Democracy Index, are constructed using Bayesian factor analysis on the IRT-derived latent variables, weighting components like electoral and liberal principle scores based on their empirical factor loadings and uniqueness estimates to capture multidimensional democratic concepts.^[23] This confirmatory factor model assumes an underlying structure derived from theoretical disaggregation (e.g., electoral vs. participatory democracy), with loadings estimated via Markov Chain Monte Carlo sampling to propagate uncertainty into index confidence intervals. Validation of these aggregates involves internal consistency checks, such as correlating indices with observable events like reported election fraud incidents from ancillary sources, though such external benchmarks are limited by their own measurement errors.^[43] Time-series comparability is addressed through dynamic IRT parameters that adjust for evolving expert measurement norms—e.g., recalibrating historical codes against contemporary anchors—but this process risks hindsight bias, as retrospective adjustments may incorporate post-hoc knowledge unavailable to original coders.^[44] V-Dem documentation notes that dataset revisions for methodological refinements can alter past scores, potentially introducing non-stationarity despite efforts to minimize it via stability priors in the models.^[45] Overall, while these techniques enhance reliability over naive aggregation, their precision depends on the assumption of informed expert priors, which empirical tests of inter-coder agreement (e.g., Krippendorff's alpha >0.7 for most variables) suggest is variable across indicators and contexts.^[41]

Datasets

V-Dem Core Dataset

The V-Dem Core Dataset constitutes the primary aggregated output of the Varieties of Democracy project, focusing on five high-level democracy indices—electoral, liberal, participatory, deliberative, and egalitarian—along with 92 sub-indices and 167 underlying indicators designed to capture multifaceted aspects of democratic governance.^[14] This structure enables detailed empirical analysis by disaggregating democracy into measurable components, such as electoral fairness, judicial independence, and inclusive participation, facilitating causal inquiries into specific institutional mechanisms rather than holistic regime scores. Data are organized in country-year format, spanning up to 202 countries from 1789 to 2024, with coverage intensifying post-1900.^[46]^[47] Version 15 of the dataset, released in March 2025, incorporates refined Bayesian item response theory models for indicator aggregation, accounting for measurement uncertainty through point estimates and confidence intervals.^[14] While primarily at the national level, the dataset's temporal granularity supports within-country variation studies over time, complemented by associated tools like the Country Graph for visualizing trends in selected indicators or indices.^[48] The Core Dataset excludes specialized extensions, concentrating on core democracy metrics to provide a foundational resource for cross-national and longitudinal research. Access to the V-Dem Core Dataset is provided free of charge through direct downloads in formats including Stata, CSV, R data files, and SPSS, with an institutional citation requirement for users.^[14] Archival versions of prior releases are maintained to allow researchers to track methodological updates and ensure reproducibility in analyses affected by revisions, such as indicator recalibrations or expanded historical coverage.^[49] This version control underscores the dataset's commitment to transparency in evolving measurement approaches.

V-Party Dataset

The V-Party Dataset, formally known as the Varieties of Party Identity and Organization Dataset, compiles expert-coded assessments of political parties' policy positions, internal structures, and operational behaviors to examine their role in democratic processes. It encompasses data on 3,467 parties across 178 countries, spanning factual records from 1900 to 2019 and expert evaluations primarily from 1970 to 2019. This granular, party-year level aggregation draws factual elements from sources like PartyFacts and election outcomes, enabling longitudinal tracking of party evolution.^[50] Coding focuses on ideological orientations, such as economic left-right positioning (v2pariglef) and welfare stances (v2pawelf); origins via metrics like party continuation (v2paelcont); organizational attributes including local branch presence (v2palocoff), candidate selection methods (v2panom), and internal discipline (v2padisa); and behaviors encompassing populism (v2xpa_popul index) and clientelism (v2paclient). These indicators are derived from inputs by 711 experts applying V-Dem's Bayesian Item Response Theory model, which aggregates multiple coders (minimum five per observation) to generate probabilistic estimates and mitigate individual biases. Factual variables supplement expert data for robustness.^[50] Distinct from aggregate country-level metrics, V-Party facilitates micro-level analysis of how party traits—such as populist appeals or clientelist strategies—affect broader democratic varieties, including participatory elements through decentralized structures or elite-driven ideologies. It integrates directly with the V-Dem Core Dataset, allowing researchers to link party-level variables to regime outcomes like electoral integrity or polyarchy indices, as supported by unified access tools in V-Dem's data packages. Updates follow V-Dem's annual revision cycle, with version 2 (2022) refining indices (e.g., anti-pluralism measures) and correcting prior entries for enhanced validity.^[50]^[51]

Electoral Risk Tracker (ERT) Dataset

The Electoral Risk Tracker (ERT) Dataset offers probabilistic forecasts of electoral risks, particularly violence and manipulation, for upcoming election cycles worldwide. Unlike retrospective measures, it emphasizes predictive modeling to identify vulnerabilities in real time, drawing on V-Dem's granular indicators of electoral processes alongside event-based data on past incidents. This forward-looking approach supports interventions by quantifying risks such as intimidation, fraud, or post-election unrest before they materialize.^[52] The methodology combines historical patterns from V-Dem's expert-coded data—covering factors like electoral authority independence and opposition suppression—with machine learning ensembles incorporating economic stressors, political instability metrics, and digital threat indicators. Models are trained to output risk levels ranging from negligible to severe violence, validated through true out-of-sample testing to mitigate overfitting. Current expert assessments refine inputs for near-term events, yielding country-specific probabilities rather than deterministic outcomes.^[52]^[53] Applied to 2024-2025 elections, the ERT highlights elevated risks in nations with recurrent violence histories or acute instability, such as those in sub-Saharan Africa and parts of Asia, where ensemble predictions exceed baseline thresholds for severe incidents. For instance, forecasts prioritize countries exhibiting combined high political repression and socioeconomic volatility, informing policy responses like observer deployments or dialogue facilitation. Outputs remain probabilistic, acknowledging model uncertainties from data gaps in autocratic contexts.^[52]

Historical V-Dem Dataset

The Historical V-Dem Dataset extends the V-Dem project's coverage to enable analyses of democracy's long-term causal drivers, reaching back to 1789 for approximately 80 polities through the adaptation of around 130 core indicators.^[12] These indicators, drawn from the broader V-Dem framework, measure dimensions such as electoral processes, political participation, and institutional accountability, repurposed for pre-20th-century contexts using archival records like constitutional texts, legislative debates, and contemporary accounts.^[12] Specialist coders with expertise in historical political institutions contribute assessments, ensuring consistency with V-Dem's expert-coding methodology while addressing era-specific evidence limitations.^[12] Initial releases began in 2022, progressively integrating historical data into the main V-Dem dataset to fill gaps in pre-1900 coverage, which previously constrained comparative studies of democratization patterns across centuries.^[12] By 2025, this extension supports over 31 million data points spanning 1789 to 2024 for up to 202 countries, though historical portions prioritize polities with sufficient archival depth, such as European states and early American republics.^[15] The dataset's design facilitates empirical tests of factors like economic development, warfare, and colonial legacies on regime trajectories, drawing on disaggregated indicators for causal inference over extended periods.^[12] Coding historical periods encounters challenges from sparser and more fragmented evidence, leading to elevated uncertainty intervals in aggregated indices compared to modern data.^[40] V-Dem's Bayesian measurement model incorporates coder disagreement and source reliability to quantify this uncertainty, producing confidence bounds that widen for earlier years where primary documents are scarcer or biased toward elite perspectives.^[40] Despite these hurdles, validation against qualitative historical scholarship confirms the dataset's utility for robust, long-run regressions on democracy's determinants.^[54]

Publications and Reports

Annual Democracy Reports

The V-Dem Institute has published an annual Democracy Report each March since approximately 2017, providing a yearly assessment of global and regional trends in democratic governance based on its expert-coded indicators.^[4] These reports synthesize data from the latest V-Dem dataset version to track changes in regime types, highlighting shifts such as the proportion of autocracies versus democracies and patterns of autocratization or democratization across countries.^[4] Each report follows a consistent structure, beginning with global overviews of democracy indices, followed by country-specific and regional analyses, and concluding with thematic essays on issues like electoral integrity and institutional erosion. Visualizations, including maps and time-series graphs derived from V-Dem metrics, illustrate key trends, such as declines in election quality in specific regions.^[10] To enhance accessibility, recent editions offer multilingual versions in languages including English, Portuguese, and Spanish.^[4] The 2025 Democracy Report, titled "25 Years of Autocratization – Democracy Trumped?", analyzes regime transformations from 2000 to 2024 using V-Dem Dataset v15, documenting that autocracies now comprise 91 countries compared to 88 democracies—the first such reversal in over two decades. It identifies 45 countries in autocratization episodes against 19 undergoing democratization, attributing these shifts to factors like weakened electoral processes and executive aggrandizement in thematic sections.^[10]^[55]^[56]

Key Thematic Outputs

The V-Dem Institute issues working papers that provide specialized analyses beyond annual reports, often employing advanced methodologies to explore regime dynamics. Working Paper No. 150, "Forecasting Electoral Violence," published in November 2024 by David Randahl and colleagues, introduces an ensemble machine learning model to predict risks of electoral violence in global elections from 2024 to 2025, integrating V-Dem indicators with economic data and historical violence records to achieve high predictive accuracy for medium-term forecasts.^[52] This output supports proactive policy interventions by identifying high-risk contexts where violence by non-state actors or government intimidation disrupts electoral processes.^[53] Another thematic focus involves sequencing in regime transformations, as detailed in Working Paper No. 141, "Chains in Episodes of Democratization," from June 2023 by Kelly Morrison and co-authors, which applies chain analysis to V-Dem data across episodes since 1900, revealing that electoral reforms typically initiate democratization sequences but correlate weakly with overall success rates.^[6] Complementary analyses extend this to autocratization reversals, with Working Paper No. 147, "When Autocratization is Reversed: Episodes of Democratic Turnarounds since 1900," published in January 2024, documenting that 48% of autocratization episodes since 1900 lead to democratic recoveries, rising to 70% in recent decades, often restoring prior democratic levels through targeted institutional reforms.^[57] V-Dem's disaggregated indices facilitate nuanced thematic insights into democratic components, as elaborated in Working Paper No. 135, "V-Dem’s Conceptions of Democracy and Their Consequences," by Michael Coppedge in February 2023, which delineates five varieties—electoral, liberal, participatory, egalitarian, and deliberative—enabling differentiation of declines, such as steeper erosions in liberal principles (e.g., rule of law) compared to egalitarian components (e.g., power alternations) in specific historical periods.^[39] The Deliberative Democracy Index, measuring inclusive public reasoning oriented toward the common good, has been integrated into external analytical platforms for comparative studies, highlighting variances where deliberative practices lag behind electoral metrics in consolidating democracies.^[2] Thematic reports further these outputs, exemplified by the March 2024 report "Democracy’s Core Institution – Clean Elections Across the World" by Marina Nord, Juraj Medzihorsky, and Staffan I. Lindberg, which uses V-Dem's election integrity indicators to assess global cleanliness trends, identifying persistent deficits in vote-buying and fraud prevention despite formal electoral frameworks.^[58] These publications underscore V-Dem's emphasis on granular, evidence-based dissections of democratic attributes to inform targeted research and policy.^[59]

Data Accessibility and Updates

The V-Dem datasets are provided free of charge for public download directly from the institute's website, supporting replicability through open access to raw and processed data files.^[14] Available formats include STATA, CSV, R, and SPSS, with options for country-year, country-date, and coder-level aggregates to facilitate analysis across different granularity levels.^[14] Supplementary tools for data visualization and graphing are integrated into the platform, enabling users to generate custom outputs without proprietary software dependencies.^[60] Versioned releases ensure traceability and allow researchers to reference specific iterations for reproducible studies, with the latest being version 15 released in March 2025.^[49] Previous versions, dating back through annual updates, are archived on the V-Dem site alongside external repositories such as the Swedish National Data Service (SND) and CurateND at the University of Notre Dame, which host additional files including full posterior distributions for uncertainty estimation.^[61] These distributions, derived from Bayesian item response theory models, provide probabilistic bounds on indicator estimates, encouraging users to incorporate measurement error in validations rather than treating point estimates as definitive.^[15] Major dataset updates occur annually in March, aligning with the release of the institute's Democracy Report and incorporating coder revisions based on emerging evidence or methodological refinements.^[62] Minor interim adjustments, such as patches for identified errors, may be issued between major versions, with backward revisions to historical ratings permitted when coders access superior information, though these are reflected in subsequent full releases rather than standalone patches.^[63] This schedule balances timeliness with rigor, as evidenced by consistent yearly increments from version 12 in 2022 to version 15 in 2025, prioritizing comprehensive expert recoding over ad hoc updates.^[61]

Reception and Impact

Academic and Policy Influence

The V-Dem datasets have facilitated extensive empirical analysis in comparative politics, particularly through their provision of high-frequency, country-year panel data spanning over two centuries, which supports advanced causal inference techniques such as difference-in-differences and instrumental variable approaches to examine democratic backsliding and regime transitions.^[64]^[65] Scholars have leveraged these disaggregated indicators to test causal mechanisms of autocratization, including the role of executive aggrandizement and electoral manipulation, yielding insights into how institutional erosion precedes electoral declines.^[66] The project's methodological innovations, such as Bayesian item response theory for aggregating expert-coded indicators, have enhanced the reliability of time-series cross-sectional models, enabling robust identification of within-country variation in democratic qualities.^[67] In terms of scholarly adoption, the V-Dem dataset has been downloaded more than 400,000 times by users in over 200 countries since 2016, reflecting broad integration into academic workflows for quantitative research on governance and political institutions.^[10] This uptake has influenced thousands of studies annually, as evidenced by the frequent citation of V-Dem-derived indices in peer-reviewed journals on topics like the effects of democratic qualities on economic outcomes and conflict prevention, where the data's granularity allows for nuanced causal claims beyond binary regime classifications.^[68]^[69] On the policy front, V-Dem data have been incorporated into World Bank resources, including the V-Dem Core dataset, which underpins analyses of democratic governance for development programming and fragility assessments.^[70] The Electoral Risk Tracker (ERT) dataset, in particular, has informed election monitoring efforts through collaborations such as the one with the Kofi Annan Foundation, which developed Electoral Vulnerability Indexes to forecast violence risks and guide preventive interventions in at-risk polities.^[71] These tools have been referenced in international policy discussions on electoral integrity, aiding organizations in prioritizing resources for observation and support in high-risk elections as of 2023.^[72]

Applications in Research and Journalism

V-Dem datasets enable researchers to conduct time-series analyses of democratic erosion by providing granular, annual indicators spanning centuries, allowing for econometric models that link democratic quality to economic variables. For example, studies have utilized V-Dem's polyarchy and electoral democracy indices to estimate the impact of democratic levels on per capita GDP growth, finding positive associations in panel regressions across developing and advanced economies from 1900 onward.^[73] Similarly, analyses of crisis periods employ V-Dem data to assess how variations in institutional accountability mitigate or exacerbate economic downturns, with findings indicating that higher democracy scores correlate with shallower recessions and faster recoveries in post-2008 data.^[74] These applications leverage V-Dem's disaggregated variables, such as electoral fairness and executive constraints, to isolate causal pathways beyond aggregate indices.^[75] In journalism, V-Dem reports and metrics inform coverage of global and national democratic trends, with outlets citing the institute's autocratization classifications to frame stories on electoral integrity and institutional decline. Annual Democracy Reports, which track episodes of democratic regression in over 70 countries since 1994, have been referenced in analyses of events like the 2020 U.S. presidential election, where V-Dem indicators highlighted deteriorations in voter suppression and media freedom sub-indices.^[76] Journalists draw on these for narratives emphasizing rapid shifts, such as the reclassification of 42 countries as electoral autocracies by 2023, though coverage often focuses on headline aggregates rather than underlying coder variances.^[77] V-Dem data extends to public-facing tools, integrating into platforms like Our World in Data for interactive visualizations that democratize access to time-series trends. There, the electoral democracy index is charted alongside economic and social metrics, enabling users to explore correlations between democratic scores and outcomes like human development from 1789 to the present, with over 30 million graphs generated via V-Dem tools since 2016.^[34]^[76] This facilitates broader journalistic and research synthesis, though it underscores the need for contextual interpretation of expert-coded estimates.^[29]

Criticisms and Controversies

Methodological Challenges

The V-Dem dataset relies on expert coders to provide ordinal assessments of latent concepts such as institutional components of democracy, which are inherently subjective and not directly observable.^[78] These assessments use scales prone to interpersonal differences in interpretation, even after calibration efforts like item-response theory (IRT) modeling to account for coder traits.^[23] Empirical analyses of coder-level data reveal significant disagreements on specific country-year observations, particularly for corruption-related indicators, undermining the reliability of disaggregated measures despite aggregation techniques.^[79] Retrospective coding of historical periods introduces risks of time-inconsistency, as experts may adjust past evaluations influenced by contemporary events or hindsight bias.^[45] V-Dem's iterative updates to historical data, aimed at refining estimates, can lead to score revisions that reflect evolving interpretive norms rather than new evidence, potentially distorting longitudinal trends.^[23] Studies on expert surveys indicate that democratic crises in recent years systematically alter retrospective judgments, with coders exhibiting availability bias toward current conditions when reassessing prior eras.^[80] Validation of V-Dem's high-level indices faces limits due to their dependence on unobservable constructs, complicating external corroboration against objective benchmarks like electoral data or institutional records.^[81] While the measurement model generates confidence intervals, critiques highlight potential over-precision in aggregated scores, as IRT assumptions about coder independence and scale comparability may underestimate uncertainty in latent variable estimates.^[82] Face validity checks by project coordinators provide internal consistency but do not fully address aggregation artifacts or the challenges of scaling ordinal inputs to continuous indices without empirical anchors for many variables.^[38]

Allegations of Political Bias

Critics have alleged that the V-Dem Institute exhibits an ideological skew toward liberal democratic norms, particularly in its assessments of regimes led by conservative or nationalist leaders. For instance, Hungary's democracy score has declined sharply since Viktor Orbán's Fidesz party consolidated power in 2010, with V-Dem highlighting rises in anti-pluralism and media control under metrics emphasizing deliberative and egalitarian principles over purely electoral outcomes.^[83]^[84] Similarly, India was reclassified as an "electoral autocracy" in V-Dem's 2021 report, citing erosion in freedom of expression and civil liberties under Narendra Modi's BJP government since 2014, a designation that Indian officials dismissed as reflecting Western ideological bias rather than empirical electoral integrity.^[85]^[86] These downgrades are said to prioritize non-electoral dimensions, such as egalitarian participation and liberal protections, which allegedly penalize policies restricting migration or cultural pluralism in conservative contexts while applying less scrutiny to comparable restrictions in left-leaning autocracies like Venezuela or Nicaragua.^[7] The composition of V-Dem's expert coder pool has fueled claims of embedded liberal hegemony, as the majority hail from Western academic institutions where surveys indicate predominant left-leaning ideological orientations.^[87] Political scientist Jonas Wolff argues that V-Dem has shifted from a pluralistic conceptualization of democracy—encompassing diverse regime types—to a decontested defense of liberal norms amid perceived global threats, effectively reconstituting ideological consensus around egalitarian and deliberative ideals at the expense of electoral-focused metrics.^[88] This convergence, per analyses, risks systematic skew in coding conservative-led regimes as autocratizing, as coders' socialization in environments favoring liberal hegemony may undervalue alternative democratic forms emphasizing national sovereignty or majority rule.^[7] V-Dem has countered these allegations by emphasizing coder diversity, with 61% of experts born or residing in the countries they assess, and by incorporating statistical models to account for measurement uncertainty and potential biases.^[76] Internal studies, such as those examining selection into the coder pool, report no evidence of systematic pessimism or ideological bias driving global backsliding assessments.^[89]^[90] However, critics note the absence of fully independent external audits of coder selection or ideological balance, relying instead on self-reported validations within an academic ecosystem known for left-leaning skews.^[87]^[7]

Discrepancies with Alternative Indices

V-Dem's multidimensional approach, incorporating over 400 indicators across various democracy components, frequently yields steeper declines in scores for Western democracies compared to indices like Freedom House's Freedom in the World and the Economist Intelligence Unit's (EIU) Democracy Index. For the United States, V-Dem's Liberal Democracy Index fell from 0.85 in 2016 to approximately 0.73 by 2020, reclassifying the country from a liberal democracy to an electoral democracy due to deteriorations in executive oversight, media impartiality, and academic freedom sub-indices. In contrast, Freedom House's aggregate score for the US declined more modestly from 90/100 in 2016 to 83/100 in 2023, retaining its "free" status, while the EIU downgraded the US to a "flawed democracy" in 2016 (score 7.98/10) with subsequent dips to 7.85/10 by 2023 but without the categorical shift V-Dem recorded. These divergences stem from V-Dem's granular sub-indices, which capture nuanced erosions in liberal components—such as constraints on executive power and egalitarian treatment—that coarser aggregates in Freedom House (focused on political rights and civil liberties) or EIU (emphasizing electoral process and pluralism) may underweight.^[91] Empirical analyses indicate V-Dem detects autocratization episodes—defined as sustained declines in electoral democracy scores—in more countries and years than Polity IV, which relies on a narrower authority spectrum and institutional checks, resulting in fewer flagged events overall. For instance, Polity IV operationalizations identify substantially fewer autocratization cases globally, including minimal change in high-scoring polities like the US (stable at +10 on its -10 to +10 scale since the early 2000s), highlighting V-Dem's heightened sensitivity to incremental shifts in sub-dimensions. Freedom House, conversely, registers more autocratization events than V-Dem in some contexts due to its binary-leaning thresholds for rights violations.^[92] Such discrepancies underscore the trade-offs in measurement design: V-Dem's disaggregation enhances detection of causal mechanisms like partisan media influence or judicial independence erosion but risks amplifying measurement noise or coder subjectivity in contested areas, potentially exaggerating trends in polarized settings.^[93] Researchers and policymakers are advised to cross-validate V-Dem with alternatives, as reliance on any single index may distort assessments of regime trajectories, particularly where multidimensionality interacts with varying coder expertise or interpretive frames across datasets.^[91]

References

[1]
V-Dem Institute
The V-Dem Institute, led by Professor Staffan I. Lindberg, is an independent research institute and the Headquarters of the V-Dem project.
[2]
V-Dem Project
The V-Dem project distinguishes between five high-level principles of democracy: electoral, liberal, participatory, deliberative, and egalitarian.
[3]
Funders - V-Dem
Funders of V-Dem Research Programs. Learn more about our Research Programs here. Knut and Alice Wallenberg Foundation University of Gothenburg.
[4]
Democracy Reports - V-Dem
Democracy Reports. Released by the V-Dem Institute annually in March. Democracy Report 2025. Download the Democracy Report 2025 in ...
[5]
[PDF] The V–Dem Measurement Model
Marquardt & Pemstein (2021) illustrate that the V–Dem measurement model tends to perform similarly or better than these approaches under a variety of assumed ...Missing: peer- | Show results with:peer-<|control11|><|separator|>
[6]
Working Papers - V-Dem
There are longstanding concerns about biases in expert-coded data. Here we attempt to causally identify the effect of these biases, specifically evaluating the ...
[7]
V-Dem and the reconstitution of liberal hegemony under threat
Jul 6, 2022 · This paper argues that the perception of serious threats to democracy in general and liberal norms in particular has given rise to a convergence around the ...
[8]
Varieties of Democracy Project | Kellogg Institute For International ...
The Varieties of Democracy Project (V-Dem) is an ambitious international effort to produce new indicators of democracy for all countries since 1900.
[9]
[PDF] Organization and Management - V-Dem
Mar 15, 2025 · As a collaborative effort between leading Swedish universities, Demscore elevates the scale of social science data to a new level, offering ...<|control11|><|separator|>
[10]
[PDF] V-DEM Democracy Report 2025 25 Years of Autocratization
Mar 6, 2025 · bias. Attacks on journalists by the government and its associates have become increasingly more frequent in 35 countries. Guatemala,. Hong ...Missing: controversies | Show results with:controversies
[11]
[PDF] CURRICULUM VITAE Staffan I. Lindberg
(2) PI, “V-Dem”, internal co-funding for data collection and establishment of V-Dem Institute from. University of Gothenburg, College of Social Science and ...Missing: Riksbank | Show results with:Riksbank<|separator|>
[12]
Historical V-Dem
Sep 7, 2022 · Historical V-Dem has mainly been funded through three larger research grants from the Norwegian (Young Research Talent Grant pnr 240505; PI Carl ...
[13]
Introducing the Historical Varieties of Democracy dataset: Political ...
May 14, 2019 · With Historical V-Dem, most indicators contained in V-Dem now extend back to 1789, offering continuous time series across more than 225 ...
[14]
The V-Dem Dataset
Includes the world's most comprehensive and detailed democracy ratings. The latest version of the dataset and associated reference documents can be downloaded ...
[15]
Varieties of Democracy (V-Dem) Data — v.15 (2025) - Curate ND
Varieties of Democracy 15 produces the largest global dataset on democracy with over 31 million data points for 202 countries from 1789 to 2024.
[16]
[PDF] Codebook | V-Dem
Mar 15, 2025 · ... Varieties of Democracy project, divided by theme. – 3.1 Elections. – 3.2 Political Parties. – 3.3 Direct Democracy. – 3.4 The Executive. – 3.5 ...Missing: inception | Show results with:inception
[17]
https://www.v-dem.net/documents/58/Organization_v15.pdf
[18]
Staffan I. Lindberg - V-Dem
Staffan I. Lindberg is the Director of the V-Dem Institute at University of Gothenburg and one of five Principal Investigators for Varieties of Democracy ...Missing: origins | Show results with:origins
[19]
Staffan Ingemar Lindberg | University of Gothenburg
Lindberg's main current occupation is as Director of the V-Dem Institute at University of Gothenburg and one of four Principal Investigators for Varieties of ...<|separator|>
[20]
V-Dem: A New Way to Measure Democracy
Staffan I. Lindberg is professor of political science and director of the V-Dem Institute at the University of Gothenburg. View all work by Staffan I. Lindberg ...
[21]
Call for Country Experts - V-Dem
Sep 27, 2023 · We are looking for mainly academics or professionals with specialist and documented knowledge in one or more domains, for a specific country.
[22]
V-Dem Institute is Seeking Country Experts | IPSA
V-Dem has a global network of more than 3,000 Country Experts supporting the data collection process by providing expert information via our online surveys.
[23]
[PDF] Methodology - V-Dem
Mar 14, 2024 · 3.6 Expert Coding Process . ... area of documented expertise (relevant for the selection of surveys the expert might be competent to code).
[24]
V-Dem Director Lindberg: If Trump Is Reelected, Democracy ... - ECPS
Apr 9, 2024 · V-Dem's Director Staffan I. Lindberg expresses his concern: “I am deeply concerned about the possibility of Donald Trump being reelected.
[25]
[PDF] A Framework for Understanding Regime Transformation - V-Dem
This research project was principally supported by European Research Council, Consolidator Grant 724191, PI: Staffan I. Lindberg; but also by Knut and Alice.
[26]
https://gupea.ub.gu.se/bitstream/handle/2077/67993/gupea_2077_67993_4.pdf?sequence=4
[27]
https://gupea.ub.gu.se/bitstream/handle/2077/70498/gupea_2077_70498_1.pdf?sequence=1
[28]
Varieties of Democracy Dataset Released to Public
Jan 4, 2016 · The V-Dem project has been made possible with funding from ... Swedish Research Council, International IDEA, Denmark's Ministry of ...
[29]
The “Varieties of Democracy” data: how do researchers measure ...
Nov 30, 2022 · True to its name, the Varieties of Democracy project acknowledges that democracy can be characterized differently, and measures electoral, ...
[30]
[PDF] Organization and Management - V-Dem
This generated funding from multiple sources as well as a set of joint activities such as the development of an Index on the Global State of Democracy. • V ...
[31]
North America – V-Dem
The Kellogg Institute for International Studies at the University of Notre Dame was an instrumental part in building and establishing V-Dem. It was initially ...
[32]
[PDF] Organization and Management - V-Dem
Mar 12, 2022 · V-Dem collaborated with the World Bank in drafting parts of the World Development Report 2017. V-Dem data is used throughout the report ...
[33]
Collaborations - V-Dem
An e-infrastructure providing free access to harmonized data from several of the world's largest datasets on Democracy, Environment, Migration, Social Policy, ...Missing: partnerships | Show results with:partnerships
[34]
Varieties of democracy - Our World in Data
The indices are by V-Dem, based on expert estimates, and range from 0 to 1 (most democratic).Missing: agreements | Show results with:agreements
[35]
About - V-Dem
V-Dem is a unique approach to conceptualizing and measuring democracy. V-Dem distinguishes between five high-level principles of democracy.
[36]
[PDF] Methodology | V-Dem
Mar 15, 2025 · days collectively reviewing all data collated at that point to validate the approach and aggregation methods. The process of face validity ...Missing: peer- | Show results with:peer-
[37]
[PDF] The Methodology of “Varieties of Democracy” (V-Dem)1
At this point, V-Dem offers separate indices of five varieties of democracy: electoral, liberal, participatory, deliberative, and egalitarian. The high-level ...
[38]
[PDF] Methodology - V-Dem
release of each new dataset. Several configurations of the V-Dem dataset are available, including country-year, country- date, and coder- level datasets ...
[39]
[PDF] V-Dem's Conceptions of Democracy and Their Consequences
Varieties of Democracy (V-Dem) is a unique approach to conceptualization and measurement of democracy. The headquarters – the V-Dem Institute – is based at ...
[40]
V-Dem Methodology
V-Dem has developed innovative methods for aggregating expert judgments in a way that produces valid and reliable estimates of difficult-to-observe concepts.Missing: core principles multidimensional
[41]
[PDF] Methodology - V-Dem
Mar 13, 2023 · Therefore, we use Bayesian item response theory (IRT) modeling techniques (Fox 2010) to estimate latent polity characteristics from our ...
[42]
[PDF] IRT models for expert-coded panel data - V-Dem
IRT models address expert disagreement in coding by accounting for varying expertise and interpretation of question scales, outperforming simple averages.<|separator|>
[43]
Data Validation (Chapter 6) - Varieties of Democracy
Jan 25, 2020 · ... V-Dem uses statistical item response theory (IRT) techniques. They help model variation in coder reliability while allowing for the ...
[44]
[PDF] Autocratization Changing Nature ? - V-Dem
This process of constant improvement may lead to a correction of scores reported in earlier versions of the dataset and the Democracy Report.Missing: hindsight adjustments
[45]
Hindsight bias in expert surveys: How democratic crises influence ...
Aug 9, 2025 · We argue that hindsight bias is a potential problem when it comes to retrospective evaluation due to the difficulty involved in separating our ...
[46]
Country-Year: V-Dem Core v15
V-Dem Core version 15. The five high-level V-Dem democracy indices, 92 sub-indices, and the 167 indicators constituting them.
[47]
None
Summary of each segment:
[48]
Country Graph - V-Dem
Site overview. About. V-Dem Project · V-Dem Institute · Regional Centers · Historical V-Dem · Funders · Internships · Contact. Research & Publications. Research ...
[49]
https://v-dem.net/data/dataset-archive/
[50]
[PDF] Codebook Varieties of Party Identity and Organization (V-Party)
Feb 2, 2022 · ... coding of one indicator depends upon the coding of other indicators ... V–Dem country coding periods, please see the V–Dem Country Coding.
[51]
vdeminstitute/vdemdata: An R package to load, explore and ... - GitHub
This package contains the most recent V-Dem (Country-Year: V-Dem Full+Others) and V-Party datasets and provides some additional functions.Issues 4 · Pull requests 0 · Actions · Activity
[52]
[PDF] Forecasting Electoral Violence - V-Dem
Nov 5, 2024 · This study uses machine learning models with economic indicators, historical violence, political instability, and digital vulnerability to ...Missing: methodology | Show results with:methodology
[53]
https://doi.org/10.1016/j.ijforecast.2025.09.003
[54]
Introducing the Historical Varieties of Democracy Dataset: Political ...
The dataset covers 91 polities globally – including most large, sovereign states, as well as some semi-sovereign entities and large colonies – from 1789 to 1920 ...
[55]
Two decades of decline in the global state of democracy
Apr 9, 2025 · The V-Dem Democracy Report 2025 states that the number of autocracies (N=91) has exceeded the number of democracies (N=88) for the first time in more than 20 ...
[56]
Autocracies outnumber democracies for the first time in 20 years: V ...
Mar 27, 2025 · The 2025 V-Dem report finds that 45 countries are affected by autocratization and 19 are in episodes of democratization.<|control11|><|separator|>
[57]
https://doi.org/10.1080/13510347.2024.2448742
[58]
https://www.v-dem.net/media/publications/Elections_Report.pdf
[59]
Thematic Reports - V-Dem
Working Papers · Policy Briefs · User Working Papers · Country Briefs · Thematic Reports · News · Sign In · Books etc · Democracy Reports · Journal Articles ...
[60]
https://v-dem.net/documents/60/V-dem-dr__2025_lowres.pdf
[61]
Dataset Archive - V-Dem
Access and download previous dataset versions. We also have external archives at the SND and at CurateND where eg, posterior distributions for variables, and ...
[62]
Varieties of Democracy (V-Dem) | University of Gothenburg
Mar 13, 2025 · The V-Dem project distinguishes between five high-level principles of democracy: electoral, liberal, participatory, deliberative, and ...
[63]
FAQ - V-Dem
V-Dem Institute · Regional Centers · Historical V-Dem · Funders · Internships · Contact. Research & Publications. Research Programs · Collaborations ...
[64]
V-Dem Reconsiders Democratization (Chapter 1)
Jun 28, 2022 · The authors review some of the prominent arguments regarding the relationship between different types of political institutions and democracy.
[65]
Causal Panel Analysis under Parallel Trends - arXiv
Over the past decade, political scientists have increasingly relied on panel data to draw causal conclusions (Xu, 2023) . A favored method for such analyses ...
[66]
[PDF] Defiance in the Face of Autocratization - V-Dem
The V-Dem Institute's Democracy Report centers on the Liberal. Democracy Index (LDI). It combines the 'core' institutions of electoral democracy with the ...
[67]
[PDF] Codebook - V-Dem
Mar 14, 2024 · ... Varieties of Democracy project, divided by theme. – 3.1 Elections. – 3.2 Political Parties. – 3.3 Direct Democracy. – 3.4 The Executive. – 3.5 ...
[68]
The effects of different democratic qualities on climate change ...
In this study, we go beyond a narrow definition of democracy by investigating how different democratic qualities influence national CO2 emissions per capita.<|separator|>
[69]
https://v-dem.net/media/publications/UWP_51.pdf
[70]
V-Dem Core - World Bank Open Data
A multidimensional dataset that captures the complexity of democracy beyond elections. It defines democracy through five key principles.
[71]
Introducing a new tool to Predict Election-Related Violence
Mar 10, 2023 · The Kofi Annan Foundation has teamed up with V-Dem to develop two Electoral Vulnerability Indexes - a tool to predict electoral-related ...
[72]
https://v-dem.net/wp.html
[73]
[PDF] Local Democracy and Economic Growth - V-Dem
A few recent studies find a positive relationship between democracy and per capita GDP growth (e.g., Acemoglu et al. 2015;. Gerring et al. 2005; Knutsen 2015).
[74]
[PDF] Democracies in Crisis - V-Dem
This paper estimates how levels and changes of democracy affect economic outcomes around economic crises, using yearly data from the Varieties of Democracy ...
[75]
The V-Dem Dataset – V-Dem
### Summary of Citations, Usage Statistics, and Impact from V-Dem Dataset Page
[76]
[PDF] V-DEM Democracy Report 2025 25 Years of Autocratization
Mar 6, 2025 · The V-Dem dataset has been downloaded by users 400,000 times in 200+ countries since 2016. 30 million graphs created using the online tools by ...Missing: milestones | Show results with:milestones
[77]
Democratic backslide a threat to free elections globally
Mar 7, 2024 · Over half of the 60 countries holding national elections this year are experiencing a democratic decline, risking the integrity of the electoral process.<|separator|>
[78]
The V-Dem Measurement Model: Latent Variable Analysis for Cross ...
May 8, 2020 · The Varieties of Democracy (V-Dem) project relies on country experts who code a host of ordinal variables, providing subjective ratings of latent- that is, not ...
[79]
[PDF] Assessing The Varieties of Democracy Corruption Measures
Our assessments of the data generating process reveal that V-Dem coders disagree significantly on particular subsets of country-year observations, indicating ...
[80]
Recent Events and the Coding of Cross-National Indicators
Aug 8, 2023 · ... coding significantly affect the reliability of cross-national indicators. ... V-dem dataset V10 codebook. Varieties of democracy (V-dem) ...
[81]
The Measurement Model and Reliability (Chapter 4)
Jan 25, 2020 · V-Dem relies on Country Experts who code a host of ordinal variables, providing subjective ratings of latent – that is, not directly observable ...
[82]
[PDF] Experts, Coders, and Crowds: An analysis of substitutability - V-Dem
Crowdsourcing—the large-scale recruitment of lay persons to code data—has emerged as a tool for data collection in traditionally expert-reliant domains (Kittur, ...
[83]
https://v-dem.net/weekly_graph/rising-anti-pluralism-in-hungarys-fidesz-party
[84]
Parliamentary Elections in Hungary and its Autocratization - V-Dem
May 4, 2022 · On April 3, Viktor Orbán's national-conservative Fidesz party won a two-thirds majority in Hungary's general elections.<|separator|>
[85]
'Electoral autocracy': The downgrading of India's democracy - BBC
Mar 15, 2021 · Last week, Sweden-based V-Dem Institute was harsher in its latest report on democracy. It said India had become an "electoral autocracy ...<|control11|><|separator|>
[86]
India 'one of the worst autocratisers': V-Dem report on democracy
Mar 11, 2024 · India's democratic decline highlighted in V-Dem report, with autocratisation worsening and freedom of expression under threat.<|separator|>
[87]
Inside the V-Dem Rankings - Indian Century Roundtable
Mar 1, 2025 · The Varieties of Democracy Institute (V-Dem) incorrectly categorizes India as an "electoral autocracy." The problems with V-Dem do not ...
[88]
V-Dem's Discursive Turn from the Contestation to the Decontestation ...
Sep 6, 2022 · In order to defend liberal democracy against its critics, V-Dem has chosen to use unequivocal, decontested and simple concepts: democracy is ...
[89]
Expert Bias and Democratic Erosion
Jan 11, 2024 · We find no evidence of bias toward pessimism at either stage of selection into the V-Dem coder pool, invitation or participation. The average ...
[90]
[PDF] Pessimism and the Assessment of Democratic Backsliding - V-Dem
Aug 25, 2025 · Our findings indicate that it is unlikely that global pessimism bias drives expert- based measures of worldwide democratic backsliding. More ...Missing: allegations | Show results with:allegations
[91]
Democracy data: how sources differ and when to use which one
Jun 17, 2022 · V-Dem treats democracy as a spectrum, with some countries being scored as more democratic than others. Other approaches instead treat democracy ...
[92]
Are non-democracies taking over? Autocratization and international ...
Aug 25, 2025 · Polity IV-based operationalizations count substantially fewer, Freedom House operationalizations substantially more autocratization ... V-Dem ...
[93]
Comparing measures of democracy: statistical properties ...
Mar 16, 2021 · This article provides novel information regarding the statistical properties, convergence, and interchangeability of some of the most frequently used measures ...