The Three Main Research Projects
The following section provides answers to the two questions below. They are organized by the program's three main projects (from the original proposal).

Project 1 – Conceptual and Empirical Structures
The aim of this project is to develop a refined and empirically validated scheme for 1) aggregating the 329 V-Dem indicators then using them 2) for aggregation of components, and components into 3) highly aggregated principles of democracy that can be used in various combinations to form full-fledged varieties of democracy. At the same time, inquiries into possible biases associated with coder characteristics, coding conditions (time on task, availability of information, etc), as well as robustness and uniqueness of expert-coders compared to other sources of data (e.g. so called "crowds") has generated new questions in need for additional research. These questions extend beyond what we originally planned for in this project in several ways. This new track and sub-project is being supported by supplementary grants from the Wallenberg Academy Fellowship program and Marcus and Marianne Walleberg foundation (PI: S. Lindberg in both cases).

Aggregaring the 329 V-Dem Indicators
The first task regarding item 1 above focused on developing statistical tools to assess measurement error in the raw coder-level data and to address issues of scale incompatibility in the subjective measures of regime characteristics. That is, assuring that the measures we provide capture theoretical concepts in identical ways across space and time. Dealing with these issues is fundamental to the wider goals of the research project because the substantive conclusions drawn from the project will hinge on the trustworthiness of the underlying data. The dataset now covers 180 countries from 1900 to 2019 and building on over 3,000 expert coders to code ˜200 variables measured at ordinal scales (i.e. rate a given country-year on a variable on a 0-1/0-3/0-4/0-5 scale). Since they are latent (concepts that are not observed directly), a team of methodologists (Pemstein, Marquardt, Wang, Tzelgov, Medzihorsky) has developed statistical tools that allow us to produce reasoned estimates of measurement error, largely solving the first of these issues. The team has developed a unique, Bayesian IRT-based measurement model aggregating experts' responses and generating country-level estimates. The model assumes that raters' are unbiased in their perception of the latent variable, but that each specific rating is erroneous. It also estimates rater-specific thresholds (i.e. the cutpoints between the different levels of latent variable), and a rater-specific reliability parameter. We utilize a hierarchical modeling strategy, in which we cluster raters' thresholds based on their main country of expertise. We report estimates of confidence in our country level variables of interest, as well as the probablity that each country-year falls into a given ordinal category. We have also made substantial headway on the (cross-national) scale incompatibility problem by developing techniques to efficiently add cross-national "bridging" information to the dataset. As a result, the face validity of the dataset has improved substantially.

Especially noteworthy in this regard is work by Project Managers Seim and Pemstein in developing vignettes. Anchoring vignettes are descriptions of specific, but hypothetical - or at least unnamed - cases that provide the information required to answer a certain question. In the context of V-Dem, they are descriptions of hypothetical country-years that focus on describing the country's status specific to one V-Dem indicator. Coders' ratings of the hypothetical cases, once combined, provide information about differences in how they translate concrete aspects of cases into ordinal ratings. Vignettes is a powerful and efficient tool for addressing differential item functioning because they do not require coders to have expertise in additional cases, we can keep them consistent across coders, and we can construct them to provide high threshold variability. This strategy was piloted in 2015/2016 and implemented in full (vignettes for all levels of all ordinal response category questions, and all country experts answered a random selection of them) since 2016. We have now incorporated hundreds of thousands of vignettes observations into the measurement model, further refining the precision of our estimates and ensuring cross-national comparability.
This research has broad implications for a variety of studies that employ cross-national expert surveys. Ongoing work also focuses on making adjustments to our bridging approach, leveraging the Computerized Adaptive Testing literature, and explores ways to achieve cross-national comparability without over-taxing expert coders, by relying on pairwise ranking tasks. While these advances are helping us to validate the data that sit at the core this project, they also speak to measurement problems faced by social scientists more generally. We are answering methodological research questions that are fundamental to the development and assessment of cross-national expert surveys across a wide variety of substantive domains. Moreover, the statistical tools that we have developed have direct applications in other areas of political science, most notably measuring the ideology of legislators, judges, and voters.

Aggregation of components and principles of democracy
The second and third tasks in this project - construction of a conceptual and empirical scheme for components and democracy-indices – are complete. It involved an intensive study of the existing literature (normative/conceptual as wel as empirical) resulting in a consensus on most issues of a new conceptual and aggregation scheme. We have achieved a consensus within the project on aggregation formulas that combine a weighted average of each component with a weighted product of all the components, leaving us with a thorny issue unresolved: how much should electoral aspects of democracy weight in measurement of concepts of other varieties of democracy such as participatory, egalitarian, or liberal democracy. We then used structured surveys of V-Dem PMs (project managers) and PIs (principal investgators) in an attempt to improve on the status quo measures of polyarchy and other high level indices. This involved the creation of candidate formulas for aggregating from status quo middle level indices to high level indices, the creation of visualizations showing the results of those formulas, and staging votes/surveys of the PM and PIs to assess preferences over these options, lead by Glynn and Seim. The top down project is largely complete at this time. The results so far indicate that, the V-Dem measurement approach to capturing democracy have high face validity and produce high-quality measures of the different conceptions, also taking measurement uncertainty into account.

As a result, the V-Dem dataset is now complete with five main measures of varieties of democracy: electoral, liberal, participatory, deliberative, and egalitarian democracy. For the latter four, we also have main component indices focusing on the distinct liberalness, participatoryness, etc. Down the disaggregation-tree, these components are broken down into sub-components that also have their own indices. The conceptual tree for varieities of democracy is attached to every dataset download.

Two manuscripts on the democracy indices have been published as working papers (#6 and #45) and in the form of a peer reviewed article in International Political Science Review as well as an article in Journal of Democracy. Two chapters in the V-Dem volume with CUP 2020, are devoted to the conceptual distinctions and aggregation, respectively. With respect to the measurement of electoral democracy, or what in the literature is called "polyarchy", the V-Dem data has proven several advantages over extant measures. We can for the first time allow for an analysis of how the different components of polyarchy, such as competition and inclusion, hang together empirically. We also allow for both minimalist and maximalist versions of electoral democracy and can thus systematically explore for the first time how these different conceptions affect our estimates of democracy in the world. An article on the construction of the polyarchy measure has been published in SCID.

The conceptual and measurement approach to V-Dem's egalitarian principle of democracy has achieved special attention. In a paper that has been published in Political Science Research Methods, Sigman and Lindberg have demonstrated how existing measures of democracy such as Polity and Freedom House, do not empirically capture egalitarian concepts such as the extent to which rights and liberties are protected equally across the population and the extent to which citizens are provided equally with resources – health, education, etc. – that enable them to participate meaningfully in civic life. The Egalitarian Democracy HLI and associated sub-indices provide opportunities to measure the many equality-related concepts, beginning even with those articulated by Aristotle, that form an important – yet often forgotten – element of democratic theory. The development of these indices has also led to an incipient research project exploring the origins of egalitarian democracy across the world.

The diverse strategies to create and then validate new indices based on V-Dem data have been aligned and dissected in a book chapter on data validation within the V-Dem project in the Cambridge University Press book co-authored by V-Dem PIs and PMs (with Seim as lead author).
Bernhard (with others) also wrote a paper that explored the way in which a broad range of democracy indicators including the V-Dem polyarchy measure could be used to capture different forms of regime change. The paper showed that the choice of democracy indicator and the way it was used to create regime change indicators led to radically different test outcomes in an important literature on international relations on regime change and conflict and is published in " International Interactions. Lührmann, Tannerberg, and Lindberg have developed an ordinal regime measure based on the V-Dem indices – Regimes of the World (RoW) published in Politics and Governance.

We have also largely completed the other aspects of Project 1 with occilliary indices of democracy. One concerns the measurement regime types, specifically the way the executive is appointed and dismissed (Teorell & Lindberg in Perspectives on Politics) and differentiate between different types of authoritarian regimes (such as single-party, military and monarchic dictatorships), and that these regime types have consequences beyond what can be explained solely in terms of the underlying level of democracy. Bernhard (with others) has also created a new composite indicator, the Core Civil Society Index, to measure of the robustness of civil society, suitable for use in cross-national time series research designs. As a result, the world will for the first time have a coherent measure of this idea with comprehensive country and year coverage. An article introducing the measure and other aspects of the civil society battery was published in Perspectives on Politics.

We have also innovated regarding the measurement of corruption (McMann, Seim, Teorell and Lindberg), an area related to yet distinct from democracy. Our tentative results indicate that the V-Dem project has been able to produce original data on corruption not only comparable across countries but also over long stretches of time going back to the early 20th century. The latter is an entirely novel contribution and will result not only in an index of corruption that for the first time provide comprehensive coverage of corruption before 1996, but also a set of valid measures beyond summary indices now forthcoming in Political Analysis.

Sundström (with others) created a three-dimensional index measuring women's political empowerment. This index captures three dimensions of women's political empowerment, including women's civil liberties, civil society participation, and political participation. Compared to existing indices measuring similar phenomena, such as the GDI, the GEM, the GII and the CIRI data, the V-Dem index allows more precise measurement and is superior in both spatial and temporal coverage for especially lower-income countries.It is published as an article in World Development.
Mechkova, Lührmann, and Marquardt developed a set of unique indices of vertical-, horizontal-, and diagonal accountability and then aggregated then together in a main accountability index using conventional Bayesian factor analysis, and then developed a Bayesian structural equation model for simultaneous estimation published in American Political Sceince Review. These indices are in extremely high demand from not only the academic community but also among policy and practitioners where accountability is a major concept of interest.

Project 2 – Endogenous Democratization
In Project 2, the research program investigates the unexplored area of "endogenous" patterns democratization. We are doing so by focusing on four closely related sub-projects:

2A) Sequences and Family Trees
Within the sequencing sub-project, Lindenfors, Lindberg, and Wang (with others) have identified and tested a number of already existing methods, gone down quite a few dead ends in terms of method development inspired by DNA sequence analyses such as 'social sequence analysis' (Abbott 1995, Abbot and Tsay 2000), multi-state problems (Gauthier et al. 2010), QCA (Ragin 1987, Rihoux and Ragin 2009) and Bayesian dynamical systems approach (Spaiser et al. 2014; Ranganathan et al. 2014). For our purposes, however, we need to investigate temporal relationships between pairs of comparable variables scored on a similar scale along a time series, as this is how the V-Dem data is structured. We have developed a graphical approach that, if paired with frequency counts of state combinations, can be used to identify when one variable consistently develops before another. By focusing on the changes, themselves we are able to identify the paths along which reform commonly occur. Going one step further, we also compare these observed reform paths to paths expected from the distribution of variables to identify 'more popular than expected' reform paths. This approach can be implemented and modified in several ways. The ground-work on themodel has been finished and was published as a working paper late 2015, with a revised version published in Political Science Research and Methods
Wang with others have applied the new method to investigate the relationship between civil liberties by gender and electoral democracy. These analyses show that civil rights for both men and women are important for a successful democratic transition. This is a particularly interesting observation given the failures of the 'Arab spring'. This first paper utilizing the new methods has been published in the European Journal of Political Research.

More specifically, we are currently working on establishing sequences between the variables that go into the Polyarchy composite variable. We have pushed the methods further and managed as a frist step to find significant differences between successful sequences of democratization (reaching high on the polyarchy/electoral democracy index) and failing ones (moving upwards but never really democratizing). There is also room for further method development, both concerning refinement of the graphical approach mentioned above, putting the developed methods on firmer statistical footing, and also designing complementary methods to fill methodological gaps. These developments led to the realization that much more effort is needed in this area to solve the research problem of detailing and then explaining sequences of democratization. The program as a whole and these first steps on the sequence agenda put Lindberg in a position to be granted for a European Research Council Consolidator Grant for the period 2017-2021 focusing specifically on "Failing and Successful Sequences of Democratization". This is a direct consequence of the current research program funded by Riksbankens Jubileumsfond, and should be considered a significant spin-off effect.

In another section of the sequencing-research efforts Bernhard, and Lindberg (with others) has completed analysis on two issues. The democratizing effect of sequences of iterated elections has been addressed using data 1900-2014 finding strong robust global and regional evidence of the democratizing effects. This was published in Democratization, with a separate article on authoritarian regime survival in Erupean Journal of Political Research.
In another paper (van Ham and Lindberg) investigates the sequencing of electoral manipulation, showing robust findings for Africa, that autocratic regimes gradually must switch from fraud and intimidation to votebuying as countries democratize, thus making vote buying a good sign, published in Government and Opposition.

2B) Democratization Waves/Diffusion
We now have the first findings for the second subproject. Coppedge (with others) have published a working paper with results focusing on diffusion of liberal democracy within colonial empires and occupations. V-Dem is the only democracy measurement project that rates colonial periods, making it possible to distinguish, for the first time, the impact of colonial rule on democratization before independence from the impact after independence. On average, over many decades, both rich and poor countries have tended to become more democratic; however, the rich countries have tended to democratize faster than the developing countries. But the colonizing countries have tended to democratize less rapidly than similarly high-income countries that never had colonies; and the colonies and former colonies have tended to democratize more rapidly than similarly low-income countries that were never colonies. The net effect of colonial rule is therefore to promote convergence especially in the oldest and longest-lasting empires.

2C) Reinforcing Critical Institutions
Bizzarro (with others) has focused on the causes and consequences of party system institutionalization. In a first paper, Bizzarro, Gerring, Knutsen, Bernhard, Skaaning, Coppedge, and Lindberg argue that the strength of political parties is a causal factor in subsequent economic growth. The publication in World Politics drawing on a novel measure of party strength from the V-Dem dataset, this theory is tested on data drawn from more than 150 countries, observed annually from 1900–2012. A sizeable effect is discovered, which is robust to various specifications, estimators, and samples. The effect operates in both democracies and autocracies and is fairly stable across regions and time periods.

We have also discovered that party system institutionalization along with the vibrancy of civil society plays a substantial role in explaining democratic consolidation and democratic breakdown. By using the party system institutionalization and civil society indicators in the dataset we provide robust evidence that these heretofore previously unmeasured institutional subsystems are robust predictors of democratic survival as hypothesized, published in SCID. The work in this area has also generated new questions. First, how do the party system and civil society interact to shape democratic stability? Are they substitutes for one another? Is there any combination of the two variables that is particularly problematic for or supportive of a stable democracy? Second, what are we to make of the finding that traditional predictors of democratic breakdown (e.g. economic crisis) are no longer significant once we include measures of party system institutionalization and civil society?

Staton (with others) have addressed the puzzle that although there is a positive association between judicial autonomy and expected compliance with judicial orders, there are many places in the world where we observe relatively autonomous judging yet considerable non-compliance (e.g., Israel, Colombia). It indicates that a reasonable degree of non-compliance serves judicial and government goals, and as such can be quite compatible with a healthy judiciary. Other questions we are asking include: 1) what is the causal effect of institutions that promote gender equality on the distribution of genders on the world's high courts, 2) what types of constitutional institutions (if any) promote independent judging, 3) what is the mechanism through which judicial independence influences democratic regime survival, and 4) what role do courts play in the stabilization of autocratic regimes? As a first step, a unique data collection has been done and completed the capture of well over 2,000 constitutional processes for the appointment and removal of judges on high courts since 1900. This is has required translating documents from multiple languages, including Spanish, French, German, Portuguese, Dutch, Polish, Russian, Romania, Finnish and Norwegian. We are finishing a final process of reconciling differences in coding among coders which will be complete by the end of the summer 2017. We have also made considerable progress on a project to capture gender diversity on high courts around the world from 1960 to the present. A first paper (Arrington, Nancy, Leeann Bass, Adam Glynn, Jeffrey K. Staton, Brian Delgado and Staffan Lindberg. 2017. "Gender Diversity on Peak Court.") has ban R&R in APSR. Staton et al. has also a book manuscript "Bending but not Breaking: Rule of law tensions and regime survival" accepted for publication by CUP.

Altmanhas also published a single author book manuscript, entitled "Hacking Democratic Malaise: The Choice of Direct Democracy." He argues that democracy is not a static state, but a constant exercise of fine-tuning. When representative government ceases to function as expected and there is a demand for change, two (non-mutually-exclusive) paths for improvement seem plausible that would allow democracies to retain their polyarchical foundations of freedom and equity among citizens. Hacking Democratic Malaise also offers a new institutional setting aimed at enlarging public views on a given contentious topic when a popular vote is on the table, offering an alternative to both sides of the policy under scrutiny.

2D) Territorial Sequencing
In a first paper, McMann (with others) investigates the causes of subnational unevenness in democratization with important empirical, theoretical, and methodological results. Empirically, the paper reveals for the first time that unevenness exists in all regions of the world, but is rare in Western Europe and North America. It also reveals that subnational unevenness existed historically and is not limited to countries with federal systems of government. These findings contribute significantly to previous research that has been limited to studies of a small number of countries examined during the contemporary area. Theoretically, the paper unlikely previous research, identifies distal causes of subnational unevenness, related to state incapacity and societal heterogeneity. Methodologically, as the first crossnational, crosstemporal study of this issue, the paper offers a benchmark model that can serve as a foundation for future work on a topic scholars are beginning to explore on a global scale. It is now published in Comparative Politics.
Project 3 – Factors External to Democracy
Project 3 seeks to establish which structural factors explain democratization in general and transitions to the variety of democracies we find in the world today. External factors include geography, inequality, modernization, demography, natural resources, religion, culture, and colonial heritage.

The influence of geography on international development has been widely heralded in recent years. Scholars to explore the relationship between geography and democracy only recently. Wig, Tollefsen, and Gerring focus on water as a mode of transport. They argue that the existence of natural waterways – navigable rivers or oceans – fosters changes in society over time, including the rise of cities, commerce, the middle class, and human capital. These changes, in turn, affect the style of governance – states that are smaller and/or more decentralized; political cultures characterized by greater emphasis on tolerance, meritocracy, and inclusion; the ability to overcome coordination problems among citizens, and thus to solve collective action dilemmas; enhanced leverage for citizens vis-à-vis their governments; and a greater emphasis on consent (rather than coercion) in the affairs of state. This argument is tested by looking at the relationship of waterways to regimes, measured by the V-Dem Polyarchy index, at the country level and the grid-cell level (defined by 50x50 km cells).

Gerring, Knutsen and Skaaning (with others) have completed a publication in European Journal of Political Research on modernization and democracy – one of the central issues in democratization studies. The ongoing debate concerns not only whether modernization affects democracy, but also – conditional on the possibility of such a relationship – whether the effect is unidirectional or bi-directional, whether it varies by historical era or according to the colonial history of a country, and what the causal mechanisms might be. We explore the insights gained from disaggregation. If modernization affects democracy it stands to reason that it might affect some aspects of democracy more than others. The approach is made possible by the V-Dem dataset, which tracks features of democracy across 400+ variables. We find that the relationship between modernization and democracy is most robust with respect to the electoral component of democracy. This, rather than enhanced state capacity or some other channel of influence, appears to be the intermediate factor that accounts for the strong connection between per capita GDP and electoral democracy. Other aspects of democracy are not as clealy, or even not at all, related to national income.
In explaining the shape of political institutions, scholars rarely look at demographic factors. Gerring and coauthors take on the question of why the exercise of political power is highly concentrated in some polities and widely dispersed in others. We argue that one persistent causal factor is demographic. Populous polities are characterized by less concentrated structures of authority. To explain this relationship, we use two mechanisms: efficiency and trust. The theory is demonstrated with a wide variety of empirical measures and in two settings: (1) cross-country analyses including most sovereign states and extending back to the 19th century and (2) within-country analyses focused on states, counties, and localities in the United States. V-Dem provides many of the cross-country indicators for this analysis of demography and institutions, presented in working paper #29.

In a working paper (#9), currently under review, Gerring and coauthors argue that electoral competition incentivizes politicians to provide public goods and services, and these, in turn, save lives. Hence, we argue that the electoral aspect of democracy should have a substantial impact on human development – while other aspects of democracy should be less consequential. Drawing on the V-Dem dataset, which provides a highly differentiated set of democracy indicators, and a new collection of mortality data, we conduct panel analyses that include most countries from 1900 to 2012 – a much more extensive sample than previous studies. We find that only indices focused on the electoral components of democracy yield a highly robust relationship to indicators of human development. This confirms the central role of competitive elections in enhancing quality of life.
Relatedly, Wang (with others) published an article in Political Research Quarterly exploring the relationships between various dimensions of democracy and several health outcomes (mortality rates, life expectancy, and public health expenditure) finding that the electoral principle of democracy especially has positive effects on the improvement of health. Importantly, we find that the positive effects are primarily found in poorer countries. That the establishment of electoral institutions is especially necessary for the improvement of health in poor countries is an important finding with clear policy implication.

Altman (with others) revisited the "democratic peace" with V-Dem data. If there is a sentence that reaches almost the character of a mantra in our discipline is that democracies do not fight each other. Some scholars even mention that this is virtually an empirical law (Levy 1989, 270). Nonetheless, they are well aware that this relationship might be contingent on the definitions used, both regarding democracy as well as conflict/war/fight. This research studies how far the idea that democracies do not fight each other goes in terms of democratic levels. In doing so, they explicitly leave aside previously used measures of democracy and supplement it with V-Dem. Unlike previous studies, they do not set an arbitrary cutting point between democracies and non-democracies, but rather they let the data talk by themselves. It is published in Journal of Peace Research.