A dataset containing all (or nearly all) extant democracy indexes.

democracy

## Format

An object of class tbl_df (inherits from tbl, data.frame) with 28492 rows and 86 columns.

## Overview

This dataset contains most democracy indexes in use today in a single panel (country-year) file. It attempts to be as comprehensive as possible, incorporating rarely used democracy indexes not easily available in country-year format.

The measurement of democracy is complicated and controversial. This dataset makes few judgments about what counts as a measure of democracy, as long as it has been used in scholarly work. It thus includes dichotomous, trichotomous, ordinal, and continuous indices; indices that focus primarily on the "competition" and indices that focus on the "participation" dimension of democracy; "thick" indices that attempt to measure a wide variety of characteristics plausibly attributed to democracy; and "minimalist" indexes that restrict themselves to the bare minimum of competition. Four broad families of democracy measures can be distinguished:

## Dichotomous indexes of democracy

These are indexes designed to distinguish between democracy and non-democracy. Most of them follow a fairly minimalist conception of democracy, focused on political competition while giving little weight to the extent of the suffrage or thicker civil or economic rights. These include the Boix, Miller, and Rosato (2012) indicator of democracy (bmr_democracy, bmr_democracy_omitteddata, and bmr_democracy_femalesuffrage); the Bernhard, Nordstrom and Reenock index of democracy (bnr and bnr_extended), originally developed for event history analysis (Bernhard, Nordstrom, and Reenock 2001); Renske Doorenspleet's indicator of democracy (doorenspleet), based on the Polity III data and a measure of the extent of suffrage (Doorenspleet 2000); Freedom House's list of electoral democracies (fh_electoral); the Cheibub, Gandhi, and Vreeland (2013) extension of the PACL (Przeworski, Alvarez, Cheibub and Limongi) democracy/dictatorship dataset (pacl); Jay Ulfelder's indicator of democracy (ulfelder_democracy and ulfelder_democracy_extended), based on a recoding of the Polity IV data (Ulfelder 2012); a calculated measure of democracy from the PIPE (Prezeworski et al 2010) dataset (PIPE_democracy - use with care as it may be incorrectly calculated) and a number of dichotomous democracy/non-democracy indicators (gwf_democracy, gwf_democracy_extended, kailitz_binary, magaloni_democracy, magaloni_democracy_extended, svolik_democracy, utip_dichotomous, utip_dichotomous_strict, reign, wth_democ1, and wth_democrobust) from datasets concerned with the identification of authoritarian regime types (Bell 2016; Geddes, Wright, and Frantz 2014; Svolik 2012, Wahman, Teorell, and Hadenius 2013, Kailitz 2013). These are all highly but not perfectly correlated with one another, with correlation coefficients ranging from 0.69 (magaloni_democracy_extended and doorenspleet) to 0.95 (reign_democracy and gwf_democracy_extended_strict)

Of these, only Doorenspleet's and Bernhard, Nordstrom, and Reenock's measures give special weight to the extent of suffrage in determining whether a country is democratic; not surprisingly, they display the lowest correlation with the other dichotomous indicators of democracy. All of these indexes have world-wide spatial scope, though they differ greatly in temporal coverage. doorenspleet, bmr_democracy, reign, and the "extended" versions of magaloni_democracy, gwf_democracy, and ulfelder_democracy have the widest temporal scope, going all the way back to the 19th century and in some cases beyond.

## Trichotomous indexes of democracy

Trichotomous indexes of democracy distinguish explicitly a "hybrid" or "semi-democratic" category. These include the measure of democracy in Central America developed by Bowman, Lehoucq, and Mahoney (blm, see Bowman, Lehoucq, and Mahoney 2005), the trichotomous measure of democracy in Latin America by Mainwaring, Brinks, and Perez Linan (mainwaring), the "Political Regime Change" dataset (prc, omitting the code for "transition" in the original dataset) described in Gasiorowski 1996 and extended in Reich 2002; and a couple of measures taken from datasets of autocratic regimes that explicitly distinguish between full democracies, electoral or multiparty autocracies, and other autocracies (utip_trichotomous, kailitz_tri). Both blm and mainwaring are regional indexes developed by scholars with Latin-American expertise; the rest have world-wide scope. These correlate with each other at levels ranging from 0.70 (kailitz_tri and blm) to 0.9 (blm and mainwaring).

## Ordinal indexes of democracy

These indexes distinguish more than three "degrees" of democracy, but they are not explicitly continuous. They include Freedom House's 14 category freedom index (fh_total_reversed); the 7 category Lexical Index of Democracy (lexical_index, from Skaaning, Gerring, Bartusevicius 2015); a 5 category indicator of democracy used by the Political Instability Task Force based on the parcomp and exrec variables of the Polity dataset (pitf, described in Goldstone et al 2010); the Polity IV polity2 variable; a calculated variable from Przeworski 2010 (PIPE_regime - use with care, as it may be incorrectly calculated) Coppedge and Reinicke's Polyarchy index (polyarchy_original_contestation and polyarchy_original_polyarchy, from Coppedge and Reinicke 1991). Of these, the Polyarchy measures are specialist measures that only cover 357 country-years, but the rest have worldwide scope, polity and lexical_index go all the way back to the beginning of the 19th century.

Most of these indexes are meant to capture "thicker" conceptions of democracy. The Freedom House measure puts some emphasis on civil and political rights; Polity and PITF focus on differences in "authority patterns" (though they tend to downplay the scope of participation); Polyarchy tries to operationalize both the participation and contestation dimensions of Dahl's "Polyarchy" concept; and LIED tries to incorporate the degree of suffrage using measures originally collected for the PIPE dataset (Prezeworski et al 2010). They correlate with one another at levels ranging from 0.81 (lexical_index and pitf) to 0.94 (polity2 and pitf).

## Continuous indexes of democracy

These indexes conceptualize democracy as a continuous quantity, and usually in "thicker" ways. They include Arat's measure of democracy (pmm_arat, from Arat 1991); Bollen's index of democracy (pmm_bollen, from Bollen 2001); the World Governance Indicator's Index of voice and Acocuntability (wgi_democracy); Axel Hadenius' index of democracy, from Hadenius 1992 (pmm_hadenius); Munck's measure of democracy (pmm_munck); several variants of the Participation-Enhanced Polity Scores (e.g., PEPS1v, from Moon et al 2006); the calculated inclusion and contestation dimensions from Coppedge, Alvarez, and Maldonado (2008) (polyarchy_inclusion_dimension and polyarchy_contestation_dimension); several indexes from V-dem (e.g. v2x_polyarchy); the posterior means and medians for the three releases of the Unified Democracy scores (uds_*, Pemstein, Meserve, and Melton 2010) and Vanahnen's index of democratization (vanhanen_democratization). Of these, the Unified democracy indexes, the V-Dem indexes, and the Polyarchy dimensions are themselves latent variable indexes.

Vanhanen and PEPS give special weight to participation in the measurement of democracy; the WGI index combines a very wide variety of indicators of democracy. Only Vanahanen, the V-dem measure, and PEPS have long temporal coverage; the WGI Voice and Accountability index has broad spatial coverage but is only available for a short time series. Arat and Bollen go back only until the 1950s, though they do have reasonable spatial coverage.

Correlations vary from a low of 0.26 (Munck and the WGI, 126 country-year overlap, nonsignificant) to over 0.97 for various for V-Dem indexes. The lowest correlations are typically with the WGI measure of democracy.

## Spatial and temporal organization

The state system is complicated, and not always amenable to presentation in tabular form. Countries change name, split, get absorbed into larger units, and are not always obviously independent. This dataset basically follows the Gleditsch and Ward list of independent states (Gleditsch and Ward 1999, updated by Gleditsch to 2013), supplemented by Gleditsch's tentative list of microstates (available at http://privatewww.essex.ac.uk/~ksg/statelist.html). The Gleditsch and Ward list is very similar, but not identical, to the Correlates of War list of independent states that is commonly used in Political Science and International Relations research; in particular, the treatment of Germany, Yemen, Vietnam, Ethiopia, Yugoslavia, Serbia, Montenegro, and a few other countries and their successor states differs between them.

I have included two different codes for each country-year: the Gleditsch and Ward country code (GWn), and the COW code (cown), as well as an indicator of whether the country year is considered to be "in the international system" by Gleditsch and Ward (in_GW_system). Since the Gleditsch and Ward "international system" begins with the Congress of Vienna in 1816, country-years before 1816 are by definition not "in system," even though this does not mean that these states were not independent then. Care should thus be taken with the "in system" indicator for years before 1816.

It is worth noting that while most measures of democracy are produced only for sovereign countries, a number of measures have also been produced, either explicitly (the V-dem project, Freedom House) or not (sometimes researchers do not agree on whether a country-year represents a "sovereign" state), for non-sovereign territories. In fact in this dataset about 6415 country-years in or after 1816 are not "in system" (most of them from the V-dem indexes of democracy); these are easily excluded by filtering the dataset using the boolean in_GW_system indicator (or a logical expression that filters data points that are not in system after 1816 but includes those before 1816).

All countries are measured as of 31 December of the given year, as is the conventional rule. Some datasets do not use this rule (e.g., Freedom House, Geddes, Wright, and Frantz 2014), and others sometimes contain more than one data point per year (e.g., the Political Regime Change dataset); I have tried my best to transform their coding so that all datapoints conform to convention, but some errors may remain.

## A note on sources

Wherever possible, I have used the original sources for these measures of democracy and put them in the right panel format myself. For details, see the democracyData package. There are three exceptions.

First, in a few cases I have relied on the replication data for Pemstein, Meserve, and Melton (2010), which contains data I have not been able to find elsewhere. These variables are marked pmm_*, and they are not always perfectly correlated with the original data (where available). For example, fh_pmm includes data for 1981, whereas fh_total_reversed conventionally does not; and there are a number of divergences between PMM's pmm_prc data and the prc dataset available from Reich 2002 that seem to have something to do with the treatment of country-years that include transition periods. A more extended discussion of the differences between the PMM replication data and the original data sources is available in a package vignette; see vignette("Differences_between_PMM_and_original_data") for details.)

Second, I have constructed the PIPE_democracy, PIPE_regime, and pitf* variables following the instructions in the source, but have not found the data ready-made. This is probably unproblematic in the case of pitf -- the instructions are very clear -- but less so in the case of przeworski, where the instructions are less clear and it is difficult to produce a test that indicates whether the variable is properly constructed. It is also worth noting that the Polity data is occasionally revised, so that the pitf variable described and used in Goldstone et al 2010 is probably different from the pitf variable here for a small number of country-years. (Similarly, pmm_polity and polity2 differ in a few instances due to revisions of the Polity data between 2010 and today.)

Third, on at least one occasion I have recalculated an index from its original sources in different ways (see the utip_dichotomous, utip_dichotomous_strict, and utip_trichotomous variables), since the original data does not provide an unambiguous measure.

## Standard descriptive variables (generated by this package)

extended_country_name

The name of the country in the Gleditsch-Ward system of states, or the official name of the entity (for non-sovereign entities and states not in the Gleditsch and Ward system of states) or else a common name for disputed cases that do not have an official name (e.g., Western Sahara, Hyderabad). The Gleditsch and Ward scheme sometimes indicates the common name of the country and (in parentheses) the name of an earlier incarnation of the state: thus, they have Germany (Prussia), Russia (Soviet Union), Madagascar (Malagasy), etc. For details, see Gleditsch, Kristian S. & Michael D. Ward. 1999. "Interstate System Membership: A Revised List of the Independent States since 1816." International Interactions 25: 393-413. The list can be found at http://privatewww.essex.ac.uk/~ksg/statelist.html.

GWn

Gleditsch and Ward's numeric country code, from the Gleditsch and Ward list of independent states.

cown

The Correlates of War numeric country code, 2016 version. This differs from Gleditsch and Ward's numeric country code in a few cases. See http://www.correlatesofwar.org/data-sets/state-system-membership for the full list.

polity_ccode

The Polity IV country code, 2017 version. This differs from Gleditsch and Ward's numeric country code and COW in a few cases. See http://www.systemicpeace.org/inscrdata.html for the full list.

in_GW_system

Whether the state is "in system" (that is, is independent and sovereign), according to Gleditsch and Ward, for this particular date. Matches at the end of the year; so, for example South Vietnam 1975 is FALSE because, according to Gleditsch and Ward, the country ended on April 1975 (being absorbed by North Vietnam). It is also TRUE for dates beyond 2012 for countries that did not end by then, depsite the fact that the Gleditsch and Ward list has not been updated since.

year

The calendar year. Democracy measurements conventionally reflect the situation in the country as of the last day of the year.

## Democracy measures

anckar_democracy

The Anckar-Fredriksson 2018 measure of democracy, as a numeric value. Up to 2010 this should be identical to bmr_democracy_omitteddata. 0 = non-democracy, 1 = democracy.

arat_pmm

Democracy score from Arat 1991. Taken from Pemstein, Meserve, and Melton 2013. Min = 29, max = 109, n = 3873.

blm

Trichotomous measure of regime type from Bowman, Lehoucq, and Mahoney 2005. 0 = authoritarian, 0.5 = semidemocratic, 1 = democratic. Available only for five Latin American countries (Costa Rica, El Salvador, Guatemala, Honduras and Nicaragua) from 1900 to 2000.

pmm_blm

Same as previous but taken from Pemstein, Meserve, and Melton 2013. 0 = authoritarian, 0.5 = semidemocratic, 1 = democratic. Available only for five Latin American countries (Costa Rica, El Salvador, Guatemala, Honduras and Nicaragua) from 1946 to 2000 in this version. Included here in order to replicate the original model from Pemstein, Meserve, and Melton 2011.

bmr_democracy

Dichotomous measure of regime type from Boix, Miller, and 2012, version 3.0. 1 = democracy, 0 = non-democracy. N = 16987.

bmr_democracy_omitteddata

Dichotomous measure of regime type from Boix, Miller, and 2012, version 3.0. 1 = democracy, 0 = non-democracy. This is the same measure as bmr_democracy, except it records an NA for countries occupied during an international war (e.g., the Netherlands 1940-44) or experiencing state collapse during a civil war (e.g., Lebanon 1976-89). The democracy variable instead fills in these years as continuations of the same regime type. N = 16762.

bmr_democracy_femalesuffrage

According to the BMR version 3.0 codebook, this is the same measure as bmr_democracy, except that it also requires that at least half of adult women have the right to vote. 30 countries change values.

bnr

Dichotomous indicator of democracy from the Bernhard, Nordstrom & Reenock 2001.

bnr_extended

Dichotomous indicator of democracy from the Bernhard, Nordstrom & Reenock 2001. This version indicator has been put in country-year format, extending to 1913, with the help of the Correlates of War panel of independent states; independent countries (not microstates) in this panel that were not included in the original dataset are assumed to be non-democratic for the period. See democracyData for more information about the additional country-years generated in this way.

pmm_bollen

0-100 index of democracy from Bollen 2001. Taken from Pemstein, Meserve, and Melton 2013.

doorenspleet

Dichotomous index of democracy from Doorenspleet 2000. 1 = authoritarian, 2 = democracy. Omits periods of interruption.

eiu

The Economist Intelligence Unit's democracy index.

wgi_democracy

Voice and Accountability index from the World Governance Indicators. Taken from http://www.govindicators.org.

fh_total_reversed

Average civil liberties + political rights score (reversed so higher values are more democratic) from Freedom House. 2018. "Freedom in the World." Original data available at http://www.freedomhouse.org. Goes from 0 (least democratic) to 13 (most democratic). In this version, the index does not include a value for 1981. This is based on the latest Freedom House data going all the way to 2017.

pmm_fh

Same as previous but taken from Pemstein, Meserve, and Melton 2013. Goes from 1 (least democratic) to 7 (most democratic). In this version the index stops at 2008. It also includes a value for 1981. For more on the differences between this variable and the current FH Freedom in the World index, see democracyData.

fh_electoral

An indicator of whether a country is an "electoral democracy" in Freedom House's estimation (1 = yes, 0 - no). Original data available at http://www.freedomhouse.org. Available only from 1989. This is based on the latest Freedom House electoral democracies list going all the way to 2016.

gwf_democracy

Dichotmous democracy/autocracy indicator from Geddes, Wright, and Frantz 2014. 0 = autocracy, 1 = democracy.

gwf_democracy_extended

Dichotmous democracy/autocracy indicator from Geddes, Wright, and Frantz 2014. 0 = autocracy, 1 = democracy. Extended beyond 1945 using Geddes, Wright, and Frantz's case variable, which encodes information about the first year of the regime. For more detail on the resulting additional country-years, see democracyData.

0-10 index of democracy from Hadenius 1992. Taken from Pemstein, Meserve, and Melton 2013. Higher values are more democratic.

kailitz_binary

Dichotomous democracy indicator from Kailitz 2013. 0 = autocracy (all types including electoral autocracy), 1 = liberal democracy.

kailitz_tri

Trichotomous democracy indicator from Kailitz 2013. 0 = autocracy (all types except electoral autocracy), 1 = electoral autocracy, 2 = liberal democracy.

lexical_index

0-6 Lexical Index of Electoral Democracy from Skaaning, Gerring, and Bartusevicius 2015. This is from V3 of the dataset, updated to 2015. 0 = nonelectoral, 1 = one- or no- party elections, 2 = limited competition multiparty elections for legislature only, 3 = Limited competition multiparty elections for both executive and legislature, 4 = Competitive elections for executive and legislative, limited suffrage, 5 = Male democracy, 6 = Electoral democracy.

magaloni_democracy

Dichotomous democracy indicator from Magaloni, Chu, and Min 2013. 0 = autocracy (all types including multiparty autocracy), 1 = democracy.

magaloni_democracy_extended

Dichotomous democracy indicator from Magaloni, Chu, and Min 2013. 0 = autocracy (all types including multiparty autocracy), 1 = democracy. Extended beyond 1950 using the duration_nr variable of the original dataset, which encodes information about the first year of each regime. For more detail on the resulting additional country-years, see democracyData.

mainwaring

Trichotomous democracy indicator from Mainwaring, Brinks, and Perez Linan 2008. 0 = non-democracy, 0.5 = hybrid, 1 = democracy.

pmm_mainwaring

Same as previous but taken from Pemstein, Meserve, and Melton 2013. In this version the indicator goes only to 1945, and it is missing some country-years. -1 = non-democracy, 0 = hybrid, 1 = democracy. For more on the differences between this variable and the original Mainwaring et al data, see vignette("Differences_between_PMM_and_original_data"). Included in order to replicate the original PMM 2010 model.

pmm_munck

0-1 index of democracy from Munck 2009. Taken from Pemstein, Meserve, and Melton 2013. Only available for 342 country-years. Higher values are more democratic.

pacl

Dichotomous measure of democracy from Cheibub, Gandhi, and Vreeland 2010. 1= democracy, 0 = non-democracy.

pmm_pacl

Same as previous but taken from Pemstein, Meserve, and Melton 2013. This is missing a couple of countries in the original dataset; for more detail on the missing cases, see vignette("Differences_between_PMM_and_original_data").

PEPS1i

Participation-Enhanced Polity Score 1, polity score adjusted using IDEA Votes/Voting age population. From Moon et al. (2006).

PEPS2i

Participation-Enhanced Polity Score 2, polity score adjusted using IDEA Votes/Voting age population. From Moon et al. (2006).

PEPS1v

Participation-Enhanced Polity Score 1, polity score adjusted using Vanhanen votes/two-thirds of Vanhanen population. From Moon et al. (2006).

PEPS2v

Participation-Enhanced Polity Score 2, polity score adjusted using Vanhanen votes/two-thirds of Vanhanen population. From Moon et al. (2006).

PEPS1q

Participation-Enhanced Polity Score 1, polity score adjusted using (mostly)IDEA votes/Voting age population, with participation coded zero for noncompetitive elections. From Moon et al. (2006).

PEPS2q

Participation-Enhanced Polity Score 2, polity score adjusted using (mostly)IDEA votes/Voting age population, with participation coded zero for noncompetitive elections. From Moon et al. (2006).

pitf

A five category indicator of democracy described in Goldstone et al 2010. Can be: 0-Full autocracy (exrec < 7, parcomp !=0 and parcomp < 3) 1-Partial autocracy (exrec < 7, parcomp = 0 or parcomp > 2) 2-Partial democracy with factionalism (exrec > 6, parcomp = 3) 3-Partial democracy (exrec > 6, parcomp = 0 or parcomp = 4 or parcomp = 5 but exrec != 8) 4-Full democracy (exrec = 8, parcomp = 5). See Goldstone et al. 2010 for full details.

pitf_binary

A simplification of the pitf indicator of democracy described in Taylor and Ulfelder 2015. A country is a democracy (1) "if its chief executive is chosen in competitive elections (exrec equal to 7 or 8) and political competition is not suppressed (parcomp equal to 0 or parcomp greater than 2)" Otherwise it is a non-democracy (0).

polity

Annual polity index, excluding special codes for interruption, interregnum, and transition (-88,-77,-66). Higher values are more democratic. From Marshall, Gurr, and Jaggers 2015.

polity2

Annual polity2 index, interpolating values for interruption, interregnum, and transition periods. Higher values are more democratic. From Marshall, Gurr, and Jaggers 2015.

polity_pmm

Same as previous, but taken from Pemstein, Meserve, and Melton 2013. In this version the indicator only goes to 1945, and it differs from polity2 in a few cases. For more detail on the differences, see vignette("Differences_between_PMM_and_original_data"). Included only in order to be able to replicate the PMM 2011 model.

polyarchy_original_polyarchy

0-10 index of democracy from Coppedge and Reinicke 1991. Revised in 2003-2006. This has been reversed so that higher values are more democratic; the original index was scored so that 0 was less democratic. The codebook for the dataset suggests using polyarchy_original_contestation instead, a more reliable version of the polyarchy scale with fewer categories. Includes a value for Western Sahara in 2000, which has been assigned COW code 605; Western Sahara is not coded by any other dataset in this compilation, and is not considered an independent state by either Gleditsch and Ward or the Correlates of War project. Exclude if necessary.

pmm_polyarchy

Same as polyarchy_reversed but taken from Pemstein, Meserve, and Melton 2013. Note that there are 22 discrepancies between polyarchy_reversed and polyarchy_pmm; these appear to be due to transcription error in PMM's replication dataset. pmm_polyarchy also misses 4 country-years in polyarchy_original_polyarchy. For more detail on the differences, see vignette("Differences_between_PMM_and_original_data"). Included only in order to be able to replicate the PMM 2011 model.

polyarchy_original_contestation

1-9 index of contestation from the revised version of Coppedge and Reinicke. 1991. Revised in 2003-2006. Includes a value for Western Sahara in 2000, which has been assigned code 605; Western Sahara is not coded by any other dataset in this compilation, and is not considered an independent state by either Gleditsch and Ward or the Correlates of War project. Meaning of the scale is as follows: 9 Meaningful fair elections are held, there is full freedom for political organization and expression, and there is no preferential presentation of official views in the media. 8 Meaningful fair elections are held and there is full freedom for political organization and expression, but there is preferential presentation of official views in the media. 7 Meaningful fair elections are held and there is full freedom for political organization, but some public dissent is suppressed and there is preferential presentation of official views in the media. 6 Meaningful fair elections are held, but some independent political organizations are banned, some public dissent is suppressed, and there is preferential presentation of official views in the media. 5 Elections are marred by fraud or coercion, some independent political organizations are banned, some public dissent is suppressed, and there is preferential presentation of official views in the media. 4 Like score 5 except that there is less contestation in one or two of the following respects: no meaningful elections are held, only nonpolitical organizations are allowed to be independent, or alternatives to the official media are very limited. 3 No meaningful elections are held, only nonpolitical organizations are allowed to be independent, some public dissent is suppressed, and alternatives to the official media are very limited. 2 Like score 3 except that there is less contestation in one or two of the following respects: all organizations are banned or controlled by the government or official party, all public dissent is suppressed, or there is no public alternative to official information. 1 No meaningful elections are held, all organizations are banned or controlled by the government or official party, all public dissent is suppressed, and there is no public alternative to official information.

polyarchy_contestation_dimension

The contestation dimension (CONTEST) in the dataset described in Coppedge, Alvarez, and Maldonado 2008.

polyarchy_inclusion_dimension

The inclusion dimension (INCLUS) in the dataset described in Coppedge, Alvarez, and Maldonado 2008.

prc

Trichotomous index of democracy from Gasiorowski 1996. Available in updated form in Reich 2002. 1= Authoritarian, 3 = semidemocratic, 4 = democratic. Transitional categories coded NA.

pmm_prc

Same as previous but taken from Pemstein, Meserve, and Melton 2013. In this version the indicator only goes to 1945. It also differs from the original data in a few cases (including the fact that it includes the transition category, coded NA in prc). For more detail on the differences, see For more detail on the differences, see vignette("Differences_between_PMM_and_original_data"). Included only in order to be able to replicate the PMM 2011 model.

PIPE_regime

Calculated regime type variable from the PIPE dataset. Przeworski et al. 2013. This variable is not found in the downloadable dataset; it is calculated according to the instructions in the codebook. The documentation gives the following instructions: "0 if electoral_age (a variable in the original dataset) is missing. There are two distinct reasons why it may be missing: either elections are not held regularly or they are held but the winner does not complete a term in office or makes an autocoup. Elections are considered not to be held if (1) there is a constitution that specifies the length of term of the chief executive (or the legislature) and this period is exceeded by more than one year, except for war years, (2) there is no constitution or the constitution does not provide for elections and no elections are held during this period. 1 if electoral_age is not missing and opposition = 0 (a variable in the original dataset) (or republic_age is missing). These are regimes in which elections are held regularly and the winners (or their constitutional successors) complete electoral terms but either elections are uncontested or there is one party and independents or parties are banned and everyone runs as independent. Note that if everyone runs as independents it is possible for the incumbent government to lose an election, as in Swaziland in 1993. It is also possible for the incumbent to lose if the regime had no opposition up to some time when a competitive election occurred but the winner did not complete a term, as in Honduras in 1852. Finally, incumbents can lose when regime = 1 if the election was competitive but at the end of the year the opposition was suppressed, as in Panama in 1968. 2 if republic_age (a variable in the original dataset) is not missing, that is if electoral_age is not missing and opposition = 1 (there is some minimal opposition), but no strong alternation has ocurred (power has not changed hands). 3 if democracy_age (democracy_age is not found in the original dataset) is not missing, that is, the country has elections, there is opposition, and a strong alternation -- a change in parties controlling the government -- has occurred."

reign_democracy

A measure of democracy from Bell 2016, obtained by coding all presidential and parliamentary democracies as 1, all other regimes as 0.

csvmdi

The continuous Support Vector Machine democracy index from Grundler and Krieger 2018.

svmdi_2016

The continuous Support Vector Machine democracy index from Grundler and Krieger 2016.

dsvmdi

The dichotomous Support Vector Machine democracy index from Grundler and Krieger 2018.

svolik_democracy

Dichotomous indicator of democracy from Svolik 2012. 0 = authoritarian, 1 = democracy.

ulfelder_democracy

Dichotomous indicator of democracy from Ulfelder 2012. 0 = authoritarian, 1 = democracy.

ulfelder_democracy_extended

Dichotomous indicator of democracy from Ulfelder 2012. 0 = authoritarian, 1 = democracy, extended back in time using the regime duration information in the dataset. For more details on the additional country-years coded this way, see democracyData

utip_dichotomous

Calculated dichotomous index of democracy from data in the UTIP dataset of political regimes (Hsu 2008). 1 if the regime is a social democracy, conservative democracy, or one party democracy, 0 otherwise. The category of "one party democracy" is not well documented.

utip_dichotomous_strict

Stricter version of the calculated dichotomous index of democracy from data in the UTIP dataset of political regimes (Hsu 2008). 1 if the regime is a social democracy or a conservative democracy, 0 otherwise. This excludes "one party democracies" from the democracy category.

utip_trichotomous

Calculated trichotomous index of democracy from data in the UTIP dataset of political regimes (Hsu 2008). 2 if the regime is a social democracy or conservative democracy, 1 if the regime is a one party democracy, 0 otherwise. The category of "one party democracy" is not well documented.

v2x_api

Additive polyarchy index from V-dem version 7.1, Coppedge et al 2017.Higher values are more democratic.

v2x_delibdem

Deliberative democracy index from V-dem version 7.1, Coppedge et al 2017.Higher values are more democratic.

v2x_egaldem

Egalitarian democracy index from V-dem version 7.1, Coppedge et al 2017.Higher values are more democratic.

v2x_libdem

Egalitarian democracy index from V-dem version 7.1, Coppedge et al 2017.Higher values are more democratic.

v2x_mpi

Multiplicative polyarchy index from V-dem version 7.1, Coppedge et al 2017.Higher values are more democratic.

v2x_partipdem

Participatory democracy index from V-dem version 7.1, Coppedge et al 2017.Higher values are more democratic.

v2x_polyarchy

Continuous polyarchy index from V-dem version 7.1, Coppedge et al 2017.Higher values are more democratic.

vanhanen_competition

Index of competition from Vanhanen 2012. From Vanahnen's documentation: "The smaller parties' share of the votes cast in parliamentary or presidential elections, or both, is used to indicate the degree of competition. It is calculated by subtracting the percentage of votes won by the largest party from 100. If the largest party gets, for example, 40 percent of the votes, the share of the smaller parties is 60 percent. If data on the distribution of votes are not available, the value of this variable is calculated on the basis of the distribution of seats in parliament. The distribution of seats is used also in cases in which it seems to indicate power relations more realistically than the distribution of votes." Its maximum value is 70. See the full documentation for Vanhanen's dataset for details.

vanhanen_participation

Index of participation from Vanhanen 2012. From Vanahnen's documentation: "The percentage of the population which actually voted in the same elections is used to measure the degree of participation (= Participation). This percentage is calculated from the total population, not from the adult or enfranchized population." It is zero by construction in cases where no popular elections exist. May be modified by referenda. See the full documentation of Vanhanen's dataset for details.

vanhanen_democratization

Index of democratization from Vanhanen 2012. Higher values are more democratic. Constructed multiplicatively from vanhanen_participation and vanhanen_competition.

pmm_vanhanen

Same as vanhanen_democratization but taken from Pemstein, Meserve, and Melton. 2013. There are some small differences between this variable and vanhanen_democratization. For more detail on the differences, see vignette("Differences_between_PMM_and_original_data"). Included only in order to be able to replicate the PMM 2011 model exactly.

wth_democ1

Dichotomous measure of democracy from Wahman, Teorell, and Hadenius 2013, obtaining by coding 1 all democracies according to the regime1ny variable, 0 all other regimes.

wth_democrobust

Dichotomous measure of democracy from Wahman, Teorell, and Hadenius 2013, obtaining by coding 1 all democracies according to the regimenyrobust variable, 0 all other regimes.

## References

Anckar, Carsten and C. Fredriksson. 2018. "Classifying political regimes 1800-2016: a typology and a new dataset". European Political Science. DOI: 10.1057/s41304-018-0149-8. Data and codebook available at https://doi.org/10.1057/s41304-018-0149-8.

Arat, Zehra F. 1991. Democracy and human rights in developing countries. Boulder: Lynne Rienner Publishers.

Bernhard, Michael Timothy Nordstrom, and Christopher Reenock, "Economic Performance, Institutional Intermediation and Democratic Breakdown," Journal of Politics 63:3 (2001), pp. 775-803. Data and coding description available at http://users.clas.ufl.edu/bernhard/content/data/data.htm

Bell 2016. The Rulers, Elections, and Irregular Governance Dataset (REIGN). OEF research. Data available at http://oefresearch.org/datasets/reign

Boix, Carles, Michael Miller, and Sebastian Rosato. 2012. A Complete Data Set of Political Regimes, 1800-2007. Comparative Political Studies 46 (12): 1523-1554. Original data available at https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/FJLMKT

Bollen, Kenneth A. 2001. "Cross-National Indicators of Liberal Democracy, 1950-1990." 2nd ICPSR version. Chapel Hill, NC: University of North Carolina, 1998. Ann Arbor, MI: Inter-university Consortium for Political and Social Research, 2001. Original data available at http://webapp.icpsr.umich.edu/cocoon/ICPSR-STUDY/02532.xml.

Bowman, Kirk, Fabrice Lehoucq, and James Mahoney. 2005. Measuring Political Democracy: Case Expertise, Data Adequacy, and Central America. Comparative Political Studies 38 (8): 939-970. http://cps.sagepub.com/content/38/8/939. Data available at http://www.blmdemocracy.gatech.edu/.

Cheibub, Jose Antonio, Jennifer Gandhi, and James Raymond Vreeland. 2010. "Democracy and Dictatorship Revisited." Public Choice. 143(1):67-101. Original data available at https://sites.google.com/site/joseantoniocheibub/datasets/democracy-and-dictatorship-revisited.

Coppedge, Michael, John Gerring, Staffan I. Lindberg, Svend-Erik Skaaning, and Jan Teorell, with David Altman, Michael Bernhard, M. Steven Fish, Adam Glynn, Allen Hicken, Carl Henrik Knutsen, Kelly McMann, Pamela Paxton, Daniel Pemstein, Jeffrey Staton, Brigitte Zimmerman, Frida Andersson, Valeriya Mechkova, Farhad Miri. 2017. V-Dem Codebook v7.1. Varieties of Democracy (V-Dem) Project. Original data available at https://v-dem.net/en/data/.

Coppedge, Michael, Angel Alvarez and Claudia Maldonado. 2008 "Two Persistent Dimensions of Democracy: Contestation and Inclusiveness".The Journal of Politics 70(03), pp. 632-647. DOI: 10.1017/S0022381608080663. Original data available at https://www3.nd.edu/~mcoppedg/crd/datacrd.htm.

Coppedge, Michael and Wolfgang H. Reinicke. 1991. Measuring Polyarchy. In On Measuring Democracy: Its Consequences and Concomitants, ed. Alex Inkeles. New Brunswuck, NJ: Transaction pp. 47-68. Original data available at https://www3.nd.edu/~mcoppedg/crd/datacrd.htm.

Doorenspleet, Renske. 2000. Reassessing the Three Waves of Democratization. World Politics 52 (03): 384-406. DOI: 10.1017/S0043887100016580. http://dx.doi.org/10.1017/S0043887100016580.

Freedom House. 2018. "Freedom in the World." Original data available at http://www.freedomhouse.org.

Gasiorowski, Mark J. 1996. "An Overview of the Political Regime Change Dataset." Comparative Political Studies 29(4):469-483.

Geddes, Barbara, Joseph Wright, and Erica Frantz. 2014. Autocratic Breakdown and Regime Transitions: A New Data Set. Perspectives on Politics 12 (1): 313-331. Original data available at http://dictators.la.psu.edu/.

Gleditsch, Kristian S. & Michael D. Ward. 1999. "Interstate System Membership: A Revised List of the Independent States since 1816." International Interactions 25: 393-413. The list can be found at http://privatewww.essex.ac.uk/~ksg/statelist.html

Goldstone, Jack, Robert Bates, David Epstein, Ted Gurr, Michael Lustik, Monty Marshall, Jay Ulfelder, and Mark Woodward. 2010. A Global Model for Forecasting Political Instability. American Journal of Political Science 54 (1): 190-208. DOI:10.1111/j.1540-5907.2009.00426.x

Grundler, K. and T. Krieger. 2018. "Machine Learning Indices, Political Institutions, and Economic Development". Report. CESifo Group Munich, 2018. https://www.cesifo-group.de/DocDL/cesifo1_wp6930.pdf.

Grundler, K. and T. Krieger. 2016. "Democracy and growth: Evidence from a machine learning indicator". European Journal of Political Economy 45, pp. 85-107. DOI: https://doi.org/10.1016/j.ejpoleco.2016.05.005.

Hadenius, Axel. 1992. Democracy and Development. Cambridge: Cambridge University Press.

Hadenius, Axel & Jan Teorell. 2007. "Pathways from Authoritarianism", Journal of Democracy 18(1): 143-156.

Hsu, Sara "The Effect of Political Regimes on Inequality, 1963-2002," UTIP Working Paper No. 53 (2008), http://utip.gov.utexas.edu/papers/utip_53.pdf. Data available for download at http://utip.gov.utexas.edu/data/.

Kailitz, Steffen. 2013. Classifying political regimes revisited: legitimation and durability. Democratization 20 (1): 39-60. Original data available at http://dx.doi.org/10.1080/13510347.2013.738861.

Mainwaring, Scott, Daniel Brinks, and Anibal Perez Linan. 2008. "Political Regimes in Latin America, 1900-2007." Original data available from http://kellogg.nd.edu/scottmainwaring/Political_Regimes.pdf.

Magaloni, Beatriz, Jonathan Chu, and Eric Min. 2013. Autocracies of the World, 1950-2012 (Version 1.0). Dataset, Stanford University. Original data and codebook available at http://cddrl.fsi.stanford.edu/research/autocracies_of_the_world_dataset/

Marshall, Monty G., Ted Robert Gurr, and Keith Jaggers. 2012. "Polity IV: Political Regime Characteristics and Transitions, 1800-2012." Updated to 2015. Original data available from http://www.systemicpeace.org/polity/polity4.htm.

Moon, Bruce E., Jennifer Harvey Birdsall, Sylvia Ceisluk, Lauren M. Garlett, Joshua J. Hermias, Elizabeth Mendenhall, Patrick D. Schmid, and Wai Hong Wong (2006) "Voting Counts: Participation in the Measurement of Democracy" Studies in Comparative International Development 42, 2 (Summer, 2006). The complete dataset is available here: http://www.lehigh.edu/~bm05/democracy/Obtain_data.htm.

Munck, Gerardo L. 2009. Measuring Democracy: A Bridge Between Scholarship and Politics. Baltimore: Johns Hopkins University Press.

Pemstein, Daniel, Stephen Meserve, and James Melton. 2010. Democratic Compromise: A Latent Variable Analysis of Ten Measures of Regime Type. Political Analysis 18 (4): 426-449.

Pemstein, Daniel, Stephen A. Meserve, and James Melton. 2013. "Replication data for: Democratic Compromise: A Latent Variable Analysis of Ten Measures of Regime Type." In: Harvard Dataverse. http://hdl.handle.net/1902.1/PMM

Przeworski, Adam et al. 2013. Political Institutions and Political Events (PIPE) Data Set. Department of Politics, New York University. https://sites.google.com/a/nyu.edu/adam-przeworski/home/data

Reich, G. 2002. Categorizing Political Regimes: New Data for Old Problems. Democratization 9 (4): 1-24. http://www.tandfonline.com/doi/pdf/10.1080/714000289.

Skaaning, Svend-Erik, John Gerring, and Henrikas Bartusevicius. 2015. A Lexical Index of Electoral Democracy. Comparative Political Studies 48 (12): 1491-1525. Original data available from http://thedata.harvard.edu/dvn/dv/skaaning.

Svolik, Milan. 2012. The Politics of Authoritarian Rule. Cambridge and New York: Cambridge University Press. Original data available from http://campuspress.yale.edu/svolik/the-politics-of-authoritarian-rule/.

Taylor, Sean J. and Ulfelder, Jay, A Measurement Error Model of Dichotomous Democracy Status (May 20, 2015). Available at SSRN: http://ssrn.com/abstract=2726962 or http://dx.doi.org/10.2139/ssrn.2726962

The Economist Intelligence Unit. 2018. Democracy Index 2017: Free Speech under Attack.

Ulfelder, Jay. 2012. "Democracy/Autocracy Data Set." In: Harvard Dataverse. http://hdl.handle.net/1902.1/18836.

Vanhanen, Tatu. 2012. "FSD1289 Measures of Democracy 1810-2012." Original data available from http://www.fsd.uta.fi/english/data/catalogue/FSD1289/meF1289e.html

Wahman, Michael, Jan Teorell, and Axel Hadenius. 2013. Authoritarian regime types revisited: updated data in comparative perspective. Contemporary Politics 19 (1): 19-34.

Other democracy: extended_uds, uds_2014