Data

Overview

TBC

Downloads

TBC

Coding definitions and rules

WPD codes parties and candidates (henceforth “parties/candidates”) that win at least 5% of the vote in national legislative and executive elections in 50 countries (henceforth “cases”). By “candidate”, we only mean candidates in executive elections, not all candidates from parties in legislative elections.

WPD only includes parties/candidates that meet a minimum definition of populism defined below. Included entries are then coded on variables that differentiate qualities and attributes of their populism. Parties/candidates are listed for each election type, including multiple rounds of presidential elections, and multiple systems of legislative elections (e.g. simultaneous proportionate and constituency systems), with results from each.

Populism (minimum definition)

Parties/candidates are only included in the dataset if their discourse during the election campaign under analysis meets the our minimum definition of populism as a discourse asserting that political power has been monopolised by an elite estranged from ordinary people, and that salient grievances of ordinary people can be attributed to the self-serving decisions of that elite.

This definition has two components. First, candidates/parties must claim or imply that ordinary people are systematically excluded from political power, and that a small group of people - commonly the “elite” or similar terms - has monopolised this power.

“Ordinary people” in this definition can take many different forms, including ethnic, racial, native, regional, class, generational, or other characteristics. It can also be an undefined and residual category, in the sense of being all people other than the elite. Candidates/parties do not need to use terms like “the people” in order to infer the existence of this people/elite dichotomy. Nor do they need to express any affinity with the cultural traits or values of ordinary people.

The “elite” can also take many different forms, including incumbent politicians and parties, racial or ethnic groups, businesses, oligarchs, media figures or others. Unlike the “people”, the “elite” cannot be an undefined or residual category to meet our definition: it must be explicitly referred to by candidates/parties. This group may be collectivised by terms like “the elite”, “the establishment” (or bespoke alternatives), or evoked by referencing sets of actors or groups who have monopolised political power. As implied by the definite article in these terms, “the” elite must be considered to be an essentially singular actor, if not a totally homogenous one.

Second, populism manifests in an election campaign by the accusation that the salient grievances faced by ordinary people are a direct consequence of the actions of the elite, which benefits from those actions on a zero-sum basis with the people’s detriment. Ordinary people’s suffering must therefore be attributed to some form of corruption by the power-weiding elite which has deliberately constructed the status quo for its own benefit.

Attributing grievances in this way is different to attributing them to factors like flawed ideas/ideologies, ignorance, or incompetence on the part of the elite (although populists may make these claims as well). This is because each of these grievance attributions could plausibly exist in an elite which, while powerful, means well for the people, and is not essentially corrupt. On the contrary, populist grievance attribution makes a moral claim about the actions of the elite, asserting that they deliberately serve themselves at the expense of ordinary people.

This second component is essential for differentiating between populism and non-populist forms of anti-elitism. It is common for non-populist candidates/parties to campaign with some form of anti-elitism that fits the first component of the minimum definition but not the second. Such is the case when politicians accuse opponents of being estranged from ordinary people’s conditions, hostile to the cultures and behaviours of ordinary people, or partial to the company of powerful figures rather than ordinary people. While these elements of anti-elitism are common in populist discourses too, their presence in campaign discourse is insufficient to fulfil our minimum definition of populism unless the elite is also accused of deliberately causing the problems affecting ordinary people for their own benefit.

This second component, while often mentioned in the populism literature, has not yet been included in the minimum definitions of other populism datasets.

Locating populism

As our minimum definition states, WPD considers populism to be a discourse. Due to this, we have not considered candidate/parties’ programmatic policy in either the application of our minimum definition of populism, not any of our variables with the exception of LRPOSITION (see below). If, for example, a candidate wins office in an election in which they employed populist discourse, but then does not actually do anything to curb the power of their supposed elite adversaries when they are in office, this has no bearing on whether that candidate is in fact a populist in prior or subsequent elections in WPD.

This also means that we are not interested in the true underlying beliefs of populist candidates/parties that may contradict their discourse. If, for example, a presidential candidate campaigns against “the elite” in line with our minimum definition, but there is evidence to suggest that they actually have deep respect and reverence for political elites (from sources like interviews or autobiographies), then we code according to the former data, not the latter.

Similarly, we make no moral or judgements about the potential merit of any instance of populist discourse in any coding decision. That is, whether we happen to believe that the populist interpretation (as defined above) is an accurate and therefore “justified” representation of the polity under observation is irrelevant to our decision to include or exclude it from the dataset. “Populism” is therefore a morally neutral term in the dataset.

Generally, we have assumed that populist characteristics of candidates/parties’ discourse are carried over from one election period to another, such that we do not require fresh evidence for each election period. In cases where we do find evidence that candidate/parties have changed over time we factor this into our coding, and as a result there are many candidates/parties that enter in and out of the dataset over time.

Difficult cases

There are many types of discourse or political scenario that are questionable inclusions within this minimum definition. We have developed the following rules to clarify our application of it.

Non-populist “outsiders”. Many candidates/parties campaign as “outsiders” to existing systems, often as “third parties” outside of two-party systems, and in doing so take on a kind of holistic anti-status quo discourse. Ross Perot in the United States and Emmanuel Macron in France are good examples of this. While these are occasionally classified as populist because of their catch-all criticisms of the political system, we do not code them as populists unless their opposition to the system meets the second element of the minimum definition (that the salient grievances of ordinary people can be attributed to the self-serving decisions of that elite) - which neither Perot nor Macron do.

Incumbents. When populist candidates/parties win elections and form government, they can still engage in populist discourse to discuss “elites” as a powerbloc even though they are in power. Often this refers to powerbrokers in the state bureaucracy, judicial system, media or business class rather than incumbent politicians. It can also evoke a kind of phantom establishment or elite that is present in the polity yet temporarily out of office. Nicolás Maduro in Venezuela is a good example of this. Because of this, we don’t exclude incumbent candidates/parties due to the fact that they are in power.

Post-authoritarian parties/candidates. Many cases coded in WPD have emerged from periods of authoritarian rule. When parties/candidates in elections immediately following these transitions campaign with discourse that meets the above definition, but the targets of their anti-elite rhetoric are the former regime rather than their contemporary opponents, then we have generally not considered them as populist. Lech Walesa in Poland or Corazon Aquino in the Philippines are good examples of this.

Satirical candidates. In a small number of cases, candidates/parties campaign as “satirical” candidates who deliberately mock and ridicule the elite whilst not really claiming to be genuine alternatives. As these are often considered “protest” parties, we have coded them as populist when their act has an underlying message of corruption similar to our minimum definition. Estonia’s Independent Royalists are a good example of this.

Communist parties. As communist parties always critique the differential between powerful and powerless classes in political systems, there is an argument that this meets our minimum definition of populism. However, we find that communist (more specifically Marxist) discourse considers the role of elites to be more epiphenomenal to the overall system (capitalism) rather than as true agents in control of it, which is different to the first element of our minimum definition. For this reason we have generally not coded communist parties as populist, other than when their leaders have explicitly engaged in discourse that meets our minimum definition.

Anti-corruption politicians. While the second component of our minimum definition requires candidates/parties to allege some form of corruption among their opponents, this does not mean that any anti-corruption discourse qualifies as populist. Anti-corruption discourse is only populist by our definition when the elite writ large are accused of being essentially and systematically corrupt - not just that there are significant cases of corruption in the status quo. In such cases, parties/candidates would be meeting the second element of the minimum definition (corrupt motives as a cause of ordinary people’s problems) but not the first (power monopolised by a unitary elite).

Variables

When parties fit the above minimum definition, WPD codes them on a number of variables relating to core features of their populism and other attributes.

“Others” and enemies of the ordinary people

The first set of variables relate to who the main “others” or “elites” are relative to the party/candidate’s discourse. Some form of this rhetoric is present in all forms of populism (as our minimal definition requires), but it can take varying and occasionally overlapping forms. Parties/candidates can also change their “others”/“elites” over time, representing subtly different forms of populism in different elections.

For each variable we use a 3-level ordinal categorisation where 1 = no, 2 = present, and 3 = yes.

OTH_POLCLASS

Does the party/candidate claim that the political class - incumbent politicians, parties, bureaucrats, judges, and other holders of state office - are the “elite” or “others” to ordinary people and their interests?

1. (No) Party/candidate does not utilise anti-political class discourse.

2. (Partial) Party/candidate is critical of political class in a way that lacks the consistency, clarity, or ferocity of 3 (strong).

3. (Yes) Party/candidate considers the political class to be the or a core element of the “elite/other”, or direct representatives of that “elite” or “other”, and therefore responsible for the grievances of ordinary people

OTH_POLCLASS gauges the candidate/party’s characterisation of established political figures as an essentially unified group of people hostile to the interests of ordinary people – which we refer to as anti-political class discourse. This is perhaps the most commonly cited attribute of populist discourse, and unsurprisingly the great majority of entries in the dataset are coded 3 or 3. It is commonly evoked by the references populists often make to “the establishment” of political decision makers whom they accuse of siding with the “elite”, or constituting the essence of “the elite” themselves.

Incumbent politicians are the most common figures of the “political class” but we also include other figures more tangentially connected to the political system and/or the state, like former politicians, party officials, bureaucrats and civil servants, and judicial figures.

As the word “class” suggests, OTH_POLCLASS rhetoric must collectivise these actors in a way that presents them as a somewhat unified grouping, often emphasising that differences between them (like parties and official ideologies) are superficial and meaningless. Consider for example the assertion by the Palikot Movement of Poland that it was “not the next configuration of the same old elite...AWS, PiS, PO, ZChN, PC, KPN, ROP, and now PJN – different names, but the same people.”

OTH_POLCLASS discourse does not imply that parties/candidates must criticise all actors in political class equally, but just that there must be some implication that the bulk of “the” political class acts as a somewhat homogenous entity. Populists frequently create terms like “the network”, “the caste”, “the political mafia” to collectivise the political class in their rhetoric.

Parties/candidates are coded 3 on this scale when these references are core and consistent elements of their characterisation of elites. This includes cases where parties/candidates claim that other groups are also essential components of the elite, but that the political class acts as a corrupt accomplice to their actions. For example, Jean-Luc Melenchon (France) and Ollanta Humala (Peru) both employed OTH_POLCLASS discourse at a level befitting 3 on this scale, although both claimed that other elements (economic elites and foreign powers in both cases) were equally core components of the “elite” opposed to the interests of ordinary citizens. Conversely, some parties - such as Italy’s Five Star Movement in its first election in 2013 - considered the political class as the singular essense of the elite.

There are very few cases of parties/candidates coded as 1 on this variable. In these cases populist discourse is always employed against other “elites”. Examples of this are the Estonian National Independence Party, and the Taiwan Solidarity Movement, both of which focused mainly on foreign powers rather than the domestic political class.

Parties/candidates coded as 2 employ some degree of this discourse, but without the clarity, consistency, or degree to reach the level of 3 as defined above. Bulgaria’s National Movement Simeon II and New Zealand’s Social Credit Party are good examples of this relatively moderate OTH_POLCLASS discourse.

OTH_POLCLASS discourse can be used by candidates against their own parties. The presidential campaigns of Donald Trump (United States) and Jacob Zuma (South Africa), for example, employed this rhetoric against established political figures within the Republican Party and the African National Congress, respectively.

OTH_POLCLASS discourse can also be employed by candidates/parties who are incumbent in office. Turkey’s Recep Tayyip Erdogan, for example, only reached a 3 on our OTH_POLCLASS classification during the 2014 election, at which point he and his AK Party had been in power for over a decade.

OTH_ECONOMIC

Does party/candidate claim that economic elites - big business, the wealthy, financiers, oligarchs or other economic powerbrokers - are the “elite” or “others” to ordinary people and their interests?

1. (No) Party/candidate does not define the “elite/other” in economic terms.

2. (Partial) Party/candidate is critical of economic elites in a way that lacks the consistency, clarity, or ferocity of 3.

3. (Yes) Party/candidate claims economic elites are core elements of the “elite” or “other” responsible for the grievances of ordinary people.

OTH_ECONOMIC discourse is present when candidates allege that economic elites – which refers to big business, billionaires, oligarchs, financiers, wealthier social classes, etc. – are an element of the elite. This can take the form of ideologically rich discourses about inequality or capitalism, such as Die Linke in Germany, Syriza in Greece, and Hugo Chavez’s Fifth Republic Movement in Venezuela – each of which are socialist parties that consider the “people” and “elite” to be principally class divisions in the material sense. It can also exist in less ideologically defined opposition to the power of businesses or wealthy individuals, such as with the ideologically ambiguous New Zealand First party, or the centrist anti-oligarch Zatlers Reform Party in Latvia.

It is also often the case that candidates/parties usually characterised as right wing in the sense of ethno-nationalism, immigration, or social conservatism evoke their supporters’ hostility to economic elites in a similar way to leftists. The more recent campaigns of the far right Front National in France are a good example of this. In other cases, right wing populists like Iceland’s Progressive Party or Greece’s Independent Greeks have turned to OTH_ECONOMIC discourse in the wake of financial crises. The economic ideology of candidates/parties and its perceived effect on economic powerbrokers is not relevant to our OTH_ECONOMIC coding decisions. For example, a party/candidate may endorse a neoliberal economic ideology that would serve the interests of economic powerbrokers while still engaging in OTH_ECONOMIC discourse.

It is often difficult to disentangle OTH_ECONOMIC discourse from other OTH_ variables. It is frequently the case that opposition to economic powerbrokers is constructed in nationalist terms, with populists specifically opposing “foreign” businesses or billionaires. Likewise, opposition to domestic economic elites can overlap with ethnic or racial rhetoric, as in the case of South Africa’s Economic Freedom Fighters (EFF). In such cases we have still considered such rhetoric OTH_ECONOMIC as long as discourse is still constructed in economic terms (which it does in the case of EFF’s).

However, when discourse against one or few economic elites appears to have more to do with an incidental personal feud than an opposition to the wealth of those figures per se we have erred on the side of not considering this OTH_ECONOMIC discourse. Viktor Orban’s hostility to the influence of expatriate billionaire George Soros in Hungary (which also takes on anti-semstic overtones) is a good example of populist opposition to wealthy individuals that is not truly OTH_ECONOMIC discourse.

Consistent with our approach of coding discourse rather than policy, we ignore policy proposals that conflict with this discourse. That is, when a candidate/party claims that billionaires are an enemy of the people, but advocates cutting taxes on the wealthy or liberalising financial regulations, we code according to the former rather than the latter.

OTH_MILITARY

Does party/candidate claim that the national military are the “elite” or “others” to ordinary people and their interests?

1. (No) The military does not feature in the party/candidate’s construction of the “elite/other”.

2. (Partial) The party/candidate is critical of the military’s involvement in politics in a way that lacks the consistency, clarity, or ferocity of 3.

3. (Yes) The party/candidate considers the military to be core elements of the “elite” or “other” responsible for the grievances of ordinary people.

OTH_MILITARY discourse is present when populist candidates/parties allege that military establishments are elements of the elite and hostile to the people’s interest. This is particularly prevalent in cases where prior military regimes are seen to remain as political foils after transitions (such as Argentina or the Pakistan), or where the military habitually intervenes in the political process (such as Turkey or Thailand). However this needn’t be evident for OTH_MILITARY discourse to be present.

Codings of 3 are used when OTH_MILITARY discourse is a clear and prominent element of populist rhetoric, for example Thailand’s Pheu Thai party, or Argentina’s Front for Victory, each of which characterise the military establishment as core opponents of the people in the popuilst terms of our minimum definition.

Codings of 2 are reserved for relatively milder forms which are neither absent nor strong, such as Brazil’s Workers Party (in the 1980s and 1990s).

OTH_ETHNIC

Does party/candidate claim that specific ethnic, racial or religious groups are the “elite” or “others” to ordinary people and their interests?

1. (No) Party/candidate does not characterise the “elite” or "others" in ethnic/racial/religious terms

2. (Partial) Party/candidate is critical of ethnic/racial/religious groups in a way that lacks the consistency, clarity, or ferocity of 3.

3. (Yes) The party/candidate considers ethnic/racial/religious groups to be core elements of the “elite” or “other” responsible for the grievances of ordinary people.

OTH_ETHNIC discourse occurs when parties/candidates characterise the categories of “elite”, or “other” in ethnic, racial, or religious terms. Parties/candidates are coded 3 when this is a clear and prominent element of populist discourse. Relevant out-groups can be either domestic minorities/majorities, or local representatives of transnational groups such as religions, races, or Roma.

There is a significant range within the level 3 category, as we consider extreme cases such as Golden Dawn in Greece or Vladimir Zhirinovski’s Liberal Democratic Party in Russia and the somewhat milder cases like France’s Front National or Australia’s One Nation party equally fitting of the language of level 3 here - even though they differ on the degree of their and content OTH_ETHNIC rhetoric.

The coding of 2 is used for relatively milder form of OTH_ETHNIC discourse among populists. This includes candidates/parties who make occasional references to ethnic/racial/religious groups in ways that are discursively salient enough to surpass a 1 (null) coding, but not enough to reach 3. Iceland’s Progressive Party, and the New Zealand First party are examples of this.

Category 2 is also used for candidates/parties who rely on deeper legacies of ethnic/racial/religious discourse and action rather than contemporary rhetoric, such that neither category 1 nor 3 is appropriate. An examples of this Narendra Modi in India (who’s controversial handling of inter-communal riots in Gujarat in 2002 made him an icon of broader Hindu majoritarian politics).

We consider regional (including separatist) parties to engage in OTH_ETHNIC discourse when differentials between regions are defined in ethnic, racial or religious terms. Canada’s Bloc Quebecois and Belgium’s Flemish Bloc are examples of this.

While many parties/candidates define the “people” in ethnic, racial, or religious terms, we do not require this to code 2 or 3 on the OTH_ETHNIC variable. This is because there are many examples of parties/candidates who present somewhat liberal or multicultural representations of people, but still focus specifically on an opposition to one or more ethnic, racial or religious groups in a way befitting level 3 on this scale. Geert Wilders’ Party for Freedom is a good example.

OTH_IMMIGRANTS

Does the party/candidate claim that immigrants are the “elite” or “others” to ordinary people and their interests?

1. (No) Immigrants are irrelevant to party/candidates characterisations of “elites” or “others".

2. (Partial) Party/candidate is critical of immigrants in a way that lacks the consistency, clarity, or ferocity of 3 (strong).

3. (Yes) The party/candidate considers immigrants to be core elements of the “elite” or “other” responsible for the grievances of ordinary people.

OTH_IMMIGRANTS discourse occurs when parties/candidates characterise immigrants as a threat to the interests of the ordinary people. Immigrants are rarely considered to be a component of the “elite”, but often constitute an “other” that is favoured by the elite over ordinary people. Parties/candidates are coded 3 when this is a clear and prominent element of populist discourse. The UK Independence Party and Brazil’s Jair Bolsonaro are examples of this.

Level 2 is used on this variable when parties/candidates utilise some degree of anti-immigrant discourse that lacks the consistency, clarity, or ferocity of true anti-immigrant discourse. An example here is the ideologically ambiguous Five Star Movement in Italy, whose populism was (initially) centred on its ridicule of the political class, but in later elections included some anti anti-immigrant narratives.

There are many cases where OTH_IMMIGRANTS discourse overlaps with OTH_ETHNIC discourse. The Slovenian Democratic Party of Janez Janša for example has long been opposed to immigrants from other former Yugoslavian states in ways that correlated with ethnocentric narratives of Slovenian nationalism. This is not always the case however, such as with the example of Carl Hagen’s Progress Party in Norway, which became strongly anti-immigrant in the mid-1990s but in a way that related more to welfare chauvinism than explicit ethno-nationalism.

In cases where parties are opposed to immigrants from certain countries or ethnic backgrounds but supportive of immigrants from others, we still consider this OTH_IMMIGRANT discourse. Pakistan’s Muhajir Qaumi Movement is a good example of this. As with other variables, we ignore actual policy proposals relating to immigration and focus on campaign discourse against immigrants (although these are naturally correlated).

OTH_FOREIGN

Does party/candidate claim foreign entities - states, institutions or other entities - are the “elite” or “others” to ordinary people and their interests?

1. (No) Foreign entities do not feature in party/candidate’s characterisations of “elites” or “others”.

2. (Partial) Party/candidate is critical of foreign entities but does not consider their domestic opponents to be totally subservient to them

3. (Yes) Party/candidate claims that their domestic opponents are totally controlled by and loyal to foreign interests hostile to ordinary people.

OTH_FOREIGN discourse occurs when parties/candidates claim that the elite responsible for the grievances of ordinary people is a foreign actor (or actors) of some kind. These foreign actors are usually foreign governments, but can also be supernational/regional entities like the European Union, or global institutions like the IMF or World bank. It can also include less well-defined references to “globalist” actors or agendas. We do not consider immigrants from other countries to be “foreign entities”.

In order to fit our minimum definition of populism, OTH_FOREIGN populism must assert that foreign power is in some ways exerted through domestic figures, be they in the political class, business, domestic ethnic groups or other.

OTH_FOREIGN discourse does not require parties/candidates to be consistently against all foreign entities in the sense of ideological isolationism and/or nationalism, only that foreign entities be among the elites and others responsible for the grievances experienced by ordinary people. Parties/candidates may therefore practice OTH_FOREIGN discourse when they are in favour of some foreign entities. As with other variables, this can be the only OTH_ variable, or can exist alongside others.

Level 3 is reserved for cases where parties/candidates accuse their domestic opponents of being wholly controlled by or representative of these foreign interests. Examples of this are Bolivia’s Movement for Socialism and Greece’s PASOK (in the 1980s).

Level 2 is used when parties/candidates consistently refer to the influence of foreign entities without reaching this threshold. Examples of this are the Japanese Restoration Party or Mexico’s Party of the Revolution.

Attributes of populist parties/candidates

Next, we use six other variables relating to other aspects of candidates’ actions and backgrounds:

LIBDEMNORMS

Does the party/candidate’s behaviour threaten established liberal democratic norms - engaging in political violence, resisting constitutional rules and institutional processes, showing hostility to judicial oversight, or attacking on media scrutiny - compared with predecessors and contemporaries?

1. (No) Party/candidate does not threaten established liberal democratic norms.

2. (Partial) Party/candidate represents a moderate threat to established liberal democratic norms.

3. (Yes) Party/candidate represents a stark break from established liberal democratic norms.

LIBDEMNORMS measures whether or not the behaviour and/or speech of parties/candidates threatens established liberal democratic norms. We consider these threats to include: a) inciting, tolerating or organising violence against opponents, b) attacking the freedom of the press, c) threatening democratic institutions (for example undermining the credibility of electoral institutions or demanding that political opponents be prosecuted), d) expressing a dislike of democratic institutions (including offering praise for domestic or foreign dictatorial regimes) and e) (for incumbents) “state capture,” in the form of personalising the resources and violence of the state. This is similar to “authoritarian” behaviour in other datasets.

While ethnonationalism or exclusion of migrants are arguably threats to liberal democratic norms, we have not included them in this variable given that they are captured by the OTH_ETHNIC and OTH_IMMIGRANTS variables. “Liberal” democratic norms here do not relate to liberal economic policies or attitudes, meaning that a party/candidate could be hostile to liberal economic norms without being opposed to liberal democratic ones in our definition of LIBDEMNORMS.

Given that liberal democratic norms vary greatly prior to the entry of any given populist parties/candidates, WPD codes relative to the “established” standards of the polity. That is, parties/candidate’s behaviour only registers on this scale when it is seen to shift norms away from liberal democracy, not merely when it is in line with the non-liberal democratic norms practiced by predecessors and contemporary rivals. The three levels of LIBDEMNORMS do not therefore relate to absolute standards of behaviour across all cases, but the degree to which populist parties/candidates shift or threaten their countries’ established democratic norms. When this is ambiguous, we’ve erred on the side of coding higher rather than lower.

Parties/candidates who are incumbent in government have a different capacity to threaten liberal democratic norms than those who are non-incumbent. For example, it is possible for parties who showed no signs of threatening liberal-democratic norms from restricting press freedom or judicial oversight.

CHARISMA

Is the personal charisma of the candidate or party leader essential to their campaign’s political appeal?

1. (No) The leader is not essential to the political appeal of the campaign, and the party or movement would likely exist without this leader.

2. (Partial) The leader is an important element of the campaign’s political appeal, but the party or movement would likely exist without this leader.

3. (Yes) The leader is essential to the campaign’s political appeal, and the party or movement would not likely exist without this leader.

The CHARISMA variable captures populist parties’ or movements’ potential dependence on their leaders - something routinely cited in the theoretical literature on populism. We have identified CHARISMA in candidates/parties when sources suggest that the appeal of a particular movement is tied up in the personality of its leader in the sense of being “leader-centric” or a “one-man/women” show. This is the case for many populist movements formed around a single individual, such as Clive Palmer’s United Australia Party, Joseph Estrada’s Forces of the Filipino Masses in the Philippines - each of which would stand no chance of existing without these leaders.

The term “movement” here relates to the fact that independent presidential candidacies could still plausibly have non-partisan foundations, such as support from coalitions of political parties (though this is rare).

As it is difficult to scale the significance of leadership, we differentiate between 3 and 2 not based on difference in charismatic appeal, but on whether evidence suggests that parties have an institutional base beyond its current leader – particularly as evidenced by a history that predates that leader – such that it probably exists “beyond” them in some way. Therefore, there is not always a difference in the presence of charismatic leadership between cases coded 2 or 3.

In some cases CHARISMA can be based on politicians who are not actually leaders of their party, but none-the-less essential to their appeal. For example, Thailand’s Pheu Thai (PT) is the successor party to Thaksin Shinawatra’s Thai Rak Tai (TRT) party, formed after TRT was overthrown in a 2006 military coup in Thailand. Even in exile however, evidence suggests that Thaksin’s image was still essential to the party’s image and appeal. In such cases we have still coded 3 if that leader remains an essential component of the party’s identity at the time of the election under analysis.

INC_PARL

Is the legislative party incumbent in government at the time of the election?

1. Not in government

2. In coalition government but not head of government

3. In coalition government with head of government

4. In government outright

INC_PARL captures whether or not legislative parties are in government at the time of the election campaign under analysis. Presidential election entries are coded “N/A” on this variable.

Each coding decision relates to the time immediately before the election in question. When a party has been in government briefly in the previous term but has left before the election, or when a party has joined a governing coalition soon before the election, we still code whatever is the case at the time of the election.

Parliamentary coalitions make incumbency difficult ot assess as there are many cases where incumbency is not clear cut in coalitions with junior/minor parties. We use the position of head of government as a proxy for this, hence the difference between 2 and 3 above. For systems where the President is the head of government, (e.g. France, Argentina) that still counts for parliamentary parties. Likewise, when a party has a parliamentary majority, but the presidential head of government is from another party, they are coded 1.

In cases where parties appear to lend support on a more informal basis to incumbent presidents from different parties or who are independents, we do not include this as in government or coalition. This is common in Latin America.

INC_PRES

Is the presidential candidate incumbent in government at the time of the election?

1. Not incumbent

2. Not personally incumbent, but same party as incumbent

3. Incumbent

As with INC_PARL, we code for whatever is the case immediately prior to the election. Legislative election entries are coded “N/A” on this variable. “In government” refers to the office that is being sought in that election, not any government office.

INSIDER

Has party leader or candidate been in government in the past?

1. (No) Leader has never featured in government before this election.

2. (Partial) Leader has featured in previous governments, but did not lead them.

3. (Yes) leader has been a head of government before this election.

INSIDER captures whether or not party leaders or candidates have been in positions of government before the election under analysis. In both legislative and executive entries, we are interested in the history of the current leader rather than the party historically.

Level 3 is reserved for leaders who are previous heads of government. This could be incumbent heads of government, or non-incumbent former heads of government. Leaders are coded 2 when they featured prominently in previous governments without serving as head of government. For example, Corneliu Vadim Tudor was leader of the Greater Romanian Party (PRM) when it was briefly in coalition government in 1992, meaning that PRM is coded 2 on INSIDER for the duration of the time that he was its leader. Similarly Cuauhtemoc Cardenas was a key figure in several PRI governments in Mexico before he contested from 1988 - he and the parties he led are coded 2 for this reason.

Only prior offices held at the national level are considered. Parties whose leaders held office at the subnational or local level are still coded 1 if they have never held national office.

All incumbent leaders are coded 3 on this scale, even if they were coded 1 or 2 during their initial election.

LRPOSITION

Where does the candidate/party stand on the left-right ideological spectrum?

FR Far right

R Right

CR Centre right

C Centrist

CL Centre left

L Left

FL Far left

N/A Not applicable (i.e., satirical candidate)

LRPOSITION codes parties/candidates according to where they fall on the left-right ideological spectrum. This is based on the general assessments in the literature as well as our own understanding of left-right ideology. When parties/candidates have differed in their conventional left-right position between, for example social/cultural politics (immigration, ethnocentrism, etc.) and economic policies (taxes, welfare, etc.) we have coded based on the relative salience of each category. For examples, we have coded parties like Ataka in Bulgaria and the True Finns party in Finland as right wing due to the relative salience of their ethnonatioalism, even though they also favour economic policies wh

In cases where left-right ideological profiles have been difficult to determine due to either the balance of clearly left- and right- leaning positions, we have erred on the side of classifying parties as C. Examples of this are Beppe Grillo’s Five Star Movement in Italy, or Andrzej Lepper’s Self-Defence in Poland.

In cases where populist protest parties make no claim to any policy position other than ridicule of the political establishment, such as Estonia’s Independent Royalists, we have coded them as N/A (Not Applicable).

Elections and electoral systems

Elections are recorded by YEAR [yyyy] and MONTH [m], and also by DATE [dd/mm/yyyy]. The dataset included candidates/parties in both presidential and (various kinds of) legislative elections. These are distinguished by the TYPE variable:

PR	Presidential
LH	Lower House (including when upper house is not directly elected)
LH+UH	Both upper and lower house elected in single ballot system
UC	Unicameral

A second column, ROUND is used to denote whether election result is in the first or second round of a two-round system, in either presidential or legislative elections:

1	First Round
2	Second round
N/A	Not applicable

A second column, SYSTEM is used to denote the different representational systems that the result in question refers to:

CON	Constituency seats (as opposed to proportional representation)
PRO	Proportional representation seats (as opposed to constituencies)
DIA	Diaspora representatives

Note: the SYSTEM variable is only used when election to one legislative chamber (LH/UC) involves two balloting systems. It does not denote different methods of legislative election via a single ballot, such as mixed constituency and proportional representation systems, etc. If there was only one ballot, and therefore only one total percentage, then the system column is listed with “N/A”.

Data Sources

Coding data

Data for these evaluations was sourced primarily from secondary academic sources, but also from reliable online analysis (reputable blogs, analysis sites, etc.), news sources, and occasionally primary sources from candidates/parties such as speeches or manifestos. By far the most common source is secondary academic literature, as this material tends to include the richest description of populist rhetoric and appeal. Data sources used for coding have been laid out in a separate Case Notes document.

Election data

Election results data was sourced from a number of datasets and outlets, each of which is listed in the SOURCE column of the spreadsheet, with a link copied in the SOURCELINK column. In order of preference, these are:

Source	Reference in dataset
Parliaments and Governments Database	ParlGov
Database of Political Institutions 2017	DPI
European Electoral Database	EED-NSD
Inter-Parliamentary Union Parline	IPU Parline
Nohlen et al source books	Nohlen
Chapel Hill Expert Survey	CHES
Adam Carr’s Election Archive	Carr
Nordsieck, Wolfram. Parties and Elections in Europe, (2017).	Nordsieck
International Foundation for Electoral Systems	IFES

Wherever possible, as few datasets as possible were used on individual countries. When these databases did not provide relevant data, we have used (in order):

Source	Reference in dataset
Official state source (eg. national electoral commission website)	Official
Statistics mentioned in secondary scholarly sources	Secondary

Cases (countries)

The beta dataset includes 50 democracies, all selected based on the general reliability of official election results, and the availability of sufficient secondary sources for coding. This meant that several countries studied by coders were ruled ill-suited to the dataset despite the clear presence of populism (Belarus, Azerbaijan, Armenia, Iran for example).

The countries included are:

Europe (27)	Albania, Austria, Belgium, Bulgaria, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Netherlands, Norway, Poland, Portugal, Romania, Russia, Slovakia, Spain, Sweden, Switzerland, United Kingdom
South America (8)	Argentina, Brazil, Bolivia, Chile, Colombia, Ecuador, Peru, Venezuela
North America (4)	Canada, El Salvador, Mexico, United States
Asia (7)	India, Indonesia, Japan, Pakistan, Philippines, Thailand, Türkiye
Africa (2)	South Africa, Zambia
Oceana (2)	Australia, New Zealand,

Numerous countries are not included for the breadth of the 1980-2020 period as they did not qualify as sufficiently democratic at every point. For example, Eastern European post-communist countries are only included after ~1990, South Africa is only included after 1994, the Philippines after 1986.

The cases of Czechia and Slovakia are included as separate entities even before their formal split in 1993 due to the fact that their electoral processes were autonomous in these years, and hence there is historical continuity with later elections.

Elections coded

The dataset has coded a total of 809 elections (543 legislative and 266 presidential).

Parties/candidates

Parties (including the independent status of some presidential candidates) are listed in the PARTY variable, and candidates (in presidential elections) are listed in the CANDIDATE variable.

The dataset includes parties/candidates which received over 5% at the election being recoded.

Rules for coding parties, and alliances

In parliamentary systems we assume that the actions and discourse of party leaders constitute the essential meaning of the party as it is understood by voters. In presidential systems this is more difficult, as it is occasionally the case that presidential candidates have estranged relationships with party leaders, and that candidates’ discourse doesn’t necessarily imply that every voter at a legislative election is voting for a populist party. For example, given the abnormally institutionalised nature of the US two-party system, we don’t find it accurate to say that every vote for a Republican congressional candidate in the 2016 US elections had the same populist meaning as the vote for the party’s presidential candidate Donald Trump. We have made this judgment on a case by case basis and explained this choice in the case notes.

When parties form alliances with each other, we have tried to treat them as individual parties whenever electoral data allows us to do so (for example, the Centre Right coalition in Italy’s 2018 election). When this is not possible due to the fact that alliances are single voting lists or electoral entities, we have only considered these alliances populist when populist parties appear to account for the majority of their candidates and or leaders (that is, when populist parties are the dominant partner). This was the case for the Japanese Restoration Party of Toru Hashimoto, a regional party which merged with another party (The Japan Sunrise Party) before the 2012 election.

When parties have merged with other parties, we have usually considered them as continuations of predecessor parties in cases where those parties were significantly bigger than others. For example, Sylvio Berlusconi’s Forza Italia (FI) merged with the National Alliance (NA) party in 2009 to create the People of Freedom (PDL). However, as FI was significantly bigger than the NA, and Berlusconi remained it’s leader after the merge, we have considered PDL as a continuation of FI, and carried coding decisions of the latter into the former after 2009.

In presidential systems it is common for parties who run separately in parliamentary elections form electoral coalitions for the presidential elections. This is usually the case in many Latin American countries. In these cases, we have coded the relevant electoral base of the populist leader. For example, for the Chilean party UDI under the leadership of populist leader Joaquín Lavin (1999-2005) we have coded “Independent Democratic Union” for the parliamentary elections and “Independent Democratic Union / Alliance for Chile” for the presidential elections.

One particularly difficult case is Argentina, which has undergone a shift in its party system. At the time of Carlos Menem and Nestor Kirchner’s leadership in the 1990s- early 2000s, we consider the Justicialist party to be representative of its leaders’ discourse, and therefore a populist entity linked to that leader. However, Cristina Fernández de Kirchner, also of the Justicialist party, formed the Front for Victory alliance with members of her party and others. Given that votes for this alliance can be tabulated separately, we treat this entity as a party rather than the Justicialists.