Populist attitudes and populist voting : the case of the 2016 legislative election in Taiwan

(1)

Populist Attitudes and Populist Voting:

The Case of the 2016 Legislative Election in Taiwan

Master Thesis Political Science

Ssu An Chen (11255218) June, 2016

First Reader: G. Schumacher, Ph.D. Second Reader: E. Harteveld, Ph.D.

(2)

1. Introduction

With the surge of populist forces across the world, in particular in Western Europe and Latin America, populism has become a term that receives great attention both within and outside of academia. Many studies have focused on the conceptualization of populism (e.g., Weyland 2001; Stanley 2008). The concept of populism is fuzzy, flexible and context-dependent (Arter 2010; Taggart 2000; Pauwels 2011). Furthermore, it has a nature of elusiveness and multidimensionality (Ionescu and Gellner 1969; Roberts 2007). Although a common definition has not yet come into being, consensus is taking shape in recent years, especially regarding some of the core features of populism (Van Kessel 2014; Marcos et al. 2016). Still, the slow progress of its conceptualization has led to a relative paucity of its empirical investigation (Stanley 2011; Rooduijn and Pauwels 2011; Hawkins 2009), posing difficulties to perform objective measurements of populism on both the supply and demand sides. Thus far, for the supply side of populism, a number of researches have assessed the populist discourses of populist parties or politicians. Comparisons of different approaches have been made (e.g., Rooduijn and Pauwels 2011; Poblete 2015). On the other hand, several studies have attempted to gauge the populist attitudes of individual citizens, that represent the demand side of populism.

With increasingly more studies focusing on both sides of populism, one question has been attracting attention: is there indeed a link between the populist tendencies on these two sides? In other words, can the populist attitudes among voters indeed explain their choice to vote on populist parties, and if so, to what extent? The electoral success of populist forces in various countries does not automatically confirm support for

(4)

populism per se among the public. The choice for a party or a candidate is usually influenced by several factors. Citizens may opt for populist parties in order to show their discontent with the established traditional parties, or simply because of the charisma of some populist leaders (De la Torre 2007). To date, only few studies have probed into this specific topic (e.g., Stanley 2011; Akkerman et al. 2014; Marcos et al. 2016; Hawkins et al. 2016), and case studies are limited to European or American countries.

Populism is being called a “global phenomenon” (Marcos et al. 2016: 2), but studies on populism have, however, predominantly focused on Europe and Latin America (for example, Pappas 2012; Mudde and Kaltwasser 2013). To be sure, sporadic studies have touched upon countries in other regions, such as Russia (Ionescu and Gellner 1969), India (Subramanian 2001), and the Philippines (Hawkins 2009), but the weight that has been put on the other regions altogether is still far from the weight on Europe and Latin America. Does this indicate little occurrence of populist forces in the other regions? And does this imply that populist attitudes are not prevalent in the countries outside of Europe and Latin America? In this thesis, I will demonstrate the role of populism in a country outside of the normal regional interest of study – Taiwan, the East Asian island-state.

This thesis will be built on previous work which examines the link between voters’ populist attitudes and their voting choices for populist parties, especially on the paper of Hawkins et al. (2016), in terms of the framework and the approach used. The purpose of this thesis is to strengthen the few previous results that prove the effects of populist attitudes on voting for the populists. The 2016 legislative election in Taiwan will be

(5)

utilized as the case of study. Since Taiwan, or Asia in general, is a whole new territory of academic research of populism, this thesis also adds to the existing literature by broadening the range of this study.

The thesis proceeds as follows: first, the definitions of populism and related theories will be presented. Second, the context of Taiwan and the 2016 legislative election will be elaborated. Third, the analytical framework of this thesis will be presented. Fourth, the populist party in Taiwan will be identified through content analysis of party discourse. Next, levels of populist attitudes among the Taiwanese will be demonstrated, and vote-choice analysis will be conducted in order to test if these attitudes do predict their decisions of voting for the populist party. Finally, conclusions regarding the role of populism in Taiwan will be drawn.

2. Concept and theory

The concept of populism understood by non-scholars is usually different from its academic definitions (Mudde 2004). Even within the academia, there is confusion and disagreement about the conceptualization of populism, resulting in various definitions of it (Gidron and Bonikowski 2013). Such variance in the understanding of populism might be due to its changing faces in history. Populist politics is by no means a recently emerged phenomenon. There have been at least three waves of populism in different periods of history distinguished by scholars.

The first wave started with the agrarian populism in Russia and the US in the late 19th century, in which people pleaded against capitalism and stood up for the peasants’ rights. Populism appeared again in Latin America in the mid-20th

(6)

nationalist and charismatic leaders claiming to be the embodiment of the ordinary people. Finally, there has been the revival of populism from the 1970s onwards. This wave of populism includes the form in Europe and the United States that is often linked with the radical right, and the left-wing populism in Latin America (Roberts 2007; Jagers and Walgrave 2007; Gidron and Bonikowski 2013). These variants of populism in different time frames and contexts have made it difficult to maintain a coherent conceptualization.

The contemporary conceptualizations of populism can be categorized into three main groups, based on its various dimensions. These groups define populism, respectively, as (1) a political strategy, (2) a discursive style, or (3) an ideology. Defining populism as a political strategy is mostly prevalent among studies on Latin America (Gidron and Bonikowski 2013). There are three sub-types under this category. The first sub-type focuses on the policy outputs by populist political actors. Based on empirical instances in Latin America, these are usually “short-sighted economic policies that appeal to the poor” (Hawkins 2009: 1042). For example, in Acemoglu et al. (2011) populism is defined as adopting policies that are supported by a mass number of the population, although they are in the end harmful to the overall economy of the country. Madrid (2008: 482) identifies populist economic policies as those aiming at income redistribution and the nationalization of natural resources. This definition nevertheless fails largely to account for ideological variation among populism in different societies, for instance between the left-wing Chavismo in Venezuela and the right-wing National Front in France (Bonikowski and Gidron 2016).

(7)

The second sub-type views populism as a form of political mobilization. According to this conceptualization, mass constituencies are politically mobilized by ‘personalistic leaders’ by using anti-establishment appeals and claiming to represent the ordinary people (Levitsky and Roberts 2011; Jansen 2011; Roberts 2007; Gidron and Bonikowski 2013). Some scholars specifically argue that these populist leaders tend to cast themselves as outsiders of the established political system (Barr 2009; Pappas 2012).

The third sub-type stresses on the organization of populist parties. It identifies some defining features, such as the presence of a strong and personalized leader, centralized and hierarchical party structure, dislike of institutional complexity, and an emphasis on their mass number of supporters (Taggart 2000; Weyland 2001; Hawkins 2009; Pauwels 2011; Gidron and Bonikowski 2013). However, while the leadership of some strong, charismatic individuals is usually cited as one characteristic of populism in the second and third sub-types (Taggart 2000; Weyland 2001: 14), many have critiqued against it. The critics doubt whether it qualifies as a sufficient or even necessary criterion of populism, since there have also been non-charismatic populist leaders, or even populist movements without firm leadership. This trait is also not limited to populist parties or movements (Barr 2009; Taggart 2000; Gidron and Bonikowski 2013; Elchardus and Spruyt 2016). The same problem goes with the trait of a centralized party structure (Taggart 2000).

The discursive approach of populism considers it as a ‘rhetoric’ that frames politics as the struggle between the moral people and the amoral oligarchy (De la Torre 2000: 4), a ‘political communication style’ that refers to the people and at the same time being

(8)

anti-establishment and exclusionary (Jagers and Walgrave 2007). It is a ‘language’ (Kazin 1998), an ‘appeal’ (Canovan 1999), a ‘form of claims-making’ (Bonikowski and Gidron 2016: 1597), a ‘characteristic of political talk’ (Deegan-Krause and Haughton 2009), which all claim to speak for ‘the people’, and pit them against their constructed enemies, ‘the elite’ (Gidron and Bonikowski 2013). Populism is assessed to be “a standard communication technique to reach out to the constituency” utilized by politicians (Jagers and Walgrave 2007: 323), which coheres with the ‘politicians’ populism’ (Canovan 1981). In this definition, populism is merely a device that is adopted verbally and literally. It does not necessary imply a set of beliefs in the minds of its adopters, and does not necessarily lead to any pattern of political outputs. If a normative dimension is added to the discursive definition, populism becomes more of an ideology than just a political style (Pauwels 2011). This ideational approach of conceptualizing populism has won ground among scholars in the past decade. One of the most influential definitions has been provided by Mudde (2004: 543), who describes populism as an “ideology that considers society to be ultimately separated into two homogeneous and antagonistic groups, ‘the pure people’ versus ‘the corrupt elite’, and which argues that politics should be an expression of the volonté générale (general will) of the people”. In this definition, populism is a set of ideas featuring the antagonism between two homogeneous groups, the people and the elite. Labels have been tagged to these two groups, with the former being virtuous and the latter unethical. Since the elite usually conspire against the people’s interest and monopolize the politics for their own gains, populism advocates for popular sovereignty and believes that government policies should be the direct reflection of the people’s will (Pauwels 2011).

(9)

Nonetheless, although populism entails its own distinct worldviews, these unique ideas can be found existing together with several different ideologies, as Roberts (2007: 4) contends, “By nature……populism tends to be ideologically eclectic and malleable.” Therefore, among those who adopt the ideational definition, most of them also stress that populism is a ‘thin ideology’ (e.g., Mudde 2004; Rooduijn and Pauwels 2011; Stanley 2011; Akkerman et al. 2013). Thin ideologies are those that only focus on a limited extent of concepts, or to be more precise, the ‘meta-political’ dimensions such as “broad ontological and normative assertions rather than the detail of policy” (Stanley 2011: 258). Consequently, thin ideologies are incapable of providing solutions to major socio-political problems, unlike the full political ideologies such as socialism or liberalism do. These ideologies are, therefore, usually attached to different ‘host ideologies’, thin or full (Mudde 2004; Rooduijn and Pauwels 2011; Akkerman et al. 2013; Gidron and Bonikowski 2013; Bakker et al. 2015). Populism as a thin ideology thus rarely exists on its own. Instead, “it can be found across ideological cleavages, fused with either left- or right-wing appeals” (Gidron and Bonikowski 2013: 6), depending on the socio-political context in which populist actors emerges (Mudde and Kaltwasser 2013).

Due to the manual classification of these different meanings of populism, part of the distinctions between them are artificial, instead of intrinsic (Pauwels 2011: 100). In other words, these definitions are not mutually exclusive, but can be combined together (Gidron and Bonikowski 2013: 14). For instance, drawing on the mobilizational and discursive understandings of populism, Jansen (2011) sees it as a mass political project that mobilizes marginalized social sectors into attention-capturing political action while

(10)

employing an anti-elite rhetoric that praises the ordinary people. The definition by Hawkins (2009) is a synthesis as well, merging the discursive and ideational approaches. To be more accurate, Hawkins considers populism as a ‘discourse’, which to him is equivalent to a set of ideas that have underlying meanings for political leaders and participants, and he argues that many other labels used for indicating the discursive definition of populism, such as a ‘language’ (Kazin 1998) or an ‘appeal’ (Canovan 1999), are all talking about the same thing. Hawkins uses the term discourse in the sense that combines elements of rhetoric and ideology, and defines populism as “a Manichaean discourse that identifies Good with a unified will of the people and Evil with a conspiring elite” (2009: 1042).

2.1 My definition of populism

In this thesis, I primarily use Mudde’s (2004) definition of populism as a thin ideology, or a set of ideas. On top of this, I would like to add that these ideas are especially reflected in a specific discourse, as stated in Pauwels’ definition (2011). To illustrate, populism is considered to be a thin ideology that sees the world in a dualistic fashion, with two homogeneous and opposed groups: the people and the elite. There is a moral dimension in this worldview, which valorizes the people as good and pure, and denigrates the elite as evil and corrupt. From a populist viewpoint, the elites are always conspiring against the people’s interest for their own benefits, albeit politics should be at all times carrying out the will of the people (Mudde 2004; Stanley 2011; Hawkins et al. 2016). Furthermore, these set of ideas are expressed by their proponents through their discourse which contains a people-centrist and anti-elitist sentiment. The

(11)

‘discourse’ I use in my definition is merely seen as a rhetoric or a political communication style without necessary connections to underlying ideas.

This definition of populism is complementary to the isolated ideational and discursive definitions, while it is also clearer and easier to be understood than Hawkins’ ‘discourse’, since it is simply a juxtaposition of two different concepts. The reason why discourse, among other things such as policy outputs or party organization, is included as a determining aspect is that, whereas the populist political styles are inconsistent across times and regions, scholars argue that populist discourse is consistent, and is what distinguishes the populists from the others (Rooduijn 2014). As an ideology, populism has its own vocabulary (Pauwels 2011: 100), and “it has a subconscious quality that manifests itself primarily in the language of those who hold it” (Hawkins 2009: 1045). Also, as Hawkins suggests, unlike common rhetoric that is merely used as a strategy, the use of populist discourse usually implies the belief in what it represents, because populist discourse embodies a radical notion of democracy with such an inflammatory style of speaking, making it unlikely for someone to fake it consistently (ibid.).

Besides the used definition of populism, it is also necessary to elaborate on the choice of the core elements that are included in this thesis. The core elements in the definition are people-centrism and anti-elitism, and the other characteristics of populism are all extensions based on these two elements. To illustrate, because populism primarily focuses on the people and how they demonize the elite as a whole, a dualistic vision of the society with two antagonistic groups takes shape. People-centrism is by far a minimal and necessary condition for populism, as the word

(12)

‘populism’ itself is derived from the Latin ‘populus’ which means ‘the people’ (Jagers and Walgrave 2007). This thesis treats anti-elitism as the other minimal and necessary condition because it is what makes the populist ideology and discourse special. Politicians constantly appeal to the people since they need their support, or more specific, their votes. Therefore, only when an anti-establishment message is added in the ideas and rhetoric of someone, we can categorize this person as an adherent of populism. As van Kessel (2011: 70) argues, “Populists are, in any case, opposed to the political powers that be.”

People-centrism and anti-elitism are present in some of the most influential academic definitions of populism (for example, Mudde 2004 and Hawkins 2009), and are included as the most common denominators of all traditions of definitions (e.g., Jagers and Walgrave 2007; Elchardus and Spruyt 2016). Stanley (2011: 258) sorts out four essential concepts which are “characteristics of all manifestations of populism”, but all four of them are either synonyms or extensions of people-centrism and anti-elitism.

The other elements considered as defining features of populism by some academics are dropped out here for good reasons. For instance, the presence of a charismatic leader is rejected as a necessary component, as explained earlier. Believers of populism have the affinity to endorse and follow someone who claims to incarnate their will and speak on behalf of them, but populism itself does not require such an existence of a spokesperson. Many others (e.g., Albertazzi and McDonnell 2008; Taguieff 1995) stress on the exclusion of some minority groups from ‘the people’, when these groups are considered hindrance for the homogeneous entity of the people to reach their

(13)

agendas, or when they are accused of being privileged by the elite (Jagers and Walgrave 2007). Nonetheless, this exclusionism is not a universal property of populism, and the difference is primarily regional: while European populism has an exclusive nature, Latin American populism has an inclusive characteristic (Mudde and Kaltwasser 2013). As Mudde and Kaltwasser claim, “populism always entails both exclusionary and inclusive features” (2013: 148). This ambivalent feature of populism is also reflected in the different references of ‘the people’, elaborated by Canovan (1999: 5). Populist actors can either appeal to ‘our people’, alienating and rejecting those who are defined as outsiders, or they can appeal to ‘the united people’, resisting the division of it by parties and factions. Indeed, other than exclusionism, it is feasible to infer an inclusionary sense from the definition of populism. Being people-centric means that populism focuses on the general interest of the people, and being anti-elite indicates that the people do not trust the elite. In other words, they trust themselves in terms of political decision-making, which implies the support of participatory democracy with a direct, plebiscitary mechanism to carry out their will. Such a mechanism, as the referendum, is inherently inclusive since it directly includes the most citizens to take part in the decision-making process. Moreover, even some European populist parties that are usually characterized as exclusionary, are inclusive regarding their political personnel. Unlike traditional mainstream parties that are usually represented by elder men, higher education and higher income, various populist parties in Europe tend to have younger, new and less educated members as their representatives (ibid.). The fact that populism can entail these two opposite traits makes it unreasonable to count either one of them as a defining element of populism.

(14)

It is important to stress that, based on my definition of populism, a populist actor can be any person or group of people, from an ordinary citizen to a political party, because they can all possess the populist set of ideas in mind and further manifest them in words.

One other point has to be made clear for analytical clarity in the following sections of this thesis. Populism is sometimes seen as an essential attribute of an actor, meaning that someone, or some political party, is either populist or non-populist (e.g., Mudde 2007). In other cases, populism is seen as a matter of degree, so it is possible to deem a party more populist than another (e.g., Pauwels 2011; Rooduijn et al. 2014). In my thesis, I go for the latter approach. According to Gidron and Bonikowski (2013: 8), the difference between these two understandings represents the distinction between the ideational and discursive conceptualizations of populism. The former one is linked with the ideational definition, and the latter is linked with the discursive definition. This distinction is, however, not evident anymore, as there are also scholars who adhere to the ideational definition of populism as a gradational inclination (e.g., Hawkins et al. 2016: 3).

Since in this thesis populism is regarded as a set of ideas expressed by a certain discourse, it will be considered as a matter of extent as well, because populist ideas and discourse can be present and utilized to different levels by populist actors. Furthermore, following this understanding enhances the possibilities for empirical investigation. Using the binary approach means that there needs to be a definite threshold for anyone to become populist. Nonetheless, there is no justified and objective way to set a threshold regarding the amount of populist ideas and statements for an individual to be

(15)

called populist (Pauwels 2011). Accordingly, forcing this either-or logic upon the objects of study is problematic. It does not consider the fact that populist ideology can be potentially held by many (Hawkins et al. 2016), as well as the situation that it varies per person in terms of the degree it is reflected in each person’s discourse. Instead, viewing populism as a matter of degree is like placing research objects on a populist spectrum, with one end being extremely populist and the other being not populist at all. Doing so not only allows us to capture different levels of populism by different actors, whether they are political parties, politicians or citizens, it also allows us to detect the variation of populist dispositions of these actors across contexts and overtime (Gidron and Bonikowski 2013; Hawkins et al. 2016; Rooduijn et al. 2014; Pauwels 2011).

2.2 The activation of populist attitudes

The activation theory, proposed by Hawkins, Kaltwasser and Andreadis (2016), will be applied in this thesis in order to understand why the orientation towards populism by the public may or may not have influence on their voting behaviors. In their paper, they begin with pointing out that studies on specific countries have already shown that the mass level populist attitudes are actually widespread and shared by half to two-third of the population of these countries, and that they correlate with support or voting for populist parties there (ibid.: 5). They then argue that, whether the electorate votes for the populist parties depends not only on the presence of populist attitudes in a society, but also on the context by which these attitudes are motivated. It is not enough to drive the voters to vote for populist parties only by believing in populism themselves, since, as explained above, populism is a thin ideology that has narrow programmatic scope and rarely exists on its own. Because there are usually several other ‘thick’ ideologies

(16)

simultaneously held by both the political parties and the individual voters that actually articulate comprehensive political programs, it is therefore not always the case that populism works as a salient factor. It is a relatively potential disposition that requires activation. This activation theory helps explain the variation of support for populist parties in different countries and periods of time.

3. Context of the study – Taiwan

Before the activation theory can be applied on the case-study of this thesis, it is necessary to delve into the political context first. Taiwan, once under Japanese colonization and then an authoritarian regime originated from China for nearly a whole century, has seen a peaceful transformation to a democracy which was praised as the “best-case democratization” (Rigger 2004). After the end of the World War II, control over Taiwan was given up by Japan and taken over by the regime Republic of China (ROC). The ROC was under the control of the Chinese Nationalist Party (the Kuomintang, or KMT) which was then based in mainland China. As defeated by the Communist Party of China (CPC) in the Chinese Civil War following the end of World War II, KMT retreated to Taiwan in 1949, and the regime, ROC, has been representing the island ever since (Tien and Shiau 1992; Rigger 2004).

Despite that the founding principle of ROC is to eventually establish a democratic political system, KMT introduced martial law in Taiwan and maintained single-party authoritarianism, based on the justification that Taiwan had to stay solidary and be prepared for restoring ROC rule over mainland China (ibid.). However, this decades-long-delay of democracy was gradually faced with severe challenges. From the

(17)

mid-1950s to the 1980s, Taiwan saw a rapid economic development with an annual economic growth of almost 9 percent, which cultivated the emergence of the middle class who demanded to participate in the island’s politics (Lee 2006). While the constant social mobilization undermined the authoritarian regime’s popular support, these pressures were intensified by KMT’s diminishing international support. Before the 1970s, ROC was the representative of China in the United Nations and had diplomatic relationships with most of the U.S.’s ally countries. Nonetheless, following the U.S.’s switch of strategic partnership to the communist People’s Republic of China (PRC), ROC stage by stage lost its international recognition. Thus, the only legitimacy ROC has to the international society was its identity of being the “Free China”, and its claim to become a democracy (Rigger 2004).

In an attempt to address all these pressures, the leader of KMT then, President Chiang Ching-kuo, chose the strategy of gradual concession over stiff suppression. He acquiesced the establishment of the first opposition party, the Democratic Progressive Party (DPP), in 1986, and one year later lifted the martial law that had been employed in the island for 37 years (Tien and Shiau 1992). Many restrictions have been removed as well, such as those on freedom of expression, assembly and association, as well as on traveling to mainland China (Lin 1999). Chiang’s successor from 1988, President Lee Teng-hui, even more progressively allied with the democratic movement in order to consolidate his position against dissidents from within KMT. He enabled a series of constitutional amendments that formed the basis of Taiwan’s current democratic system, especially the one which stipulates direct presidential elections (Rigger 2004). In 1996, the first direct presidential election ever in Taiwan took place. Out of the four

(18)

candidates Lee was elected, which gave him another term of four years as the head of ROC (Lee 2006). When his presidential term ended in 2000, Chen Shui-bian from the biggest opposition party, DPP, was elected as the new president, and Taiwan welcomed its first party alternation. This was followed-up by another alternation in in 2008 and a third one in 2016, and all three were peaceful and smooth (Epoch Times 2016). All in all, several factors combined made this transition from an authoritarian regime to a democratic one possible, including social and economic developments, domestic and international political landscapes, and pragmatic leadership. As a result, Taiwan did not have to go through the social upheavals that many others countries have gone through, while still achieving the same goal (Tien and Shiau 1992).

Throughout the years, Taiwan did not only stay as an exemplar for its process of democratization. More importantly, democratic governance in Taiwan has been successfully maintained without any major disturbance, and is in general functioning well (Bertelsmann Stiftung 2016). According to the Democracy Index 2016 from The Economist Intelligence Unit, Taiwan ranked 33rd

among all 167 countries, and is in the same category “Flawed Democracy” together with Japan, the U.S., France, etc (EIU 2017). By “Flawed Democracy”, it means that these countries “have free and fair elections and, even if there are problems……, basic civil liberties are respected” (ibid.: 54). Taiwan was classified into this category because it scored low on political culture and participation, but it performs well in terms of liberty, the functioning of government and elections (ibid.). Regarding public institutions and officials, the 2016 Corruption Perceptions Index (CPI) from Transparency International ranked Taiwan as 31st

out of 176 countries and 7th

(19)

3.1 The emergence of populism in Taiwan

In spite of the smooth transition to a democracy and the nowadays rather well-functioning democratic system, some factors exist that could make the political environment in Taiwan susceptible to populism. According to Hawkins et al. (2016), populist predispositions are widespread across all democracies but are only activated in some of them, and the most ideal context that serves this purpose is one with major failures of governance that result from elite collusion. Possible conditions include corruption and political unresponsiveness. In this sub-section, I will discuss what factors in the context of recent-year Taiwan might have activated the dormant populist attitudes among the Taiwanese, and provided a fertile ground for parties that hold the populist ideology.

Economic depression or crisis, although proven to be not a sufficient nor necessary condition for a populist surge, certainly helps to intensify the feeling of ordinary people that they have been betrayed or left out by their government, who only cares about the interest of large enterprises (Hawkins et al. 2016). Accordingly, economic recession might have created the background in which people in Taiwan are constantly dissatisfied with the government. After the steady economic growth in the second half of the twentieth century, Taiwan’s economy has been relatively stagnant, also when comparing to its neighbors such as South Korea, Hong Kong and China (NST n.d.; Hsiao 2016). For the salaried class in Taiwan, the real wages that are adjusted for the price level have only annually grown 0.1% (LTN 2017). Taiwan’s economic growth rate still has an average of around 4% annually from 2000 to 2016 (NST n.d.), and the Real GDP per capita has been steadily increasing as well (Chen and Kuo 2013). These

(20)

conflicting facts might indicate the problem of income distribution, that is, who gets the benefits from the economic growth, and the answer seems not to be the majority of the Taiwanese people. Wage stagnation has been a serious issue in Taiwan for two decades. In fact, the CPI adjusted average monthly wage has even been declining, and on a faster pace comparing to neighboring countries such as Japan and South Korea (Huang 2015; Yu 2013; Lin et al. 2017). Moreover, wage stagnation has, combined with other factors, led to some bad social and economic consequences. For example, statistics shows that Taipei City, the capital of Taiwan, had the highest housing-to-income ratio in the world in 2014, meaning that houses in Taipei are greatly unaffordable for its residents (Haridasani 2015). While the two mainstream parties, KMT and DPP, are both right-leaning parties that stress heavily on fostering economic growth, free market economy and bilateral or multilateral free trade agreements with other countries, they do not pay enough attention to make sure if the ordinary citizens also share the gains of national economic growth instead of suffering from rising prices and stagnant salaries (Young 2015). As a consequence, more and more Taiwanese people, especially the younger generations, have focused on and demanded distributive justice within the society (Zhong and Jiu 2016). This does not only lead to the support of left-leaning, socialist parties or policies, but might also activate the underlying perception of some people that considers the government and big companies to be a group of elites who conspire together against the interest of the common people. Another possible reason that might have potentially activated the populist tendencies among Taiwanese people and contributed to the emergence of populist forces, is the declining (traditional)party identification. KMT, as the party that singly

(21)

ruled Taiwan for more than half a century, and DPP, as the first opposition party founded which united most of the opposition forces back then, have been dominating Taiwan’s politics ever since it turned into a democracy, basically making it into a two-party system. Although several small parties do exist, in the nine legislative elections that have been held in Taiwan, KMT and DPP together often won at least 80% of the total seats (CEC n.d.). Moreover, the small parties who have ever entered Legislative

Yuan (i.e. the parliament of Taiwan), or LY, before the 2016 election were usually

formed out of splits from KMT or DPP, and as the representative color of KMT being blue and that of DPP being green, these small parties are always parts of either the pan-blue or the pan-green coalition (Sullivan 2012). Accordingly, party identification is usually one of the most important determinants of the voting decisions of the Taiwanese electorate, and politics in Taiwan has been characterized by the competition and confrontation of the two parties and their followers (Bao 2013). However, this political antagonism has made them boycotting each other’s bills, polarized the society, while at the same time none of them has resolved to fix the plight of the ordinary people, such as the wage stagnation and housing price issue mentioned above. Increasingly more people, especially the younger generations, have perceived this traditional party division as a hindrance to the effective functioning of Taiwan’s political system (Tong 2016; Fan 2015). They consider the two major parties, or the political establishment as a whole, as corrupt, unresponsive, and careless of the needs of the people (Chen 2016; Hong 2015). This is emphasized by the increase of self-proclaimed neutral voters, who do not have a specific party identification, especially in the context of Taiwan’s two-party system. According to a monthly survey conducted since 2016, the proportion of

(22)

these neutral voters has been fluctuating between 27% to 43%, and many times over 35% (TPOF 2017; Yin 2016). This is a substantive increase compared to before, for example, to the number of approximately 20% in 2008 (Sheng 2002). This boost of independent voters has created a market gap for new political forces, and considering the reasons of their lack of party identification, a populist party might fit their appetite. The collective financial frustration, and hence life-frustration, of many Taiwanese citizens, combined with their long-term disappointment with the government and aversion of the old establishment, can theoretically contribute to the activation of populist attitudes among the public, because they correspond with the conditions of corruption and political unresponsiveness by Hawkins et al. (2016). Just before the 2016 election, it was said that the Taiwanese people felt there was a huge distance between them and the government (Van der Horst 2016). Therefore, a party that claims to stand with the people and against the elites might particularly attract votes because of its populist discourse to which the voters relate.

Another possible condition that makes a society susceptible to populism, according to Doyle (2011), is when political institutions are perceived to be illegitimate. In 2014, a massive social movement, known as the Sunflower Movement, erupted in Taiwan. It lasted for a total of 23 days with the LY and surrounding roads occupied by numerous protesters. Participants were mainly university students in the beginning, but they were joined by people of all ages and backgrounds later on (Chao 2014). The incident that triggered this movement was an ignorance of the legislative procedure by KMT legislator Chang Ching-chung, who abruptly suspended the review of a highly controversial trade deal between Taiwan and China and declared that it was ready for a

(23)

final vote of ratification (Smith 2014). The inability of the LY to prevent this illegitimate, unilateral action by KMT from happening was considered as a malfunction of Taiwan’s democratic system, and people thus craved for changes and alternatives (Van der Horst 2016; Hong 2015). Accordingly, some leaders and well-known supporters of the movement formed two new, third-power parties in 2015, including one that has sometimes been described as a populist party (for elaboration see the

Section 3.2).

3.2 The 2016 election and the 4 parties of study

The focus of this thesis is the 2016 legislative election, as a part of the 2016 general election. There are currently 113 seats in the LY, the law-making body of Taiwan. The members are elected through three different methods: (1) 73 legislators are directly elected through geographically-based single-member districts; (2) 34 are elected by party-list proportional representation according to the nationwide party votes each political party get; and (3) six are exclusively for aboriginal legislators. Accordingly, every citizen gets two votes for the legislative election, one to directly vote for a candidate in his or her district, and one to vote for a political party. Because voters’ decisions in the single-member districts are largely influenced by factors other than party identification/preference, such as candidates’ personal characteristics, in order to better study the link between voters’ populist attitudes and the parties they support, I will only look at their party votes in the 2016 legislative election.

The reason for centering on this particular election instead of taking more previous elections into account is that, there are some significant changes in Taiwan’s politics in the past two to four years. Observers have asserted that “the political base of Taiwan

(24)

has changed structurally and revolutionary”, and partisan realignment has happened (Horwitz 2016; Newtalk 2016; You 2016). These changes were manifested in the results of the 2016 election. First of all, this was the first time DPP won more than half of the total 113 seats, meaning the first time that LY is not under KMT’s, or its pan-blue coalition’s control. The pan-pan-blue coalition had always taken more than half of the legislative seats in all previous 8 elections, dominating LY for 23 years even with the competition of other parties. Out of the 8 elections, KMT singly won more than half of the seats for 6 times.

The power and influence of the richest, biggest (in terms of the number of party members), and long-standing party in Taiwan is clearly declining. Monthly polls after the election also show that the support of KMT is continuously low (TPOF 2017). Secondly, it was the first time a third-power party, namely a party independent from the traditional blue-green division, made its way into the LY. The New Power Party (NPP), founded only in 2015 by former Sunflower Movement activists, raised up to be the third biggest party in the LY after DPP and KMT. In addition, the conditions of activating populism mentioned above only happened or intensified in recent years and same did the establishment of NPP, who has been sometimes called “populist” (Zhang Y.S. 2016; Zhang Z.C. 2016; Zhang 2017). All in all, since the focus of this thesis is not the temporal change in Taiwan’s politics or voter behaviors, a focus on the most recent election and years under a context that theoretically fits this study is both feasible and suitable.

Four parties that took part in this legislative election will be the targets of study: KMT, DPP, NPP, and Social Democratic Party (SDP). While the first three parties have

(25)

been touched upon already, I will briefly introduce SDP and elaborate some more on NPP as well. In this election, SDP formed an alliance with another small party, Green Party Taiwan, which had been established for two decades but never made it into the LY. Because there was no joint official online site of this alliance that addresses its external communication, and SDP has been the center of attention both before and after the election (Fang 2016; Tseng 2015), I will only study SDP instead of the alliance in this thesis. The SDP is a young party that, just as NPP, grew out of the activists of Sunflower Movement and was founded in 2015. It carries a vivid left-wing banner that highly stresses on issues as distributive justice, labor rights and social welfare (Wen 2015; Jiang 2016; YC 2016). While NPP also focuses on redistribution and social justice, it does not brand itself that much as a left-wing party, and some of its other emphases that characterize it include reformations of the current political system and the independent Taiwanese identity (NPP n.d.; YC 2016). The legislative candidates of NPP are mostly well-known public figures, such as Freddy Lim, the vocalist of the internationally renowned heavy-metal band Chthonic who also once served as chair of Amnesty International Taiwan. Hence, NPP enjoys a higher publicity than SDP, whose leader is a university professor and gets its support mainly from people in the academia (YC 2016; Chou 2016; TPN 2016). Some commentators or politicians have labeled NPP as a populist party, based on its grassroots feature, its constant appeal to the people, and its popular, charismatic candidates (Zhang, Y.S. 2016; Zhang, Z.C. 2016; Wang 2016; Huang 2016). Nevertheless, most of these are not part of the theoretical definition of populism. Thus, whether NPP is more populist than the others remains to be tested.

(26)

The four parties are all in the top six in terms of the party votes they have received (see Table 1) in the 2016 legislative election. For all the other parties, the sample sizes from the TEDS 2016 survey database are too small to be considered usable. Among the top six parties, the third, People First Party and the fifth, New Party, were left out due to the disproportion of the votes they have gained and the “likes” they have received on their Facebook pages. Facebook posts from the official pages of political parties are used in this thesis to analyze populist discourse (for justification, see Section 5.2). Thus, it is necessary to make sure these parties have considerable amounts of followers on their Facebook pages, which guarantees that at least part of their supports do receive their populist message, if there is any.

Table 1. Results of the 2016 legislative election, proportional representation.

Party Votes % Seats

DPP 5,370,953 44% 18 KMT 3,280,949 26.9% 11 PFP 794,838 6.5% 3 NPP 744,315 6.1% 2 NP 510,074 4.18% 0 SDP 308,106 2.1% 0

Note: Retrieved from Central Election Commision (CEC).

4. Analytical framework

With the theoretical and conceptual notions of populism made clear, I can now apply my definition of populism and present the analytical framework of this thesis. As

(27)

mentioned above, I argue that populist actors can be anyone, which implies that the populist set of ideas are held not only by politicians or leaders of political movements, but also by ordinary folks. As a worldview that centers on how politics should be, two types of populist actors can be drawn: those who provide populism and those who demand it. The supply-side of populism consists of politicians and political parties, since these are the actors that are capable of transforming politics and creating a political system that fits the populist ideas. The demand-side is formed by the politically powerless people who crave for a change in politics based on their populist ideology. While these two sides seem to be naturally interrelated, most researches have however been only focused on either one of them, instead of considering both of them at the same time and studying the link between both (a few exceptions that to a greater or lesser extent address this point are: Stanley 2011; Akkerman et al. 2013; Hawkins et al. 2016; Marcos et al. 2016). This thesis mainly builds on Hawkins, Kaltwasser and Andreadis’ study (2016), in which they propose a theory of populist voting that directly considers the voters’ populist tendency. So far, most of the studies that tap into the electoral support for populist forces have relied on indirect or proxy measures of populism among the public, such as their economic and employment conditions under the circumstance of trade openness and economic globalization (Oesch 2008), their levels of trust in political institutions (Doyle 2011), their attitudes towards immigrant or refugee policies (Ivarsflaten 2008), and their specific personality trait (Bakker et al. 2015). However, as Hawkins et al. point out (2016: 4), “the point of the ideational definition of populism is that there is an additional layer of ideas that politicians are expressing and that voters may respond to”. Thus, it is crucial to prove that the affinity

(28)

of these populist ideas actually influences the people to vote for populist parties. Doing so will also answer the doubt on the real-world importance of both populist discourse and populist attitudes: populist parties convey their populist ideas through discourse to the public. Those with a higher populist attitude respond to these parties through voting for them. In other words, both the populist discourse and attitudes have concrete political consequences (Hawkins 2009).

Before assessing this relationship between having populist attitudes and voting for populist parties, the more populist parties and people have to be identified by measuring their levels of populism. Nevertheless, the ways of measuring the levels for these two groups of objectives are different due to practical concerns. For the voters, there is a lack of available record of what they say or write. The most possible space to find their daily discourse is nowadays through social media, but people’s accounts there are usually private, or only accessible for their friends. Interviews may be utilized to detect individuals’ populist discourse and attitudes, but it is not feasible for a large-N research. Therefore, at least in recent studies on the levels of populism of the public, including in this thesis, only their attitudes towards populist ideas can be uncovered, through opinion surveys with related questions. For the political parties, on the contrary, it is impossible to know what they think, the only way to detect populism is through what these people openly express, what they write and say out loud.

Using the case of the 2016 legislative election of Taiwan, the purpose of this thesis is to carry out a complete operation in order to study the link between populist attitudes and voting decisions for populist parties. The analysis is separated into two steps. In the first study, I will attempt to get an overview of the populist levels of the four

(29)

Taiwanese parties through the method of content analysis, identifying some of them as more populist than others. As mentioned in Section 3.2, the party NPP has been labeled “populist” several times in recent years, thus I expect the result to show that NPP is the most populist party among all four. This is the first hypothesis in my thesis:

Hypothesis 1 (H1): NPP is the most populist party among all four parties that are analyzed, in terms of their level of populist discourse.

In the second study, I will first assess the affinity for populism of the Taiwanese public based on existing data. Thereafter, I will examine if these populist attitudes that voters possess actually play a role in their voting decisions for populist parties, or if other factors are still far more significant in this vote-choice model. Based on the activation theory, populist attitudes of the electorate should be at least a modest predictor of the vote in Taiwan, comparable to the result in Chile in the study of Hawkins et al. (2016). In terms of the context, there are many similarities between modern Taiwan and Chile. They are both part of the third democratic wave, have experienced successful economic modernization, and are socially and politically stable. That said, they also share the common situation in which political unresponsiveness is causing major dissatisfaction with government, and triggering massive waves of protests. Hence, here lies my second hypothesis:

Hypothesis 2 (H2): Taiwanese voters’ populist attitudes predict their populist voting behavior.

(30)

5. Study 1: Measuring Populist Discourse

Since there is no research that studies the tendency of populism for any of the Taiwanese political parties, the first step is to identify which ones of them are (more)populist, in order to further test my hypotheses. To make sure there is systematic and empirical justification to label any party as populist, a refined method to measure the populist discourse of a party is needed.

5.1 Existing literature and approaches

Mudde (2004) uses minimal definition to sift out parties that meet the minimal criteria to be called populist parties. Although minimal definition provides a good starting point about what elements a party’s discourse should contain, this method is a dichotomous classification, treating populism as a fixed ideology, meaning that a party can only be either populist or not. As explained above, this understanding of populism, and therefore this method of minimal definition, is not sufficient to measure different levels of populism. Accordingly, scholars in this field have developed several more quantitative ways to measure populist discourse, which are all under the broad category of content analysis.

The first researchers to employ content analysis to especially measure populism were Jagers and Walgrave (2007). Armony and Armony (2005) preceded them by assessing the discourse in two Argentine presidents’ speeches, but measuring populism was not their aim. Jagers and Walgrave use the so-called classical content analysis, in which human coders are trained to analyze texts by means of a codebook. They first look at the manifest content of the texts through measuring the number and intensity of references to the defining elements of populism they list, and further code these

(31)

references differently based on their types and nature, which required more human interpretation. The unit of analysis of this approach is usually the sentence or the paragraph. Using classical content analysis to fathom populist discourse has gained some prominence in recent years (Pauwels 2011), and has been applied in several other researches (for example Rooduijn and Pauwels 2011; Rooduijn et al. 2012). Nevertheless, although this approach is more rigorous than that of minimal definition, it has some noticeable drawbacks: it is extremely time-consuming, and there are concerns about coder subjectivity and unreliability (Pauwels 2011: 98).

A quantitative text analysis can circumvent these problems. As a variant of content analysis, quantitative text analysis sees texts simply as data in the form of words without the need to understand or interpret underlying meanings. Hence, computers, instead of human coders, are involved, making this method efficient for large-scale analyses (Pauwels 2011). While there are multiple approaches within this emerging field, the one that has been used for measuring populism is the dictionary based approach (see Pauwels 2011; Rooduijn and Pauwels 2011). Before conducting such an analysis, a dictionary has to be created. In the dictionary, categories are made depending on which dimensions of the texts the researcher is interested in, and specific words are allocated to these categories based on a priori reasoning and empirical observations. Thereafter, a computer scans through each and every word of the texts, which is the unit of analysis of this method, codes them according to the categories they belong to and counts the proportion of them. The advantages of this approach are apparent: it is more systematic and reliable since no human-codification is involved, and it is the least time-consuming among all the methods discussed so far. Although some sceptics have

(32)

doubts on this computer-based approach since the context might change the meaning of a word, most of the time it is possible to identify words unambiguously (Laver and Garry 2000; Pauwels 2011).

Because there are strengths and weaknesses of both the human-based and computer-based content analysis, Rooduijn and Pauwels (2011) suggest to combine the two of them, taking advantage of the strong points from each method. This integrated man-machine approach they proposed begins with the normal procedure of quantitative text analysis. After yielding the results by a computer, human coders step in and interpret those results with a codebook, double-checking if the selected words are indeed meaningful for the research purpose. This combined method takes less time than a classical content analysis and produces more valid results than a merely computer-coded one. The same concept is adopted by Bonikowski and Gidron’s work (2016), where they apply a so-called dictionary-based automatic text analysis. Having over 2000 documents to be evaluated, they first classify each of them dichotomously as either populist or not by their algorithm. The algorithm also identifies those passages with dictionary terms in them in the populist documents, so the researchers can further read through all these excerpts to check if there is any incorrect classification. The way Bonikowski and Gidron develop their dictionary relies more on the computational tool. They begin with reading a random part of the texts and selecting potentially relevant terms, using these terms to run the analysis, then the algorithm will identify what additional terms in the texts are relevant as well. They repeat this process and keep on building up their dictionary until no more new relevant terms are found.

(33)

A less quantitative approach of content analysis has also been attempted. Hawkins (2009) innovatively applies the means of holistic grading to measure populism in speeches from chief executives in 34 countries across continents. The unit of analysis of this technique is the entire text, so the readers assess the overall qualities of a piece of speech and assign it a single grade based on a pre-designed rubric. Grading is conducted on a three-point scale with 0 indicating no populist elements in the text, 1 indicating a moderately populist text and 2 indicating an extremely populist text. Because Hawkins sees discourse as “a broad, latent set of meanings in a text” (p. 1049), and the ideas that form the discourse are held subconsciously, measuring populist discourse should therefore also take into account the tone and style of the language other than only the actual words. As a result, holistic grading is suitable for this purpose, whereas a computer-based codification is inappropriate, and a human-coded classical content analysis takes too much time. Some of the problems of this method are, for example, insufficient reliability, possible subjectivity of the coders, and rough measurement due to the 3-point-scale (Pauwels 2011: 102).

5.2 Data and method

To assess the level of populism of the four Taiwanese parties, I will rely on their official Facebook pages. While the previous studies on populist discourse have focused on speeches, party manifestos, television debates, broadcastings, et cetera., I believe that in the Taiwanese case, using Facebook as the data of analysis can be an adequate and promising option. On the one hand, all four parties are active on Facebook with an average of one post per day, which makes it reasonable to assume that their ideologies or beliefs are more or less expressed through this platform, including their possible

(34)

tendencies towards populism. On the other hand, Taiwan has the highest penetration rate of Facebook in the whole world as of 2016 (Epoch Times 2016). 82% of the population is at least monthly active on this social media site. Since Facebook is so prevalent in Taiwan, and someone as a supporter of a political party is likely to “like” its Facebook page, it is thus reasonable to assume that the Taiwanese voters are more or less exposed to the discourse of political parties through Facebook. Although, admittedly, this choice of data results in the necessity to drop out two of the top six parties from the analysis, the value of analyzing Facebook posts for the four remaining parties can compensate for this loss.

The quantitative text analysis of Pauwels (2011) is used to measure the populist discourse of the four Taiwanese parties. As mentioned, minimal definition is not capable of detecting different levels of populist discourse that I want to identify. Given the difficulty of manually coding over all the posts from four parties in a period of six months all by my own, a classical content analysis is also not considered. Even though I adopt the integrated method and only double-check those passages with populist terms that are sifted out by the computer, still it will leave me 589 passages to read (because among all the texts there are 589 populist terms found). The same concern about a labour intensive and time-consuming process applies to holistic grading, due to its level of measurement as a full text and the type of data I am analyzing. Using this method means that I will have to read through and grade each and every post (i.e. at least 720 pieces), which is barely feasible. Therefore, taking the practical considerations into account, a purely computer-based analysis is evidently my best option. Moreover, this method is entirely compatible with my definition of populism as a thin ideology that is

(35)

expressed through a certain language. Unlike Hawkins (2009) who stresses the latent characteristics of populist ideas, my definition implies that populist discourse can be explicit, especially by iterating the populist ‘vocabularies’ listed in the pre-determined dictionary. While this quantitative approach comes in great reliability, since there is no issue of coder subjectivity, its validity is lower than that of an approach involving human-coding. However, previous experiences point out that the computerized method is still sufficiently valid (Pauwels 2011; Rooduijn and Pauwels 2011).

In order to come up with these ‘vocabularies’, I first listed the categories corresponding to the core elements of populism. Accordingly, there are the categories of people-centrism and anti-elitism. As the central features of populism, a discourse should contain terms that indicate these two ideas for it to be identified as populist. These two concepts are also taken as the targets of scrutinizing by other studies on populist discourse (for example Pauwels 2011; Rooduijn & Pauwels 2011; Rooduijn et al. 2014). In the people-centrism category, six ways of referring to the people in Chinese are included, such as “people” (人民 rén mín), “citizen” (公民 gōng mín) and “voter” (選民 xuǎn mín). In the anti-elitism category, two kinds of words are listed. The first are those that directly refer to the elite, such as “elite” (菁英 jīng yīng), “the rich and powerful” (權貴 quán guì)or “plutocrat” (財團 cái tuán). The second type of words reflect the “anti” part of this category, thus words with negative connotation or criticism are included, such as “corrupt” (腐敗 fǔ bài), “shameless” (無恥 wú chǐ) and “distrust” (不信任 bù xìn rèn). Besides these two categories of terms derived from the core concepts of populism, a third, auxiliary category is added, based on the common strategy of populists that demands direct democracy and its devices. “Direct” (直接 zhí

(36)

jiē) and “referendum” (公投 gōng tóu) are included in this category of direct democracy. For the selection of the words, I first consulted the dictionaries in pre-existing studies on populist discourse that have used this technique (including Pauwels 2011; Rooduijn and Pauwels 2011; Bonikowski and Gidron 2016) and picked words that have counterparts in Chinese and fit the Taiwanese context. Thereafter I randomly read passages from all the texts to get inspirations for more terms that could be included. For the analysis, I will gather all the accessible posts from each party’s Facebook page in the six months prior to the election day, which is from 15/08/2015 to 15/01/2016. I will simply use the search function in Microsoft Word to find out how many times those words that are included in the dictionary appear in each party’s texts. After knowing the numbers of occurrence, I will calculate the proportion of populist vocabularies of the overall posts of each party. These numbers of proportion will be the indication of the level of populist discourse of each party. The reason why I look at all the posts instead of randomly selecting some of them from each party so as to acquire a mean populist score for each of them is that most of the posts are quite short, thus doing so is likely to render the result that a lot of the randomly chosen posts will contain no populist term.

5.3 Results

Table 2 shows the results of the text analysis. The process of data collection results in

a variant amount of texts for each party, ranging from 20,186 words to 86,178 words, due to the differences in their frequency of posting and length of posts. However, since I only take into account the proportion rather than the actual number of terms, this variance should not be a problem. Results of the analysis reveal that NPP is by far the

(37)

most populist party in terms of their public rhetoric, H1 (NPP is the most populist party among all four parties that are analyzed, in terms of their level of populist discourse) is therefore confirmed. The oldest, at that time incumbent party KMT is the least populist. The weakness of the result is, however, that the proportions of populist words are rather marginal for all four parties. This might be because of the fact that not many terms are actually included in the dictionary, or it might indicate that some Chinese terms that better capture the essence of populist ideas have yet to be discovered. Nonetheless, the differences between their levels of populist discourse are still discernable, and corresponding to the pre-analysis expectations as well. Notably, both KMT and DPP did not include any of the word in the category of direct democracy – “direct” and “referendum” – at all, in their numerous posts during a period of six months. This expresses the common impression that direct democracy is usually a tool utilized by populist parties (Crick 2005), and KMT and DPP, as the two long-lasting established political parties in Taiwan, are not particular fans of it.

Table 2. Proportion of populist discourse of the four parties.

KMT DPP NPP SDP

The people 72 137 263 39

Anti-elite 6 2 43 11

Direct democracy 0 0 15 1

Total numbers of populist

terms 78 139 321 51

Total words 44794 62450 82178 20186

Proportion of populist

(38)

6. Study 2: Populist Attitudes and Voting Decision

There are crucial reasons to measure the latent populist attitudes among the people. So far, a number of studies have focused on explaining the electoral support for populist parties (e.g., Van Kessel 2011; Bakker et al. 2016), but the voters who opt for these parties do not necessarily do so because of the populist images of them. Voters are usually recruited based on multiple issues and concerns, therefore, the support for populist parties and the support for populism per se should be distinguished. People’s support for populism is considered equivalent to their populist attitudes, because, as emphasized above, the essential ideas of populism inhere not in tangible policies, but in meta-political views and beliefs. Hence, if some voters are said to be responding to populism, it must indicate their explicit convictions to these ideas, rather than resulting from their socio-economic identities, their issue stances, or their policy preferences (Stanley 2011: 261). Furthermore, many scholars have argued that these populist attitudes are held by a significant part of the electorate, independent from party preferences (Stanley 2011; Akkerman et al. 2014; Hawkins et al. 2012; Rooduijn 2014; Spruyt et al. 2016; Elchardus and Spruyt 2016). Instead of being created by populist leaders or parties, these sentiments are preexisting among the population and can resonate with populist political actors or be activated by certain contexts (Hawkins et al. 2016). Given these facts, the direct measurement of populist attitudes is a necessity to improve the understanding of populism. In addition, although it is not the purpose of this thesis, measuring populist attitudes can be extended to assessing who holds populist attitudes, and even further to explore the cause of this affinity.

(39)

Once the populist attitudes of the voters are detected, it is possible to test the effect of these attitudes on their party preferences or voting decisions. Moreover, it is crucial to see if populist attitudes have impact on the support or vote for the populist parties in particular, so as to talk about their real-world influences.

6.1 Existing literature and approaches

The first attempt to measure populist attitudes among individuals was published by Robert Axelrod (1967), using cluster analysis to assess public opinions. Nonetheless, the populist attitudes he was gauging half a century ago are different from the contemporary conceptualizations of populism. Although one of his clusters is named the “populism cluster”, it is named this way because it corresponded to the attitudes of the American Populist movement of the 1890’s, which focuses on U.S. government policies on, for example, education aid, medical care and the involvement of foreign affairs. Thus, these indicators of populist attitudes are outdated and irrelevant to what we are studying nowadays (Hawkins et al. 2012; Akkerman et al. 2014). A few other studies have partly tapped into the assessment of micro-level populism but are not meant for measuring populism per se. For instance, Dryzek and Berejikian (1993) find that a certain type of discourse, which closely matches with the modern definition of populism as a thin ideology, exists among their sample of Americans, but their finding is never applied to larger samples of the public. Hibbing and Theiss-Morse (2002), on the other hand, use statements such as “elected officials would help the country more if they would stop talking and just take action on important problems” to measure the sentiment of popular cynicism the U.S., which resembles one of the key ideas of populism that politics is the antagonism between the elite and the people, but the term

Populist attitudes and populist voting : the case of the 2016 legislative election in Taiwan