Political polarization, a phenomenon when subsets of a population divide into two sharply opposing political camps and diminishingly share political ground (Carothers & O’Donohue, 2019; Fiorina & Abrams, 2008; Iyengar et al., 2019), is one of the crucial causes of democracy “backsliding” around the world during the past decades (McCoy et al., 2018; Stavrakakis, 2018). In Thailand, the colored-shirt confrontation that has twice invited the military to step in to punctuate the political turmoil in 2006 and 2014 is one of the clear examples of serious cases of political polarization. This political polarization, usually referred to as a political contest between “Yellow Shirt” and “Red Shirt” supporters, has been described in much of the literature as a deep societal division that cuts across many types of social classes, regions, and party identifications (Hewison, 2015; Huang & Thananithichot, 2018).
On the one side, the Yellow Shirt movement that emerged and was first united as the People’s Alliance for Democracy (PAD) in February 2006 in order to oppose the government of Thaksin Shinawatra includes a large proportion of well-off urban people with royalist sentiments (Hewison, 2015). On the other side, the Red Shirt movement that first formed as the United Front for Democracy against Dictatorship (UDD) in September 2006 to oppose the military coup, which overthrew the government of Thaksin five weeks before the scheduled election (Forsyth, 2010), has been delineated as the political movement of the poor rural-born masses (Nishizaki, 2014; Taylor, 2012; Thabchumpon & McCargo, 2011). Both movements grew rapidly with hundreds of thousands of citizens joining each camp’s several demonstrations during the past decade.
Is Thailand really as politically polarized along color lines as is commonly argued? This paper provides an empirical study in answer to this question. Studies providing explanations regarding the divergence between the Yellow Shirt and Red Shirt movements are numerous. However, an answer to the aforementioned question is still important because almost all of those previous studies relied on qualitative methods. Survey studies dealing with the demographic, identical, and ideological compositions of the two camps’ supporters are rare, and some of them are outdated and do not show results that represent all eligible Thai voters. In addition, and more importantly, very few of them test models showing the direction and extent to which the key factors have an impact on the polarization. Assisted by improved survey data and analytical techniques, this paper not only identifies the key differences between the Yellow Shirt and Red Shirt supporters, but also indicates the power of each driver in explaining the political polarization between the two opposing sides.
The ongoing political polarization between Yellow Shirt and Red Shirt supporters that followed the 2006 coup has had a massive impact on Thailand’s democracy. A number of studies have paid attention to explaining the root causes of this phenomenon. In the early period of the polarization, Hewison (2015, p. 57) depicted this phenomenon as “a tussle of competing elites, with a rising elite, associated with Thaksin, challenging the long-dominant conservative elite of palace-connected military leaders, big business/old money and technocrats.” However, as the conﬂict deepened, especially after the 2006 military coup, society-wide mobilization and mass polarization has existed in Thai society (Hewison, 2014). For many scholars, because the divide between the two opposing groups of citizens is rooted in class (the urban middle-class Yellow Shirts versus the unprivileged rural Red Shirts) differences, the political polarization in Thailand has been described as a conflict of social class in which a minority urban middle-class has tried to hold the upper hand against the rural masses (Charoensin-o-larn, 2013; Funston, 2009; Jäger, 2012). This explanation is quite similar to that of Anek Laothamathas’ “tale of two democracies” thesis, which is one of the most recognized and influential arguments about Thailand’s democracy in the 1990s (McCargo, 2017; Thananithichot, 2012).
According to this thesis, the reason why democracy has failed to be firmly established in Thailand is to be found in the conflicting attitudes over democracy and voting behavior of the well-educated middle-class in Bangkok and poor rural voters residing in villages (provincial areas) (Laothamathas, 1996). In particular, the urban middle-class voters ideally view democracy as a form of legitimate rule adopted by most civilized nations. They also expect that elections should be mechanisms for recruiting honest and capable persons to serve as lawmakers and political executives rather than a process through which voters become “parochial in outlook, boorish in manner, and too uneducated to be competent lawmakers or cabinet members” (Laothamathas, 1996, p. 208). However, in practice, elections in Thailand, in the eyes of the urban middle-class, remain “an invalid source of regime legitimacy” because of vote-buying, and it is felt that the majority of Thai voters cannot choose their representatives independently and do not have “responsible judgment” (Laothamathas, 1996, p. 214-215).
In contrast to the urban middle-class electorate, rural voters, according to Laothamathas (1996), view democracy as a mechanism to draw greater benefits from the politicians to their communities and to themselves. Elections, in the rural electorate’s opinion, are therefore very much local, not national affairs, dealing with the exchange of votes for constituency services, and relieving grievances or tangible benefits such as public works for the communities rather than abstract rewards such as laws, policies, or public interest (Laothamathas, 1996, p. 212). The existence in society of the conflicting views on democracy and voting behavior of the urban middle-class and the rural poor, as described above, has led to instability in democracy in Thailand, in which the rural majority votes to set up a government while the less in number but louder-voiced urban middle-class criticizes and weakens the cabinet, which finally ends with either its own internal conflicts or an external military coup (Laothamathas, 1996).
However, for many other scholars, the tale of two democracies thesis is appropriate for explaining what created the Yellow Shirt and Red Shirt difference only at the beginning of the period when the conflicts and polarization occurred (before the 2006 coup). As many studies (e.g., Siamwalla & Jitsuchon, 2012; Sinpeng & Kuhonta, 2012; Thabchumpon & McCargo, 2011) have asserted, people that identify themselves as being close to either the Yellow Shirt or the Red Shirt movements are socioeconomically mixed and are composed of heterogeneous groups. There are many Red Shirt protesters and supporters that are middle-class, earn a high income, and have had educational opportunities, and many Yellow Shirt protesters and supporters are working-class, earn a low income, are less-educated, and come from provincial areas, even though the majority of the Red Shirt supporters are rural residents that have lower incomes and a lower level of education than the majority of the Yellow Shirt supporters, who tend to come from big cities.
Moreover, emerging as anti-Thaksin on the one side and as pro-Thaksin on the other, both political camps’ supporters have identified themselves with one specific party over the other party—i.e., the Yellow Shirt supporters with the Democrat Party, the opposition to the pro-Thaksin parties in the 2007 and 2011 elections, and the Red Shirts with the pro-Thaksin parties (the People’s Power Party in the 2007 election and then the Pheu Thai Party in the 2011 election). Based on their party identification, the majority of the Yellow Shirt movement’s supporters came from the Bangkok middle class and from the Democrat Party’s strong electoral base in the south (Pye & Schaffar, 2008; Thabchumpon & McCargo, 2011). Large numbers of Red Shirt movement supporters, in contrast, tend to be of the Bangkok working class and came from the densely populated north and northeastern parts of the country, the pro-Thaksin party’s stronghold (Hewison, 2015; Nishizaki, 2014). Primarily driven by the establishment of well-off urban people with royalist sentiments, key aspects of the Yellow Shirt movement’s rhetoric were its anticorruption focus, protection of the monarchy, and a growing opposition to electoral politics (Hewison, 2015), while the Red Shirt movement consisted largely of the awakening of the rural people with the consciousness of inequality and unfair allocation of power and resources (Nishizaki, 2014; Thabchumpon & McCargo, 2011).
The above discussion shows that the nature of the color-based political polarization in Thailand is more difficult to understand than simply as a population divide between the two opposing groups of people that have differences in socioeconomic status (lower-higher levels of income) or in areas of living (Bangkok-province). What divides the Yellow Shirt and Red Shirt supporters also includes the regional base, party identification, and political predisposition. However, little is known about the extent to which each of these key factors has an impact on the polarization and which is the most powerful driver of the polarization between the two opposing camps. Aiming to fill this gap, this paper tests models incorporating all of the influential factors identified by previous studies—demographics or socioeconomic status, party identification, and democratic values—into the model.
Data and Methods
This paper uses a national survey conducted by the author between July and August 2017. The respondents to this survey were obtained in a statistically representative national sample of a minimum 800 adults (18 years old and above). The baseline information and addresses were drawn from the household information compiled by the Community Development Department, Ministry of Interior, for the respondents in the northern, northeastern, central, and southern regions. For the respondents in Bangkok, the baseline information and addresses were from the district offices of the Bangkok Metropolitan Administration.
The sample was allocated in Bangkok and four regions of Thailand by using a three-step stratified sample selection. The stages include: (1) stratified sampling for specifying the locations to be studied, which were divided into four regions and Bangkok, for a total of five provinces; (2) systematic random sampling of four legislative constituency voting units from the entire unit in Bangkok and systematic random sampling of two legislative constituencies voting units in the urban area (in the area of the municipality) and two legislative constituency voting units in the rural area (outside the area of the municipality) from the entire unit in the other four provinces; and (3) systematic random sampling of respondents from across 20 voting units producing an N of 800 and following distribution by region compared to the official numbers of Thai eligible voters in the constitutional referendum held on August 7, 2016:
An indicator of data quality is the standard error of the estimate, on which the margin for sampling error is based. As survey statistics are mostly proportions, the key measure of data precision is the standard error of a proportion taken from a sample. It is computed as follows:
where Z, at a 95% confidence level, is 1.96; p is the sample proportion estimate, and n is the sample size. The overall sample size of 800 voting-age adults gives a maximum error margin of ± 3.46% at the 95% confidence level, assuming a simple random sampling design.
The data obtained from the survey were used in the two stages of the research. The first stage employed tables and graphs in order to learn about the Red Shirt and Yellow Shirt supporters. Both supporters are measured here using the color affiliation variable, and using a survey question asking the respondents: “If you had to choose one, which of the following would most closely describe your political opinion?” The responses to this question could be: no color/neutral; Red (UDD); slightly leaning toward Red; slightly leaning toward Yellow; or strongly Yellow (PAD/PDRC). At this stage of the examination, the color affiliation variable was measured as a 3-category variable, where -1 represented the Yellow (those that identified themselves as slightly leaning toward or strongly Yellow), 0 represented those that were neutral (those that identified themselves as having no color or neutral), and 1 represented the Red (those that identified themselves as slightly leaning toward and strongly Red). In order to examine the demographic and socioeconomic differences between the Yellow Shirt and Red Shirt supporters, three socio-demographic indicators, region, rural-urban divide, and wealth, were then employed in order to test the following: (1) whether the Yellow and Red supporters could be regionally differentiated; (2) whether the Yellow supporters consisted of a larger proportion of urban-based Thais than the Red supporters; and (3) whether the Yellow supporters were wealthier than the Red supporters.
First, the regional variable was measured as a 5-category variable, where 1 represented the north, 2 represented the northeast, 3 represented the central region, 4 represented the south, and 5 represented the capital city, Bangkok, respectively. Based on earlier literature considering the Yellow-Red disparities discussed above, the Yellow Shirt supporters were expected to be more numerous than the Red Shirt supporters in the south, while the Red Shirt supporters were expected to be more numerous than the Yellow Shirt supporters in the north and northeast.
Secondly, the indicator for the rural-urban divide was measured as a dichotomous variable, where 0 represented the rural individuals, the respondents that live in a small town or village outside the municipal area, and 1 represented the urbanites, the respondents that live in a large city (in a municipal area) and Bangkok. A large proportion of the Yellow Shirt supporters were expected to be in the urban area, whereas a large proportion of the Red Shirt supporters were expected to be in small towns or villages.
Finally, using respondent’s self-report of their average monthly household income during the past 12 months, wealth was measured as a 4-point scale variable, where 0 was low income (those whose household earns less than 9,000 baht [$287US] a month), 1 was lower middle income (those whose household earns between 9,001 and 20,000 baht [$287-$638US] a month), 2 upper middle income (those whose household earns between 20,001 and 80,000 baht [$2,553US] a month), and 3 high income (those whose household earns more than 80,000 baht [$2553US] a month). The Yellow Shirt supporters tend to be wealthier than their Red Shirt counterparts.
Apart from the three sociodemographic indicators, this paper utilized party identification and democratic values as other indicators in order to examine the Yellow-Red divide in Thai society. The measures of the party identification variable relied on an 11-point scale questions, asking the respondents about the extent to which they usually think that their political opinions are close to the following parties—the Chart Thai Pattana Party, the Democrat Party, the Pheu Thai Party, and the Phumjaithai Party. Each of these questions asked the respondents to rate their degree of affiliation to each political party, ranging from 0 to 10, where 0 was “feel totally not close to this party” and 10 was “feel very close to this party.” According to previous studies regarding the Yellow-Red divide in Thai society, the Yellow Shirt supporters were expected to be more strongly affiliated with the Democrat Party than the Red Shirt supporters, while the Red Shirt supporters tend to be more strongly affiliated with the Pheu Thai Party than the Yellow Shirt supporters.
The second stage of the examination then employed multivariate analysis in order to test whether the Yellow Shirt and Red Shirt supporters differed in demographics and socioeconomic status, political identification, and on democratic values after controlling for several socioeconomic and political influences. Three regression models were constructed, using color affiliation as a dependent variable. In all three models, color affiliation was measured on a Yellow-Red scale, ranging from -2 to 2, where -2 was Yellow, 0 was neutral, and 2 was Red. In order to ascertain how powerfully demographic and socioeconomic variables can explain the differences between the Yellow Shirt and Red Shirt supporters, Model 1 was created by including all of the above three sets of sociodemographic variables. However, regional variables were then measured as four dichotomous variables—north, northeast, central, and south. The rural-urban divide was measured as a 3-point scale variable, where 0 represented the rural individuals, the respondents that live in a small town or village outside the municipal area, 1 represented the urbanites, the respondents that live in a large city (in a municipal area), and 2 represented the respondent that live in Bangkok, whereas the wealth variable was measured in the same way as described above.
Models 2 and 3 were constructed by adding the party identification and democratic values as the main explanatory variables for each model. The main explanatory variables for Model 2 were the four party identification variables, including an 11-point scale of the attachment to the Democrat Party, Pheu Thai Party, Phumjaithai Party, and Chart Thai Pattana Party, in which 0 meant feeling not close to this party and 10 meant feeling very close to this party, with higher values indicating a closer attachment to each political party.
The main explanatory variables for Model 3 were the democratic values, which were measured using the four survey questions, asking whether the respondents strongly agreed (3), agreed (2), disagreed (1), or strongly disagreed (0) with the statements that provided insight into the four democratic principles that are the controversial issues between the parties alienated by conflict —(1) elections, (2) political ethics, (3) rule of law, and (4) redistribution through vote buying. Regarding elections, the respondents were asked how strongly they agreed with the following statement: “Competitive and periodic elections are a necessary and indispensable element of sustained efforts to maintain the right of everyone to take part in the government of their country.” In terms of political ethics, the respondents were asked how strongly they agreed with the following statement: “Thailand needs an ethical leader, regardless of how they get into power, to take the country back from the corrupt politicians.” Regarding rule of law, the respondents were asked how strongly they agreed with the following statement: “When the country is facing a difficult situation, it is okay for the government to disregard the law in order to deal with it.” Further, regarding redistribution, the respondents were asked how strongly they agreed with the following statement: “Elections based on vote-buying by offering money to legitimate voters are a beneficial aspect of Thai politics, because it is a matter of income distribution to Thai people.”
According to previous studies on the Yellow-Red divide discussed above, because one of their movement’s most featured topics is anticorruption, the Yellow Shirt supporters were expected to support political ethics as a key element of democracy to a greater extent than the Red Shirt supporters. In contrast, the Red Shirt supporters, because their movement is driven by a consciousness of inequality, and they believe in elections as the most important right that can provide them with equality, are more likely than the Yellow Shirt supporters to agree that elections, the rule of law, and socioeconomic equality should be the essential elements of democracy.
Before exploring the key differences between the two opposing groups, this paper begins with the proportion of the Thai people that identify themselves as Yellow Shirt or Red Shirt supporters. According to the survey data used in this paper, more than two-thirds of the respondents (67.5%) claim to be completely color neutral, and only 16.1% have a Yellow inclination, and approximately 16.4% have a Red inclination (Figure 1). These findings are slightly different to those presented in a report presented by the Asia Foundation (2011), in which approximately a quarter of the total respondents claimed to have either a Yellow or Red inclination. More precisely, according to the Asia Foundation (2011), those that identified themselves as strong Yellow Shirt supporters represent 5.1% while those that identified themselves as strong Red Shirt supporters represent 6.6%. In addition, around 5.2% identified themselves as leaning toward Yellow Shirt supporters whereas about 7% identified themselves as leaning toward being Red Shirt supporters.
Sociodemographic Differences Between Yellow Shirt and Red Shirt Supporters
The regional differences between the two groups are in line with the expectations. Figure 2 shows that the northern and northeastern regions are, as expected, a Red Shirt stronghold, with a small number of Yellow Shirt supporters. The southern region, in contrast, is a Yellow Shirt stronghold, with very few Red Shirt supporters. The color clash seems to be the most relevant to respondents in the central region and Bangkok, as both groups have supporters in almost similar numbers. Bangkok and its surrounding provinces, in particular Nonthaburi, Samutprakan, and Pathumthani, are places where violent clashes have taken place. Arguably, Bangkok is also the area where people have been most affected by the Yellow-Red conflict in their everyday life.
The rural-urban divide between the Yellow Shirt and Red Shirt supporters is also in line with expectations. As shown in Figure 3, approximately three-fourths of the respondents that identified themselves as Red Shirt supporters are living in a rural area. This proportion is about 20% higher than the proportion of those that identified themselves as Yellow Shirt supporters. In addition and as expected, the proportion of the urban residents that identified themselves as Yellow Shirt supporters is almost double those that identified themselves as Red Shirt supporters. However, because the proportions of both groups in the rural and urban areas are more similar than what might have been expected, this result suggests that what is going on in today’s Thai politics cannot be explained by simply adopting the tale of two democracies thesis, as some previous studies asserted. Rather, this result confirms what recent scholars have claimed, i.e., that people who identify themselves as being close to either the Yellow Shirt or Red Shirt movement are socioeconomically mixed (see e.g., Siamwalla & Jitsuchon, 2012). The difference between these two groups of respondents in terms of their wealth, as shown below, makes this claim more emphatic.
Based on the cash-income indicator used in this paper, the Yellow Shirt supporters tend to have higher income than their Red counterparts, as expected. Figure 4 shows that almost 60% of the respondents that identified themselves as Yellow supporters are those whose household earned monthly income is in the upper-middle-income and high-income categories. In contrast, approximately 70% of the respondents that identified themselves as Red supporters are those whose household earned monthly income is in the low- and lower-middle- income categories.
Differences in Party Identification Between Yellow Shirt and Red Shirt Supporters
According to previous studies regarding the Yellow-Red divide in Thai society, the Yellow Shirt supporters are expected to be more strongly affiliated with the Democrat Party than the Red Shirt supporters, while the Red Shirt supporters tend to be more strongly affiliated with the Pheu Thai Party than their Yellow Shirt counterparts. The results in Table 2 reveal the expected party polarization between the Yellow Shirt and Red Shirt supporters. That is, the respondents who identified themselves as Yellow Shirt supporters are much more likely than the respondents who identified themselves as Red Shirt supporters to report that their political opinions are close to the Democrat Party. In contrast, the respondents who identified themselves as Red Shirt supporters are much more likely than the respondents who identified themselves as Yellow Shirt supporters to report that their political opinions are close to the Pheu Thai Party. The table also presents a very small difference in party identification among the respondents that identified themselves as having no color or being neutral, as the mean scores for how close these respondents think that their political opinions are to all four political parties ranged between 2.11 and 2.80.
The Key Drivers of Political Polarization between the Yellow-shirt and Red Shirt Supporters
Model 1 in Table 3 shows the associations among six demographic and socioeconomic variables and color affiliation. Northern and northeastern residence were significantly and positively (i.e., more Red) related to color affiliation, while southern and urban residence, and wealth, were significantly but negatively (i.e., more Yellow) related to color affiliation. According to the standardized regression coefficients, among these variables, southern residence had the largest impact on color affiliation, followed by northern and northeastern residence. However, the R2 was only 0.362, indicating that only a moderate part of the variation in color affiliation was accounted for by the variables considered in this model.
When the four variables of party identification were added to the regression analysis of Model 2, the R2 sharply improved. Because color affiliation was measured on a Yellow-Red scale, where the minus values represented the Yellow Shirt and the plus values represented the Red Shirt, the attachment to the Democrat Party was significantly and negatively related to color affiliation (more Yellow), whereas the attachment to the Pheu Thai Party was significantly but positively (more Red) associated with color affiliation, as expected. Only three of the six sociodemographic backgrounds—southern residence, urban residence, and wealth—remained significant with the same direction of association as color affiliation. However, based on the standardized regression coefficients, southern residence lost its position as having the most significant impact on color affiliation and was replaced by attachment to the Democrat Party. Although adding party identification to the regression model eliminated the significance of only two sociodemographic variables (i.e., north residence and northeast residence), this finding indicated that party identification is the most powerful factor driving the Yellow-Red divide in Thai society.
When the four variables related to democratic values were added to the regression analysis of Model 3, the R2 slightly improved. Three of these democratic values were significantly related to color affiliation. As expected, support for elections and equality as an essential element of democracy had a positive impact on color affiliation (more Red), while political ethics had a negative effect in relation to color affiliation (more Yellow). As with the results shown in the Model 2, attachment to the Democrat Party maintained the most significant impact on color affiliation, followed by attachment to the Pheu Thai Party, based on the standardized regression coefficients. This finding indicated that adding democratic values to the regression model did not eliminate the significance of any of the sociodemographic variables but barely reduced the strength of the impact of all of them and one of the party-identification variables (i.e., attachment to the Pheu Thai Party).
Political polarization has been a phenomenon in modern democratic societies where political conflicts between two opposing groups of people are present. In Thailand, the phenomenon that emerged decades ago is known as “color politics,” a political contest between Yellow Shirt and Red Shirt supporters. The analyses carried out in this paper showed evidence mostly confirming what previous studies have described about the differences in sociodemographic backgrounds, party identification, and political attitudes between Yellow Shirt and Red Shirt supporters. However, this paper found that color-based polarization in Thailand is more difficult to understand than has been explained in the existing literature.
In terms of sociodemographic differences, the color-based political polarization in Thailand could not be explained as an absolute division between the urban rich and the rural poor because there are both urban and rural residents as well as the rich and poor support both “color camps.” Rather, the better way to explain this phenomenon is that the political polarization between the Yellow Shirt and Red Shirt supporters in Thailand is a division between one political camp where the majority of its supporters are the urban and richer residents, and another political camp where the majority of its supporters are rural residents and poorer people.
In addition, the findings regarding the differences in party identification and democratic values between the Yellow Shirt and Red Shirt supporters shown in this paper indicate that the struggle between the Yellow Shirt and Red Shirt supporters is not just about competing interests but is about the party identifications and basic values underlying the rules of “the political game.” Moreover, among these three groups of factors, party identification was seen to be the most powerful factor driving the political polarization between the Yellow Shirt and Red Shirt supporters, followed by the democratic-values factor. This result reflects what has happened in Thailand, where political polarization often seems extreme and could damage democracy. However, it is interesting that a majority of the respondents indicated no strong color preference, and that this evident indifference stands in contrast to the political disruption the color conflict has caused in the country. It would therefore be desirable for future research concerning theoretical frameworks to attempt to explain why a political division that does not seem to be acutely felt by the majority of the population has been able to dominate political events for nearly two decades.
In fact, Sondhi Limthongkul, a media mogul and one of the PAD leaders, had already been promoting anti-Thaksin campaigning for several months, largely under the Thailand Weekly talk show label, and his supporters had already informally adopted yellow, the color of Monday (the day of King Rama IX’s birth), as their color. In November 2013, this group newly formed the People’s Democratic Reform Committee (PDRC), a coalition of the Democrat Party (DP), the PAD, and pro-military groups.
While the UDD is the dominant Red Shirt group, the movement includes other groups only loosely affiliated and not subordinate to the UDD.
Although 11 parties had members elected in the 2011 general election, only these four political parties won more than 10 seats.