2.1 Confucianism in Asia

Confucianism presents two contrasting faces in academic research. While many scholars perceive Confucianism as an inflexible official doctrine that impedes personal freedom and political equality (Chan, Reference Chan2007; Huntington, Reference Huntington1991; Li, Reference Li1997), proponents of neo-Confucianism argue that it embodies numerous beneficial elements essential for liberal democracy, such as duty and civil virtue (Herr, Reference Herr2010; Shils, Reference Shils and Tu1996). Consequently, the visage of Confucianism appears somewhat ambiguous, sparking intense debates. We attempt to minimally define Confucianism from three aspects: non-individualistic, moral hierarchy, and harmonious order.

First, Confucianism is a doctrine of collectivism or communitarianism rather than atomic individualism (Fei et al., Reference Fei1992; Rosemont, Reference Rosemont2015; Bai, Reference Bai2019), with the family serving as the smallest unit of the social network (Fei et al., Reference Fei1992; Zhai, Reference Zhai2017, Reference Zhai2018, Reference Zhai2022; Liang, Reference Liang2021; Park and Shin, Reference Park and Shin2006). Interpersonal relationships, such as those between father and son, husband and wife, and ruler and people, constitute the central focus of Confucianism. Each individual is a node within the interconnected network of relationships, and it is imperative that they revere these connections without separation or betrayal (Fei et al., Reference Fei1992; Rosemont, Reference Rosemont2015; Liang, Reference Liang2021; Yao, Reference Yao2000; Bai, Reference Bai2019).

Second, Confucianism is committed to a moral hierarchy, which promotes ethical guidelines for various interpersonal relationships ranging from family dynamics to political governance (Fei et al., Reference Fei1992; Park and Shin, Reference Park and Shin2006; Zhai, Reference Zhai2017, Reference Zhai2018, Reference Zhai2022; Liang, Reference Liang2021; Yao, Reference Yao2000), emphasising the importance of adherence to these ethical norms (Fei et al., Reference Fei1992; Park and Shin, Reference Park and Shin2006; Rosemont, Reference Rosemont2015; Liang, Reference Liang2021; Yao, Reference Yao2000). The partner with higher social standing within the relationship is attributed with greater moral authority, often resulting in an expectation for the lower-status partner to demonstrate respect.

Third, Confucianism advocates harmonious order. According to Confucian doctrine, harmony is the fundamental essence of the world, as opposed to chaos or conflict (Fei et al., Reference Fei1992; Park and Shin, Reference Park and Shin2006; Zhai, Reference Zhai2017, Reference Zhai2018, Reference Zhai2022; Liang, Reference Liang2021). Confucianism believes that unrest arises when ethical rules are broken. Therefore, the ideal society in Confucianism is one characterised by orderly adherence to ethical norms (Fei et al., Reference Fei1992; Yao, Reference Yao2000; Li, Reference Li2006, Reference Li2013; Wei and Li, Reference Wei and Li2013; Liang, Reference Liang2021).Footnote ²

2.2 The ‘Asian values’ debate

The relationship between Confucianism and democracy remains highly contentious, known as the ‘Asian values’ debate (Kim, Reference Kim1994; Sen, Reference Sen1997, Reference Sen1999; Thompson, Reference Thompson2001; Lu and Liu, Reference Lu and Liu2022). According to incompatibility theory, Confucianism has been treated as an obstacle to the adoption of democracy in Asia because it emphasises obedience and obligations are inconsistent with the individualistic basis of democratic norms (Hu, Reference Hu1997; Huntington, Reference Huntington1991, Reference Huntington1996; Pye, Reference Pye1968, Reference Pye and Pye1985). Massive evidence has accumulated to indicate that Confucian values inhibit democratic values (Chang and Chu, Reference Chang and Chu2002; Chang et al., Reference Chang, Chu and Tsai2005; Chu et al., Reference Chu, Diamond, Nathan, Shin, Chu, Diamond, Nathan and Shin2008; Nathan, Reference Nathan2007; Park and Shin, Reference Park and Shin2006). On the other hand, the compatibility theory has noted that Confucianism is a modification of liberal democracy (Ackerly, Reference Ackerly2005; Flanagan and Lee, Reference Flanagan and Lee2000; Inglehart and Welzel, Reference Inglehart and Welzel2005). Recent literature has observed that adherence to Confucian values reinforces support for democracy rather than diminishing it (Dalton and Ong, Reference Dalton and Ong2005; Fetzer and Soper, Reference Fetzer and Soper2007, Reference Fetzer and Soper2010). Democratic support does not have to generate tension with orientations toward authority associated with Confucian values, offering a more optimistic outlook on the prospects for political development in Asia. Overall, the ‘Asian values’ debate constitutes a facet of the struggle between traditional values and democratic attitudes.

3.1 Support for democracy: the MSFD and the PSFD

The longevity of a democratic system relies on sustained backing from its citizens. Political scientists explore diverse factors to understand individuals’ endorsement of democracy, encompassing economic inequality (Andersen, Reference Andersen2012), government effectiveness (Magalhães, Reference Magalhães2014), and performance assessments (Camacho, Reference Camacho2019). However, none of these factors adequately explain the paradox of democratic support. We argue that the conceptual ambiguity and measurement uncertainty within current indices assessing democratic support create a disconnect between operationalisation methods and theoretical goals. The most commonly used operationalisation method (the MSFD) in surveys explicitly asks respondents whether they prefer a democratic system or consider it appropriate. Evidently, this approach contains significant measurement errors (e.g., social desirability bias) and fails to capture genuine supporters among the general public. Respondents may be reluctant to express their true opinions due to apprehensions regarding government censorship (Chu et al., Reference Chu, Huang and Lu2015; Shi, Reference Shi, Chu, Diamond, Nathan and Shin2008). As a consequence, the public’s explicit support for democracy sometimes primarily reflects their desire to project a certain image to others (Chu and Huang, Reference Chu and Huang2010). Affirmative responses only obtain their vague attitudes toward the D-word, providing little insight into the extent to which people are willing to sacrifice for democratic ideals.

This study conceptualises public support for democracy as the MSFD and the PSFD, with the latter meriting further exploration. All supporters of democracy have the MSFD, meaning that whatever understanding of democracy they adopt, they agree that ‘democracy is good’. At best, the MSFD indicates that respondents possess a positive recollection or impression of the D-word. The PSFD involves a trade-off between democracy and economic development, signifying a refusal to abandon intrinsic and procedural values in favour of the instrumental value of democracy. As a robust predictor of loyal democratic support, the PSFD is fundamental to both the conceptualisation and measurement of substantive support for democracy (Chang et al., Reference Chang, Chu and Huang2006; Lu and Chu, Reference Lu and Chu2021). In summary, this approach aims to delineate supporters of democracy with diverse perceptions by artificially juxtaposing the instrumental and other values of democracy, thereby highlighting their distinctions. Researchers can ascertain whether individuals are willing to pay the opportunity cost of democracy by examining the trade-off between democracy and economic development. It is much more meaningful to explore the question of whether ‘democracy is worthwhile’ under the circumstance that nearly all citizens agree that ‘democracy is good’.

3.2 Understanding of democracy: instrumental and others

Before exploring the cultural origins of the MSFD and the PSFD, it is necessary to recognise the existing disparity in democratic cognition among respondents, with variations observed across different countries (Chu et al., Reference Chu, Diamond, Nathan, Shin, Chu, Diamond, Nathan and Shin2008; Chu et al., Reference Chu, Huang and Lu2015). Many scholars contend that diverse understandings of democracy influence people’s attitudes toward it (Chapman et al., Reference Chapman, Hanson, Dzutsati and DeBell2024). Hence, it is imperative to incorporate cultural norms into research on democratic support among the populace.

The concept of democracy encompasses three distinct values:

1. (1) The intrinsic value of democracy emphasises its fundamental principles – accountability, civil liberties, and a series of political rights of citizens, such as freedom of speech and freedom of belief (Kim, Reference Kim2017; Dalton et al., Reference Dalton, Sin and Jou2007; Diamond, Reference Diamond1999; Bratton and Mattes, Reference Bratton and Mattes2001);

2. (2) The procedural value of democracy pertains to the methods that ensure fair and equal participation in decision-making (Chu et al., Reference Chu, Huang and Lu2015; Dalton et al., Reference Dalton, Sin and Jou2007 ; Kim, Reference Kim2017). Scholars such as Schumpeter (1976 [Reference Schumpeter1943]), Sartori (Reference Sartori1987), and Dahl (Reference Dahl1956, Reference Dahl1989) define democracy from an institutional perspective, emphasising that the unpredictability brought about by competitive elections is an essential feature of a democratic system;

3. (3) The instrumental value of democracy focuses more on the performance of democratic governance, such as prosperity, welfare, and integrity (Chu et al., Reference Chu, Huang and Lu2015; Dalton et al., Reference Dalton, Sin and Jou2007; Kim, Reference Kim2017; Bratton and Mattes, Reference Bratton and Mattes2001). It stems from instrumental rationality, which posits that democratic value lies in its outputs, especially its potential to foster economic development.

Supporters’ understanding of democracy can thus be divided into two categoriesFootnote ³ : instrumental understanding and non-instrumental understanding. People with an instrumental understanding perceive democracy as a means to pursue effective governance outcomes. The instrumental understanding of democracy necessitates that support for democracy be contingent upon the performance it delivers. Given the inherently volatile nature of governance performance, an instrumental understanding of democracy often reflects pragmatic support based on a balance of interests. They are not staunch supporters but rather opportunists who may abandon democratic ideals if autocracy can deliver good governance. Conversely, those with a non-instrumental understanding emphasise the procedural value (institutions) and intrinsic value (principles) of democracy as paramount.

3.3 Hypothesis

This paper aims to investigate the intricate relationship between Confucian values, democratic understanding, and support for democracy. Confucianism exerts an essential role in shaping citizens’ level of support for democracy (Shin, Reference Shin2012; Hu, Reference Hu2021; Lu and Chu, Reference Lu and Chu2021, Reference Lu and Chu2022; Ma and Huang, Reference Ma and Huang2021). According to Confucian doctrines, the principle of minben emphasises rulers treating ‘people’ (min) as the ‘root’ (ben), which can be understood as a concept emphasising the centrality of the people in governance (Shin, Reference Shin2012). It highlights the enduring relevance of placing people’s interests at the heart of political and administrative decision-making. Mencius, regarded as one of the sources of Confucianism, said, ‘The people are of greatest importance, the altars of the soil and grain are next, and the ruler is of least importance’ (Mencius, 2009). He believed that rulers should govern for the prosperity and welfare of the people due to Tianming Footnote ⁴ , just as a tree needs a solid root (Nuyen, Reference Nuyen2000; Shin, Reference Shin2012; Spina et al., Reference Spina, Shin and Cha2011; Murthy, Reference Murthy2000; Xu, Reference Xu2006). Heaven serves as the supervisor of the ruler, mandating legitimacy upon the ruler – ‘Heaven sees as its people see; Heaven listens as its people listen’ (Kern and Meyer, Reference Kern and Meyer2017, p. 311). If the ruler fails to treat the people well, Heaven will withdraw his mandate, leading to rebellion among the people (Spina et al., Reference Spina, Shin and Cha2011; Xu, Reference Xu2006).Footnote ⁵

In the absence of a distinction between intrinsic value, procedural value, and instrumental value, there is no contradiction between the Confucian minben principles and the democratic spirit of ‘of the people, by the people, and for the people’ (Spina et al., Reference Spina, Shin and Cha2011). Within the Confucian context, there is a tendency for Confucianism and democracy to rhetorically intertwine, with proponents of Confucianism also embracing democracy as a commendable value. The theoretical discussion leads to Hypothesis 1.

However, when faced with the trade-off between democracy and economic development, the potential adverse consequences of democratic principles and institutions become amplified. A democracy stripped of its instrumental value struggles to align with Confucianism. In fact, Confucian values emphasise harmony over chaos, obedience over freedom, and obligations over rights and discourage actions that undermine the authority structure, thereby limiting public participation or dissent (Bell and Ham, Reference Bell and Ham2003; He, Reference He2010; Shin, Reference Shin2012). Clearly, these values are in tension with the democratic system, as competitive elections seek to safeguard civil liberties and political rights by institutionalising conflict. Confucianism holds that the well-being of its inhabitants serves as the primary foundation for governmental legitimacy and national stability (Chan, Reference Chan2007; Lu and Shi, Reference Lu and Shi2015; Shi and Lu, Reference Shi and Lu2010), which drives East Asian citizens to prioritise economic development over democratic ideals.

Confucianism assesses political systems by their capacity to deliver socioeconomic benefits. When democratic principles and institutions fail to yield socio-economic benefits or when alternative political systems can guarantee such interests, their appeal diminishes. As a consequence, we posit that those with an instrumental perspective on democracy are prone to misunderstanding its fundamental value, blurring the distinction between democracy and authoritarianism, thereby providing inconsistent support for democracy. By discerning the diverse comprehension of democracy among people, we can elucidate the crisis faced by democratic systems and unravel counterintuitive phenomena observed in existing surveys. On the basis of the theoretical discussion, we propose Hypothesis 2 and Hypothesis 3.

The three research hypotheses utilised in this paper are succinctly summarised in Figure 1 Effects of ConfucianValues, which visually depicts the relationship between Confucian values, democratic understanding, and support for democracy.

Figure 1. Effects of Confucian values on support for democracy.

Note: Solid line refers to direct effects; dashed lines refer to indirect effects. The sign on the lines indicates the relationship hypothesised in this study.

4.1 Data and sample

We test our theory using data from the latest (fifth) Asian Barometer Survey (2018–2022), which utilises a probability proportional to population sampling design. Table A1 in the appendix presents questions and corresponding codes that we utilised in the ABS. We focus on seven Asian societies, both democratic and authoritarian, namely Japan, South Korea, Mainland China, Hong Kong, Taiwan, Singapore, and Vietnam. Their shared Confucian traditions have been recognised by scholars as forming the Confucian cultural sphere (Lu and Chu, Reference Lu and Chu2022; Shin, Reference Shin2012).

The dataset comprises 11,915 valid respondents. To address missing values, we adopt random forest imputation (RFI) proposed by Tang and Ishwaran (Reference Tang and Ishwaran2017). The RFI is founded upon machine learning principles, showcasing exceptional computational efficiency and prediction accuracy. Additionally, it exhibits versatility in handling diverse types of data, including continuous and discrete variables, as well as high-dimensional datasets. The K-fold cross-validation results demonstrate that the RFI model exhibits superior performance. A comparison of the K-fold cross-validation outcomes is reported in Figure A1 in the appendix.

4.2 Dependent variable: support for democracy

We conducted regression analyses on two types of democratic support: the MSFD and the PSFD. The MSFD refers to respondents’ favourable assessment of democratic ideals and their agreement with the notion that ‘democracy is good’. We measure the MSFD through respondents’ answers to the following two questions on democratic suitability and democratic preferability:Footnote ⁶

• Democratic suitability: Is democracy suitable or not? (1–10)

• Democratic preferability: Do you agree or disagree that ‘Democracy may have its problems, but it is still the best form of government’. (1–4)

The PSFD aims to assess citizens’ engagement in profound contemplation regarding the essence of the democratic system. We specifically examine the trade-off individuals face between democracy as an intrinsic and procedural value and economic development as an instrumental value, which reflects their level of endorsement for the notion that ‘democracy is worthwhile’. We measure the PSFD based on respondents’ answers to the following questions on democratic priority and democratic necessity:

• Democratic priority: If you had to choose between democracy and economic development, which would you say is more important? (1–5)

• Democratic necessity: As long as a government can solve our country’s economic problem, it does not matter if it is democratic or not democratic. (1–4)

4.3 Independent variable: Confucian values

While Confucianism adheres to a common ethical framework, its specific manifestations vary across different dimensions (Fei et al., Reference Fei1992). According to the minimal definition of Confucianism outlined in the theoretical section, we mainly categorise it into three distinct dimensions: family, society, and politics. Concretely, the non-individualistic Confucian family values emphasise solidarity, loyalty, respect for parents, and the interests of the family as primary values; Confucian social values advocate for social cohesion, respect for higher-status partners, maintaining harmony, and avoiding disorder; and Confucian political values, with an inherent orientation toward authority, uphold respect for authoritarian structure and the supremacy of national interests (Chang and Chu, Reference Chang and Chu2002; Chang et al., Reference Chang, Chu and Tsai2005; Zhai, Reference Zhai2017, Reference Zhai2018, Reference Zhai2022). Given the multifaceted nature of Confucian values, distinct dimensions of Confucianism may exert varying influences on democratic support (Zhai, Reference Zhai2017, Reference Zhai2018, Reference Zhai2022).

A series of questions representing the three dimensions was combined into a single independent variable for each dimension through principal component analysis (PCA). The results of the KMO test and Bartlett’s test can be found in Table A2 in the appendix. We assess respondents’ level of agreement with Confucian values based on their alignment with the following perspectives:Footnote ⁷

• A: For the sake of the family, the individual should put his personal interests second.

• B: Even if parents’ demands are unreasonable, children still should do what they ask.

• C: In a group, we should sacrifice our individual interests for the sake of the group’s collective interest.

• D: Being a student, one should not question the authority of their teacher.

• E: In a group, we should avoid open quarrels to preserve the harmony of the group.

• F: Even if there is some disagreement with others, one should avoid the conflict.

• G: A person should not insist on his own opinion if his co-workers disagree with him.

• H: Harmony of the community will be disrupted if people organise lots of groups.

• I: If people have too many different ways of thinking, society will be chaotic.

• J: For the sake of national interest, individual interest could be sacrificed.

• K: Government leaders are like the head of a family; we should all follow their decisions.

• L: The government should decide whether certain ideas should be allowed to be discussed in society.

• M: If we have political leaders who are morally upright, we can let them decide everything.

Based on the scores of questions A and B, we have constructed an index for Confucian family values, an index for Confucian social values derived from questions C to I, and an index for Confucian political values derived from questions J to M. It is worth noting that although the operationalisation of Confucian values draws on extensive literature based on the ABS dataset (Chang and Chu, Reference Chang and Chu2002; Chang et al., Reference Chang, Chu and Tsai2005; Fetzer and Soper, Reference Fetzer and Soper2007; Park and Shin, Reference Park and Shin2006; Zhai, Reference Zhai2017, Reference Zhai2018, Reference Zhai2022; Shin, Reference Shin2012; Kim, Reference Kim2010; Chang et al., Reference Chang, Wu and Weatherall2017), certain limitations remain. Due to the limited scope of inquiry within the ABS, it is challenging to comprehensively measure Confucian values, including aspects such as equality and tolerance. Nevertheless, we still consider the operationalisation in this study to be valid, as it avoids overgeneralising Confucian values to the point of ambiguity (the omitted aspects themselves are subject to debate). Therefore, the selected measurement indicators are well-designed to capture the primary characteristics of Confucian values.

4.4 Mediator: understanding of democracy

As discussed in the theory section, we suppose that Confucianism facilitates an instrumental understanding of democracy, which serves as a mediator between Confucian values and support for democracy. We examined whether individuals exhibited an instrumental understanding based on their responses to the following question:

• If you have to choose only one, which of the following do you think is the most essential element of a democracy?

• a: People are free to express their political views openly.

• b: Basic necessities, like food, clothes, and shelter, are provided for all.

• c: People choose the government leaders in free and fair election.

• d: Politics is clean and free of corruption.

Answers b and d exemplify the instrumental value of democracy, which pertains to the focus on democratic governance outcomes such as basic necessities and clean politics. Conversely, answers a and c, respectively, imply the intrinsic value and procedural value of democracy, emphasising democratic principles (freedom of expression) and the process (competitive election procedure) by which they are realised. If respondents choose answers b or d for this question, a dummy variable of instrumental understanding will be constructed and assigned a value of 1. Conversely, if respondents select answer a or c, the mediating variable will be assigned a value of 0.

4.5 Covariates

The control variables account for an array of subjective and objective indicators that have the potential to exert an influence on an individual’s cultural values and political attitudes (Butler and De La O, Reference Butler and De La O2011; Ceka and Magalhaes, Reference Ceka and Magalhaes2020; Cho, Reference Cho2014; Albacete, Reference Albacete2010; Lago, Reference Lago2022; Lu and Chu, Reference Lu and Chu2022; Osborne et al., Reference Osborne, Yogeeswaran and Sibley2015; Placek, Reference Placek2020; Putnam et al., Reference Putnam, Leonardi and Nonetti1993; Robbins and Tessler, Reference Robbins and Tessler2012; Shin, Reference Shin2012; Weinschenk et al., Reference Weinschenk, Dawes, Oskarsson, Klemmensen and Nørgaard2021). The subjective indicators include respondents’ assessment of the economic status of their country and family (country economic evaluation and family economic evaluation), their awareness of their family’s social status (family status), their interest in politics (political interest), their sense of political efficacy (political efficacy), and their degree of social trust (social capital). The objective indicators are mainly individual indicators of respondents, including gender, age, education, employment, household income, internet use, residence, and election participation. These confounding factors need to be controlled to mitigate omitted variable bias. Summary statistics of all variables are provided in Table A3 in the appendix.

4.6 Estimation method

The effects of Confucian values on the MSFD and the PSFD are estimated using the least squares dummy variable (LSDV) approach. For cross-sectional data with continuous dependent variables, LSDV estimation is able to capture the impact of group differences by controlling for country and year fixed effects, which account for time-invariant country-specific characteristics and year-specific factors shared by all countries.

Considering the greater theoretical interest in the PSFD in this study, we also seek to investigate the underlying mechanisms through which Confucian values influence the trade-off between democracy and economic development. We present mediation analysis in two steps. We first adopt a stepwise regression (Baron and Kenny, Reference Baron and Kenny1986) to estimate the mediating effect of democratic understanding. However, the mediator, being a binary variable, may not be suitable for this traditional method (Hicks and Tingley, Reference Hicks and Tingley2011). Instead, we employ a two-step mediation analysis. Subsequently, we apply the causal mediation analysis (CMA), Sobel test, and bootstrap test to further estimate the average mediation effect (AME) and the average direct effect (ADE). Compared to the traditional two-step method, CMA better controls for confounding factors and reveals complex causal mechanisms (Hicks and Tingley, Reference Hicks and Tingley2011; Imai et al., Reference Imai, Keele and Tingley2010; Imai et al., Reference Imai, Keele, Tingley and Yamamoto2011; Celli, Reference Celli2022).

The equation used in the main analysis is as follows:

(1) $$MSF{D_{ijt}} = \alpha Value{s_{ijt}} + \beta {X_{ijt}} + {\mu _j} + {\varphi _t} + {\varepsilon _{ijt}}$$

(2) $$PSF{D_{ijt}} = \alpha Value{s_{ijt}} + \beta {X_{ijt}} + {\mu _j} + {\varphi _t} + {\varepsilon _{ijt}}$$

(3) $$PSF{D_{ijt}} = \alpha Value{s_{ijt}} + \theta Understandin{g_{ijt}} + \beta {X_{it}} + {\mu _j} + {\varphi _t} + {\varepsilon _{ijt}}$$

where i represents the interviewed individual, j denotes the respondent’s country, and t signifies the survey year. MSFD _ijt and PSFD _ijt measure two forms of democratic support expressed by respondents. Values _ijt reflects the extent to which respondents hold Confucian values. Understanding _ijt is a dummy variable that measures whether people’s understanding of democracy is instrumental. X _ijt represents a set of observed covariates. The equation also includes country-fixed effects μ _j and year-fixed effects φ _t, capturing unobserved heterogeneity. ε _ijt represents the error term.

Equations (1) and (2) serve as baseline models for testing the first two hypotheses. We anticipate a positive (or insignificant) coefficient $\alpha $ for Equation (1), while a negative coefficient $\alpha $ for Equation (2). Equation (3) is the outcome model with the mediator included, designed to test the final hypothesis. We preliminarily estimate the mediating effect by comparing the magnitude of the coefficient $\alpha $ of Equations (2) and (3). We expect that the coefficient $\alpha $ in Equation (3) to also be negative, but with an absolute value smaller than that in Equation (2).

5.1 Baseline model

The results of the baseline regression, as presented in Tables 1 and 2, demonstrate the association between distinct dimensions of Confucian values and various indicators of support for democracy. In general, the three dimensions of Confucian values do not exhibit a significant negative correlation with the MSFD. Even Confucian family values are positively correlated with democratic suitability and democratic preferability. It provides empirical evidence supporting Hypothesis 1, which posits that Confucian ideas promote the MSFD. When the dependent variable shifts to democratic priority and democratic necessity, a reversal of the signs of coefficients is observed. There is a significant negative association between Confucian values and the PSFD, indicating that individuals with stronger adherence to Confucian beliefs are more inclined to prioritise economic development over democracy. This result verifies Hypothesis 2, that is, Confucian values undermine the PSFD. These results remain robust after controlling for fixed effects and relevant covariates.

Table 1. Regression results of Confucian values and minimum support for democracy (MSFD)

Note: Models (1) – (3) use democratic suitability as the dependent variable, while Models (4) – (6) use democratic preferability as the dependent variable. Columns 1 and 4 include only the core independent variables, columns 2 and 5 add fixed effects, and columns 3 and 6 incorporate control variables. The estimator is OLS with fixed effects. The unit of analysis is the individual. Robust standard errors in parentheses. *P < 0.1, **P < 0.05, ***P < 0.01.

Table 2. Regression results of Confucian values and the priority support for democracy (PSFD)

Note: Models (7) – (9) use democratic priority as the dependent variable, while Models (10) – (12) use democratic necessity as the dependent variable. Columns 1 and 4 include only the core independent variables, columns 2 and 5 add fixed effects, and columns 3 and 6 incorporate control variables. The estimator is OLS with fixed effects. The unit of analysis is the individual. Robust standard errors in parentheses. *P < 0.1, **P < 0.05, ***P < 0.01.

The baseline regression results align with our theoretical hypothesis and present an intriguing contrast. They suggest that democratic ideals appear appealing, but material well-being is more desirable.

5.2 Robustness tests

Additional tests were performed to estimate the robustness of the main findings, which can be found in Appendix B. The first robustness test altered the calculation method of independent variables. Three Confucian value indicators were constructed by calculating the mean scores of respondents for each dimension question instead of using PCA. The second test changed the sample scope, focusing on five narrowly defined Confucian societies: Japan, South Korea, Mainland China, Hong Kong, and Taiwan, while the respondents from Singapore and Vietnam were excluded. The third test narrowed the sample based on the survey period. Given that all seven Confucian societies were affected by the COVID-19 pandemic from 2020 to 2023, it is plausible that this crisis may influence respondents’ preferred order of various political values. Under the threat of a pandemic, individuals might prioritise evaluating government effectiveness over assessing democratic systems. Hence, to isolate the potential impacts stemming from COVID-19, only samples surveyed in 2018 and 2019 were selected for regression analysis. Figures A2, A3, and A4 provide 95% confidence intervals for the core independent variables in the three robustness tests.

The fourth and fifth robustness tests used propensity score matching (PSM) and entropy balancing matching (EBM) before regression to correct sample selection bias and alleviate the endogeneity problems in observational studies (Dehejia and Wahba, Reference Dehejia and Wahba2002; Ho et al., Reference Ho, Imai, King and Stuart2007). Building upon the relevant research, we selected gender, age, education (Shin, Reference Shin2012), residence (Lago, Reference Lago2022), employment, family status, household income (Ceka and Magalhaes, Reference Ceka and Magalhaes2020), election participation (Robbins and Tessler, Reference Robbins and Tessler2012), and internet use (Cho, Reference Cho2014; Placek, Reference Placek2020) as covariates to match individuals. The regression results after PSM are presented in Table A5, and the results of balance diagnostics are provided in Figure A5 and Table A4. Table A6 displays the regression results after EBM. Meanwhile, the sixth robustness test turns to the data from the seventh wave of the WVS (2017–2021). We retain only a sample of the seven Confucian societies and construct variables that align with the baseline models. Table A7 provides the detailed questions and codes. The regression results based on the WVS can be found in Table A8 in Appendix B.

The results of the six robustness tests are consistent with the baseline models and our arguments, establishing the reliability of the relationship between Confucian values and democratic support in this study.

5.3 Mediation analysis

The stepwise regression results are reported in Table 3. Baseline models reveal a significant negative correlation between Confucian political-social values and the PSFD. Nevertheless, upon the inclusion of an instrumental understanding of democracy, there is a decrease in absolute coefficients for both social and political values. The coefficient of the mediating variable is significant, and the sign is negative, as is the coefficient of social and political values. This suggests that the instrumental understanding plays a mediating effect between Confucianism and the PSFD.

Table 3. Regression results of outcome and baseline models

Note: Models (1) and (2) use democratic priority as the dependent variable, while Models (3) and (4) use democratic necessity as the dependent variable. Columns 2 and 4 include only the core independent variables, and columns 1 and 3 add the mediating variable. All models include country and year fixed effects. The estimator is OLS with fixed effects. The unit of analysis is the individual. Robust standard errors in parentheses. *P < 0.1, **P < 0.05, ***P < 0.01.

The CMA results presented in Table 4 validate the significance level of the mediating role played by an instrumental understanding of democracy, demonstrating that AME, ADE, and total effect for each model have 95% confidence intervals that do not encompass zero. This indicates that the instrumental understanding significantly mediates the relationship between Confucianism and the PSFD. To validate the mediation analysis, we also conducted the Sobel test and bootstrap test, as shown in Table A9. Both tests yield significant results at a level of 1%, further confirming the strong robustness of the proposed mediation mechanism. Therefore, Hypothesis 3 is corroborated.

Table 4. The results of the causal mediation analysis (CMA)

Note: The estimation method is the causal mediation analysis (CMA). The independent variables are Confucian political-social values, the dependent variable is priority support for democracy (PSFD), and the mediator is an instrumental understanding of democracy. AME refers to the average mediation effect, and ADE refers to the average direct effect. They constitute the total effect.

It can thus be concluded that proponents of Confucianism possess a pragmatic perspective on democracy, recognising its significance primarily in terms of its instrumental value rather than intrinsic or procedural value. The contention is that democracy should be endorsed as a mechanism for achieving effective governance outcomes such as prosperity and welfare. However, when democratic ideals clash with material well-being or interests that can also be achieved through other regimes, the proponents of Confucianism may no longer consider democracy necessary.