President Xi Jinping's 习近平 daring anti-corruption campaign, particularly its cooling effect on the country's economy, has rekindled the debate about the resilience of the authoritarian regime.Footnote 1 Echoing Gordon Chang's decade-old prediction,Footnote 2 David Shambaugh argues that “the endgame of communist rule has now begun.” Among other omens of a looming crisis, he notes that “[L]arge numbers of citizens and party members alike are already voting with their feet and leaving the country or displaying their insincerity by pretending to comply with party dictates.”Footnote 3 Referring also to public confidence, Dingding Chen contends that the regime is not going to collapse because Xi's anti-corruption drive “has the public's support.”Footnote 4 A longstanding question re-emerges. How much confidence do ordinary Chinese citizens have in the central leadership?
Assessing popular trust in the central leadership or government is anything but straightforward. In fact, one of the most puzzling results of national surveys conducted since 2000 is that over 80 per cent of Chinese citizens consistently show strong or moderate trust in the central government or central leaders. The 2000 wave of the World Values Survey, for instance, shows that over 95 per cent of respondents from China trust the central government.Footnote 5 Similarly, the 2008 China Survey shows that nearly 85 per cent of respondents trust central government leaders. Sceptics dismiss such findings on the grounds that self-reported political trust is merely a “response to social pressures and political control” under one-party authoritarian rule.Footnote 6 Survey researchers, however, have found little evidence that political caution seriously compromises the quality of data on sensitive issues.Footnote 7
Drawing on five national surveys, this paper reassesses popular trust in the central government. It first describes the phenomenon of hierarchical trust, i.e. having more trust in the central government than in local government. Then it shows that, compared to those who express equal trust in central and local governments, individuals who hold hierarchical trust have significantly stronger latent doubts about the central government. It argues that the fact that between one- and two-thirds of respondents hold hierarchical trust indicates that trust in the central government is significantly weaker than it looks. After a brief analysis of the sources and significance of hierarchical trust, the paper concludes with a discussion about how to use an integrated approach to triangulate trust in the central government.
The Phenomenon of Hierarchical Trust
Political trust is commonly defined as citizens’ belief that the political system, government and politicians will work to produce outcomes consistent with their expectations.Footnote 8 Despite its simple definition, however, political trust is hard to measure in any society. Above all, it has multiple targets. As regards political institutions, citizens may have different levels of confidence in the legislature, the executive and the judiciary.Footnote 9 Second, the substance of trust in a given target has multiple dimensions. For instance, people may have different levels of confidence about a politician's commitment, capacity or integrity.Footnote 10 Furthermore, public confidence regarding a single aspect of a political leader – for example, his competence – may vary according to different domains, such as domestic affairs and foreign policy.Footnote 11 Lastly, yet another complication emerges when the object of trust is a multilevel government,Footnote 12 as people may have (1) equal trust in all levels; (2) equal distrust in all levels; (3) stronger trust in local authority than in the national government, which is known as “paradox of distance”;Footnote 13 and (4) stronger confidence in the central government than in local government, which can be called hierarchical trust.
On top of these conceptual complexities, scholars face additional challenges when they attempt to measure political trust in China. Jie Chen, for example, finds that longstanding one-party domination has rendered the distinction between diffuse trust in regime and specific trust in the central government practically irrelevant.Footnote 14 There is also evidence that many people do not clearly differentiate between the central government and central leaders.Footnote 15 Moreover, nationwide surveys, owing to various constraints, have relied on a single measure to measure trust in the central government, which causes two problems. First, the surveys fail to make the crucial distinction between confidence in the central government's policy intent and confidence in its capability to have local deputies implement its policies.Footnote 16 Second, they fail to take into account the fact that popular trust in the central government varies considerably across domains of issue. For instance, people may have strong confidence in the central government on issues which involve little conflict of interests between central leaders, local authorities and ordinary citizens – for example, developing the economy. However, they may feel far less confident about the central government on issues which involve serious tensions, such as controlling corruption, protecting the lawful rights and interests of ordinary citizens, and developing political democracy. The surveys, in other words, fall short of explicitly assessing what people trust about the central government on what issues. As a result, survey findings about trust in the central government are hard to interpret.
Fortunately, five national surveys – the 2002 and 2008 Asian Barometer Surveys, the 2003 and 2006 AsiaBarometer Surveys, and the 2008 China Survey (see Appendix) – collect two additional sets of data which make it possible to reassess trust in the central government. First, without using the terms “trust” and “central government,” the surveys ask respondents to evaluate policy processes and outcomes for which the central government has a distinctive share of responsibility. In other words, along with direct observation of general trust in the central government, which is considered subjective in that respondents are left to define the meaning of trust, there is a set of observations of more objectively assessed trustworthiness of the government with regard to specific issues.Footnote 17 The two sets of observations allow researchers to reassess trust in the central government by examining whether individuals who express general subjective trust also have objective trust on particular issues.
Equally important, the surveys collect information on trust in local government, which turns out to be considerably weaker than that for the central government. As is shown in Table 1, although over 85 per cent of respondents trust the central government, between one- and two-thirds of respondents hold hierarchical trust.Footnote 18
Table 1: Patterns of Trust in Government
Notes:
Column entries are percentages; column totals may be above or below 100 owing to rounding errors. Missing responses of the China Survey are multiply imputed while those of the other four surveys are excluded listwise.
Scholars have long noted that many people in China are less trusting of local government than they are of the central government.Footnote 19 As they attempt to explain the intriguing gap, however, researchers often implicitly assume that trust in local government and trust in the central government can be assessed independently of each other. Xiaobo Lü, for instance, argues that negative education experience undermines trust in local government without weakening trust in the central government, and that the awareness of education reform policy enhances trust in the central government but not trust in local government.Footnote 20 Similarly, Ernan Cui et al. argue that land requisitions in the countryside undermine trust in local government without affecting trust in the central government.Footnote 21 Tang and Huhe recently note that trust in the central government and trust in local government should be examined “simultaneously,” yet they nonetheless share the assumption that the two kinds of trust can be assessed independently.Footnote 22
The assumption, however, may be too simplistic. Trust in local government may well have a dual meaning in China. While it certainly indicates how ordinary citizens assess local government's commitment and capacity to serve their interests, manifest trust in local government may also indicate how citizens evaluate the central leadership's commitment and capacity to ensure that its local deputies serve the public interest. After all, despite the constitutional principle that leaders at all levels of government are elected by the corresponding people's congresses, in reality it is central leaders who are ultimately responsible for selecting local leaders through a top-down chain of appointment.Footnote 23 Under such an integrally nested hierarchical system, explicit distrust of local government is almost bound to reflect implicit scepticism about the central leadership. Comments such as “if the upper beam is not straight, the lower beam will be crooked” may insinuate doubts about the central government's policy intent, while complaints about local officials deceiving the centre may indicate doubts about the central leaders’ monitoring and disciplining capabilities.Footnote 24 Two recent studies support this hypothesis. Michael Lewis-Beck et al. find that people who are dissatisfied with local government practices tend to be significantly less satisfied with the central government.Footnote 25 Similarly, Yang Zhong finds that urban residents who have a less positive evaluation of the performance of municipality governments express stronger distrust of both central and local governments.Footnote 26 Both findings suggest that assessments of policy processes and outcomes for which the central government has a distinctive share of responsibility may indicate latent trust in the central government.
Pursuing this line of reasoning further, this study adopts an integrated approach to triangulate trust in the central government. It employs observed trust in central government and observed trust in local government as reference points to identify an assessment of a policy process and outcome which indicates latent trust in the central government. Then it uses the identified indicator as a criterion of comparison to determine if holders of hierarchical and equal trust have the same amount of latent trust.Footnote 27 If hierarchical trust holders turn out to be less trusting, the fact that there are so many of them indicates that trust in the central government is weaker than it looks.
The Meaning of Hierarchical Trust
The 2008 China Survey generates the richest data for exploring if hierarchical trust holders have the same amount of latent trust in the central government as equal trust holders (see the Appendix for information on sampling, weighting and missing data imputation). The survey measures the trust respondents have in county, provincial and central government leaders with a four-level scale: (1) “do not trust at all,” (2) “do not trust much,” (3) “trust somewhat” and (4) “trust very much.” Respondents trust central leaders much more than they do provincial leaders and county leaders. While 43 per cent of 3,989 respondents trust central leaders very much, respectively 22.7 and 15.8 per cent feel the same about provincial leaders and county leaders (see Table 2).
Table 2: Trust in Government Leaders
Source:
China Survey.
Note:
N = 3,989. Column entries are percentages; column totals may be above or below 100 owing to rounding errors. Missing responses are multiply imputed.
Four major patterns emerge when it comes to examining how respondents assess the trustworthiness of three levels of government leaders against each other. The most common pattern is hierarchical trust. Among 3,989 respondents, 44.8 per cent express strong or modest confidence in central leaders but have less confidence in provincial and/or county leaders. The second common pattern is equal trust, which is held by 39.9 per cent of respondents. The other two patterns are clear minorities, as 10.5 per cent hold equal distrust and 4.8 per cent hold the pattern of “paradox of distance.”
John Stuart Mill's “method of difference” is used to test whether hierarchical trust holders have weaker latent confidence than equal trust holders in central leaders on particular issues. The analysis proceeds in three steps. First, the distinction between hierarchical trust and equal trust is operationalized as relative distrust of local authorities. Second, satisfaction with political democracy in the country is identified as a proxy indicator of latent trust in central leaders. Lastly, hierarchical trust holders are compared with equal trust holders in terms of their satisfaction with democracy.
Relative distrust of local authorities
Hierarchical trust holders and equal trust holders are indistinguishable in terms of their manifest trust in central leaders. What separates them is whether they express less confidence in provincial leaders and/or county leaders. For simplicity, the groups that express relative distrust of provincial leaders and of county leaders are merged into a broader category of relative distrust of local authorities, which is operationalized as follows. First, relative distrust of provincial leaders is defined as the positive differential between trust in central leaders and trust in provincial leaders.Footnote 28 Second, relative distrust of county leaders is defined as the positive differential between the trust in central leaders and trust in county leaders. Third, relative distrust of local authorities is operationalized as a dichotomous variable, with “0” indicating having neither relative distrust of provincial leaders nor relative distrust of county leaders, and “1” indicating having one or both kinds of relative distrust. By this measure, 67.7 per cent of 1,714 respondents who show strong confidence in central leaders hold hierarchical trust. Among 1,582 respondents who express modest trust in central leaders, 34.4 per cent hold hierarchical trust.
Proxy indicator of latent trust in central leaders
The survey asks respondents to assess a wide range of policy processes and outcomes, including political democracy. The results reveal a high level of complacency on the issue. When asked to indicate the degree to which they think that political democracy in the country is a problem, 3,989 respondents score an average of 4.2 on an 11-point scale, which ranges from “0” (it is not a problem at all) to “10” (it is an extremely serious problem). For convenience, perceived severity of the problem of political democracy is reformulated as satisfaction with democracy. It turns out that the average degree of satisfaction is 6.8 on an 11-point scale (or 61.8 on a 100-point scale).
Assessment of political democracy in China is a likely proxy indicator of latent confidence in central leaders. Although the term “democracy” has multiple meanings in China, including Confucian paternalism, “mass line” and electoral accountability,Footnote 29 it turns out that central leaders are held to bear a distinctive share of responsibility on the issue. Regardless of their demographic backgrounds, i.e. gender, age, education, household registration and Party membership, respondents who are less satisfied with political democracy tend to be significantly less trusting of central leaders, controlling for their trust in provincial and county leaders.Footnote 30 In other words, individuals who are unsatisfied with political democracy, in addition to blaming provincial and county leaders, hold central leaders responsible for the problem. They may have either doubts about central leaders’ commitment to build democracy, or doubts about central leaders’ capacity, or both.
In order to ascertain that satisfaction with political democracy can be treated as a proxy indicator of latent trust in central leaders, regression analysis is employed.Footnote 31 Observed trust in central leaders is treated as the dependent variable.Footnote 32 Satisfaction with democracy is the predictor of interest. In addition to trust in provincial and county leaders, several other factors that are theoretically expected to affect trust in central leaders are controlled. First, satisfaction with government policies is controlled. When asked to indicate the degree to which they agree with “in general, I am basically satisfied with government policies” on a five-level ordinal scale, 33 per cent of 3,989 respondents choose “strongly agree,” 53.2 per cent “somewhat agree,” 6.4 per cent “neither agree nor disagree,” 5.7 per cent “somewhat disagree,” and 1.7 per cent “strongly disagree.” It is well understood that in Chinese political discourse, the term “government policies” (zhengfu zhengce 政府政策) usually refers to laws and policies promulgated by the central government, although it does not have “central” as a qualifier. To the extent that survey respondents regard central leaders as makers of “government policies,” satisfaction with government policies is expected to affect trust in central leaders.
Second, perception of local government corruption is controlled because people may blame both central leaders and local officials for the problem.Footnote 33 Respondents are generally quite critical of local government corruption. When answering the question, “how serious is the problem of cadre corruption in this locality?” 4.7 per cent choose “not serious at all,” 31.2 per cent “not too serious,” 40.7 per cent “somewhat serious” and 23.5 per cent “very serious.”
Third, life satisfaction is controlled.Footnote 34 Respondents are asked to indicate their degrees of satisfaction with household income, life in general and current job on an 11-point scale, which ranges from “0” (not satisfied at all) to “10” (satisfied very much). The three indicators constitute a reliable simple summation index (α = 0.84). The respondents score an average of 16.2 on the 33-point scale (or 49 on a 100-point scale).
Fourth, five background variables are controlled. Membership of the Chinese Communist Party is controlled as Party members are expected to be more confident about central leaders than non-members. Age, gender and education are controlled because previous studies have inconsistent findings about their effects on political trust in China.Footnote 35 Lastly, household registration is controlled as rural residents are expected to have stronger confidence in the central government than urban dwellers owing to a few recently promulgated popular policies such as the abolition of agricultural taxes and fees.
As shown in Table 3, satisfaction with democracy has a robust correlation with trust in central leaders.Footnote 36 All other things being equal, individuals who are more satisfied with political democracy tend to have greater confidence in central leaders. Conversely, people who are less satisfied tend to be less trustful.Footnote 37 Although it falls short of conclusively proving that satisfaction with democracy is an indicator of latent trust in central leaders, the analysis shows that satisfaction with democracy quite likely indicates latent confidence in central leaders’ commitment and/or capacity to build democracy.
Table 3: Predicting Trust in Central Government Leaders
Source:
China Survey.
Notes:
N = 3,989. Entries are unstandardized ordered logit regression coefficients, with robust standard errors in parenthesis beneath them. †p ≤ 0.10; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001. Missing data are multiply imputed. Data are weighted.
Results of comparison
Using satisfaction with democracy as a proxy indicator of latent trust in central leaders, a comparative analysis is conducted to explore whether hierarchical trust holders are less trusting of central leaders than equal trust holders in this regard. In other words, satisfaction with democracy is used as the criterion of comparison against which hierarchical trust holders and equal trust holders are compared with each other. The comparison of means shows that hierarchical trust holders are significantly less satisfied with democracy in China. Among 1,714 respondents who express a strong trust in central leaders, 554 equal trust holders have a mean score of 7.9 on the 11-point scale of satisfaction. By contrast, 1,160 hierarchical trust holders have a mean score of 7.1. The difference of means between the two groups is highly significant (p < 0.001). Similarly, among 1,653 respondents who express modest trust in central leaders, 1,038 equal trust holders have a mean score of 6.9 on the 11-point scale. By contrast, 544 hierarchical trust holders have a mean score of 6.0.Footnote 38 The difference of means between the two groups is also highly significant (p < 0.001).
Regression analysis is employed to examine if the observed correlation between holding hierarchical trust and being less satisfied with political democracy is significant among people with different demographic backgrounds. Satisfaction with political democracy in the country is treated as the dependent variable.Footnote 39 Holding hierarchical trust or equal trust is the predictor of interest. Observed trust in central leaders is controlled so that individuals who express the same level of confidence are compared with each other. Also controlled are five demographic variables. The results are summarized in Table 4.
Table 4: Predicting Satisfaction with Democracy
Source:
China Survey.
Notes:
Entries are unstandardized OLS regression coefficients, with robust standard errors in parenthesis beneath them. †p ≤ 0.10; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001. N = 3,296. Missing data are multiply imputed. Data are weighted.
The result shows that individuals who hold hierarchical trust tend to be less satisfied with political democracy in the country, regardless of their demographic backgrounds.Footnote 40 To the extent that satisfaction with political democracy indicates latent trust in central leaders in this regard,Footnote 41 the result suggests that hierarchical trust holders are less trusting of central leaders than they sound.Footnote 42 The fact that nearly 45 per cent of respondents hold hierarchical trust indicates that the trust in central government leaders is considerably weaker than it looks.
Evidence from the other four surveys
The 2002 and 2008 Asian Barometer Surveys and the 2003 and 2006 AsiaBarometer Surveys corroborate the finding that hierarchical trust holders have less trust in central leaders with regard to developing political democracy. All four surveys ask respondents to evaluate political democracy in China, although they use different measures. The Asian Barometer Surveys observe that about 70 per cent of people are satisfied with the way democracy works in the country.Footnote 43 The AsiaBarometer Surveys find that Chinese people are fairly satisfied with their democratic rights.Footnote 44
All four surveys find that satisfaction with democracy has a systematic and independent correlation with observed trust in the central government. As shown in Table 5, individuals who are more satisfied with political democracy tend to have more confidence in the central government, controlling for 1) trust in local government; 2) evaluation of the government's performance in developing the country's economy;Footnote 45 3) perception of local government corruption; Footnote 46 4) life satisfactionFootnote 47 and demographic backgrounds. Conversely, individuals who are less satisfied with democracy tend to express weaker trust in the central government.Footnote 48 The finding suggests that satisfaction with democracy indicates latent trust in the central government with regard to developing democracy.
Table 5: Predicting Trust in the Central Government
Notes:
Entries are unstandardized ordered logit regression coefficients, with standard errors in parenthesis beneath them. “–” indicates that no data are collected. Missing responses are excluded listwise. *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001.
Multiple regression technique is used to examine whether hierarchical and equal trust holders have the same amount of latent confidence in the central government with regard to democracy. As shown in Table 6, controlling for demographic backgrounds, hierarchical trust holders tend to be less satisfied with democracy than equal trust holders. The results corroborate China Survey's finding that hierarchical trust holders have less confidence in the central government than equal trust holders, although they sound equally confident.Footnote 49 The fact that between 33 and 63 per cent of respondents of the four surveys hold hierarchical trust indicates that trust in the central government is significantly weaker than it appears.
Table 6: Predicting Satisfaction with Democracy
Notes:
1 Entries are unstandardized ordered logit regression coefficients, with standard errors in parenthesis beneath them. 2 Entries are unstandardized OLS regression coefficients, with standard errors in parenthesis beneath them. “–” indicates that no data are collected. Missing responses are excluded listwise. *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001.
To the extent that hierarchical trust holders have in real terms less trust in the central government, the four surveys also indicate that public confidence in the central government declines considerably from 2002 to 2008. The two waves of the AsiaBarometer Survey show that the number of hierarchical trust holders increases by nearly 14 per cent from 2003 to 2006. Similarly, the two waves of the Asian Barometer Survey show that the number of hierarchical trust holders grows by 15 per cent from 2002 to 2008.
Sources and Significance of Hierarchical Trust
The analyses above show that hierarchical trust has two layers of meaning. Explicitly, it is the combination of stronger trust in the central government and weaker trust in local government. Implicitly, it is a pattern of partial trust in the central government in that manifest distrust of local authorities reflects latent doubts about central leaders. While more refined data are needed to determine the exact substance of hierarchical trust, existing studies suggest that hierarchical trust is most likely a mixture of stronger confidence in the central government's policy intent and weaker confidence in its capability to monitor and discipline local deputies.Footnote 50
Hierarchical trust may have three related sources. First, it may derive from observing how the government system appears to operate. The multilevel principal-agent system is prone to generate hierarchical trust. On one hand, central leaders can cultivate public confidence in their commitment by promulgating policies that look appealing to the public. On the other hand, local authorities often lack the required incentives and resources to implement the policies that ordinary people find beneficial.Footnote 51 Worse still, it is politically suicidal for local government leaders to inform the public that many popular policies are in fact “unfunded mandates.” Without knowing that the central policies that they find favourable are often “empty promises,” ordinary citizens may believe that such policies are genuine and give credit to the political intent of central leaders. For the same reason, people who are frustrated about poor policy implementation tend to attribute the problem to the ill-intent of local authorities. More critical individuals may suspect that central leaders share the responsibility for poor policy implementation, but they may put the problem down to central leaders’ lack of capacity to monitor and discipline local officials. In other words, among people who are confident about central leaders’ policy intent, those who are unsatisfied with processes and outcomes of policy implementation may remain confident about central leaders’ policy intent, although they lose some confidence in central leaders’ abilities to monitor and discipline local authorities.
Second, the regime employs political propaganda and censorship to foster hierarchical trust. The official media keeps glorifying central leaders by trumpeting the country's “great achievements.” Meanwhile, it shields central leaders from popular discontent by scapegoating local officials for things that have gone wrong.Footnote 52 In addition, the official news media selectively exposes corrupt local authorities, while a tight censorship system blocks negative news about central leaders and their families.Footnote 53 As a result, people who receive political information primarily from official sources or via the internet, which is restricted by the “Great Firewall,” may develop hierarchical trust.Footnote 54
Lastly, there is the cultural myth of the emperor as a paternalistic owner of the kingdom which may predispose people to believe that central leaders must want to protect common people from excessive exploitation by abusive local authorities.Footnote 55 The myth has a particularly strong favourable effect on trust in the top leader, whose commitment to governing the country well is often taken for granted. Meanwhile, the same cultural myth fosters distrust of local authorities, who are believed to have little interest in the long-term stability of the regime and will merrily exploit all opportunities for corruption.
Other than the three substantive sources, political caution may have contributed to the prevalence of hierarchical trust. Since survey questions do not differentiate between dimensions of trust on different issues, cautious respondents who have partial trust in central leaders on some issues may sound as if they have full confidence on all issues.Footnote 56 By contrast, political caution tends to have a weaker inhibitive effect on the expression of distrust of local authorities. Political caution thus makes the “attitude generalization” effect work only in favour of central leaders.Footnote 57 Honestly overstated trust in central leaders and truthfully expressed distrust of local authorities thus explain why so many people hold hierarchical trust.
Hierarchical trust may have a dual effect on behavioural orientations. Relative distrust of local authorities may enhance a sense of rights deprivation while at the same time strengthen confidence in the efficacy of appealing to central leaders. Hierarchical trust may thus be a cornerstone of the ideational foundation for rightful resistance, which typically involves citing central policies to justify claims and seeking favourable intervention from the central government.Footnote 58 However, steeply hierarchical trust may work just like distrust of central leaders. Underneath the belief that all subnational authorities are totally corrupt may lie a belief that central leaders are well-meaning but totally incapacitated.Footnote 59 People who have strong relative distrust of local authorities may be more inclined to adopt disruptive tactics when they protest against local authorities.
Conclusions
This study finds that there is significantly less popular trust in the Chinese central government than five national surveys suggest. The evidence is indirect but strong. First, the surveys show that between one- and two-thirds of respondents hold hierarchical trust. Second, all other things being equal, people who are less satisfied with political democracy tend to be less trusting of the central government. Lastly, hierarchical trust holders tend to be less satisfied with political democracy in China than those who express equal trust for central and local governments. Put together, the three findings indicate that hierarchical trust holders are less trusting of the central government than equal trust holders with regard to developing political democracy, although they sound equally confident. The fact that so many respondents hold hierarchical trust indicates that trust in the central government is significantly weaker than it looks.
The finding calls for a re-examination of survey data about trust in China's central government. Sceptics are right in warning against uncritically accepting survey results about trust in the central government; however, simply dismissing survey results risks “throwing the baby out with the bath water.” This study shows that the data collected in five nationwide surveys are in general valid and reliable. The evidence is that correlations between observed trust in the central government and evaluation of policy processes and outcomes conform to theoretical expectations. The challenge for researchers is to contextualize survey data carefully.
Methodologically, this study demonstrates how to use an integrated approach to triangulate trust in the central government. Researchers can use manifest trust in the central government and manifest trust in local government as reference points to identify assessments of policy processes and outcomes for which the central government has a distinctive share of responsibility as indicators of latent trust in the central government. Then, they can use the identified indicators as criteria of comparison to determine if holders of hierarchical and equal trust have the same amount of latent trust in the central government with regard to the assessed policy processes and outcomes. The integrated approach generates a more accurate assessment of trust in the central government at a given time as well as its shifts in trend.
Last but not least, the integrated approach enables researchers to obtain a more balanced assessment of public confidence in the current central leadership. As his anti-corruption campaign continues to break new ground, President Xi will most likely amass a huge amount of popular trust, which will be reflected in broadly assessed trust in the central government should another national survey be conducted in the coming year or two. In the meantime, however, the exposure of the outrageous corruption of senior leaders such as Zhou Yongkang 周永康 and high-ranking provincial leaders is likely to encourage even more people to develop hierarchical trust. Moreover, widespread hierarchical trust is likely to become not only greater but also highly personalized, as more people conclude that only Xi and his closest allies are trustworthy. In the absence of an institutional mechanism to convert popular trust in a strong leader into support for the one-party system, however, Xi's rise as a political strongman is likely to have a dual effect on the regime's survival. On the one hand, the regime may well remain stable as long as he stays in power; on the other hand, his success as a political leader may create a major succession crisis in the near future, not least because highly personalized hierarchical trust can easily degenerate into a dangerous cult of personality which, as we saw in the Maoist era, jeopardizes peaceful power transition.
Appendix
The China Survey is a project carried out by the College of Liberal Arts at Texas A&M University in collaboration with the Research Center for Contemporary China at Peking University. It is based on a stratified multi-stage probability sample of all Chinese adults, which was drawn using GPS/GIS Assistant Area Sampling (see Landry and Shen Reference Landry and Shen2005). A total of 5,525 target respondents were selected from 75 counties/districts. The end result was a national probability sample of 3,989 individuals aged 18 or over, drawn from 73 county-level administrative regions, representing a response rate of 72.2 per cent. To adjust for survey design effects, each primary sampling unit is treated as a cluster. Data are weighted in terms of strata, age and gender, based on the 2000 Census data. Data are used with permission.
Like other surveys, the China Survey encounters the problem of missing responses. Nearly 30 per cent of the 3,989 respondents did not answer one or more of the three questions about trust in government leaders. To improve the efficiency of estimation by reflecting additional variability owing to the missing values, this study assumes that observations are missing at random (MAR) and adopts the multiple imputation approach (King et al. Reference King, Honaker, Joseph and Scheve2001). Five multiply imputed datasets were generated using Amelia II (James Honaker, Gary King and Matthew Blackwell. 2012. Amelia II: A Program for Missing Data. R package version). All analytic models were also fitted to the original data using listwise deletion. Results obtained from the two alternative treatments of missing values are highly consistent.
