Chinese nationalismFootnote 1 has traditionally been examined through the lens of foreign relations. However, the focus is shifting towards the nationalist values of people from all segments of Chinese society. Most scholars agree that Chinese nationalism is on the rise and constitutes a populist, mass movement rather than a product of official policy: while initially fanned by the state for its own legitimization, Chinese nationalism has since developed beyond its control. It springs partly from the vacuum created by the demise of communist ideology, from a deeply-felt national humiliation suffered during its recent colonial history, and also from greater international self-confidence gained with China's rising economic power. Recent international confrontations provoked spontaneous and genuine public outrage. However, the greater affluence of ordinary Chinese has increased their appetite for Western goods, and foreign travel and education, thereby creating contradictions in their nationalist aspirations. The diverse motives of different societal groups have led to much scholarly discussion, especially concerning the nexus between social class and nationalism. The middle class in China is both growing and pivotal in the nation's future political and economic development, and arguments have been made both for an increasing and decreasing level of nationalism amongst its members.
Current scholarship thus depicts Chinese nationalism as an increasingly important but contradictory and multifaceted phenomenon shaped by the interaction between different societal groups which pursue it for diverging reasons. In response to this theoretical discourse, a small number of empirical studies exist that illuminate the relationships between different demographic dimensions and nationalism in China, including education, age, social and political class. However, most of these are based on qualitative or case study research. While these approaches are valid, the bigger picture of the level of Chinese nationalism and its relationship to demographic characteristics remains relatively unexplored. There is a dearth of reliable data on its level, development and demographic correlates. One reason lies in the generally embryonic polling institutions in China.Footnote 2 In addition, the political sensitivity of these issues makes it difficult to conduct academic surveys of the kind which are common elsewhere. As Johnston argues, “in both the scholarly and political punditry worlds, the conclusions about ‘rising’ Chinese nationalism … come mostly from anecdotal evidence or relatively unsystematic reliance on high profile, popular publications in China.”Footnote 3 Even when feasible, surveys can suffer from reliability issues, especially when dealing with sensitive or emotive topics. Nationalism may be overstated in surveys when it is associated with social approval, positive self-image and imposes no cost to the respondent. Surveys may also be subject to bias when they annoy or sensitize respondents to the issue under examination.
This article seeks to redress the comparative paucity of reliable data regarding nationalism in different segments of Chinese society. It reports on a study designed to provide a more reliable measurement of nationalism with respect to possible response or sample biases. Using subjects recruited in a public space, we conducted an incentive-compatible field experiment designed to measure the level of Chinese nationalism overall and among a range of demographic groups. We use the data to provide answers to the following two questions: to what extent do Chinese subjects display nationalism? Which demographic groups exhibit more or less of those nationalistic behavioural tendencies that we observe? Our approach differs from existing studies in three ways. First, rather than targeting particular demographic groups, we use a sample of respondents drawn from the general urban population in a provincial Chinese city. While our approach cannot make the same claims to representativeness as nationwide polls, it did generate enough of a demographically heterogeneous cross section of Chinese people to allow inferences about varied demographic characteristics. Secondly, we elicit nationalist behaviour (rather than attitudes) in an unobtrusive way, which avoids response bias when subjects become aware of the study's purpose. Thirdly, we devised an incentive-compatible experimental task, i.e. one where material consequences are associated with its expression. Our findings support the existing literature that suggests that nationalism is strongest among females, older, rural and less affluent individuals. However, more controversially, we find strong nationalist sentiment in university-educated and professional people. This provides, for the first time, tentative empirical support for arguments for a middle-class Chinese nationalism.
This research report first reviews previous empirical findings which inform our hypotheses. It then provides a methodological discussion of measuring nationalism in China that provides the motivation of our work in more detail. It follows with a report on our own method, followed by the results, and concludes with a short discussion.
Chinese Nationalism and its Demographics
Few studies within the literature on Chinese nationalism have examined its relationship to various demographic variables. A better understanding here is important for at least three reasons. First, coupled with general demographic trends in China such as urbanization, rising living standards and education levels, as well as increasing exposure to foreign influences, insight into which groups tend to have higher nationalism can help assess future development trends. Secondly, to the extent that different demographic groups have diverging stakes in, and access to, nationalist discourse, demographics can reveal the origins of nationalist sentiment. Thirdly, demographic correlates of nationalism can inform the responses of commercial or political decision makers abroad. The debate about consumer ethnocentrism, i.e. the nationalistic obstacles facing foreign exporters of goods or capital to China, provides an example.Footnote 4 It has been shown that the demographic characteristics of ordinary Chinese, especially education and income, can be powerful explanators of their political values.Footnote 5 The following demographics are relevant in the context of Chinese nationalism:
Age and gender
Physical demographics such as age and gender are standard correlates of many types of attitude and behaviour. Little evidence of their relationship with nationalism exists for China. However, they are important: age is a potential predictor of future levels of nationalism, and in China in particular, a series of social and cultural upheavals, such as the Cultural Revolution, suggest that nationalism may be age sensitive. There is a view that the younger generation faces a conflict between identifying with a global community through greater exposure to foreign media and educational opportunities abroad, and nationalist sentiments. This conflict can be resolved by viewing China as an ailing parent, a notion that appeals to filial duty.Footnote 6 Little is known about the relationship between gender and nationalism in China. However, for both demographic characteristics, clues come from other societies where older people and females have been shown to be more patriotic.Footnote 7
Socio-economic status
Social class is commonly related to political attitudes, but is difficult to measure unambiguously. Survey questions regarding income group, self-ascribed status and occupation have been used. The recent emergence of a Chinese middle class with an increasing political and economic voice has generated much debate and opposing hypotheses regarding their nationalism. According to Johnston, there are arguments that the middle class has taken up the nationalism used by the Communist Party for its own legitimization following the decline of Marxist ideology.Footnote 8 Alternatively, a case has been made for a liberal nationalism in the asserting of individual rights against both the state and foreign powers.Footnote 9 Pye identifies a ‘vibrant’ type of nationalism among the middle classes in the treaty ports (such as our study site). This nationalism, although a potential force for the modernization of China, is being suppressed by the state.Footnote 10 By this account, nationalism, as expressed by middle-class entrepreneurship, is driving China's growing economic power.Footnote 11 There are also arguments against middle-class nationalism. The higher education levels and increasing international exposure of middle-class individuals may foster liberal values and a rejection of nationalism as a force against democratization. Higher income may additionally be associated with a greater vested interest in international free trade and demand for foreign imports. Here, nationalist sentiment within a nascent and affluent consumer class may be prone to the patriot's paradox: a growing desire for branded and luxury products, which cannot be met by domestic consumption alone.Footnote 12 Johnston finds some supporting evidence in relatively liberal middle-class attitudes towards international relations and military issues.Footnote 13 Similar findings have been obtained in empirical work in other countries, which shows poorer people to be more patriotic.Footnote 14 The question of nationalism and the middle class is clearly empirically open, and provides one of our main research questions.
Education
The relationship between educational attainment and nationalism, often examined through student respondents, provides important clues about the future political and economic leadership of China. Again, there are conflicting theories. Some suggest a greater student nationalism owing to compulsory patriotic instruction in university curricula after 1989. Educated Chinese have more opportunity to participate in the nationalist discourse via the internet or other media, which exposes them to more sophisticated nationalistic ideology.Footnote 15 On the other hand, another patriot's paradox arises here when a university student's aspirations for international work or to study abroad inhibit the public expression of his/her nationalistic sentiments.Footnote 16 The rising number of Chinese students abroad entails a greater exposure to foreign cultures, which could affect nationalism either way, depending on individual experiences. However, there is little empirical evidence on student nationalism compared to other groups.Footnote 17 An exception is Zhao, who finds no evidence of anti-American nationalism among elite students, possibly because of the economic and pragmatic rationales mentioned above.Footnote 18 However, how these findings compare with the background population has not been established. Previous findings from other countries suggest that higher education levels reduce nationalism and related values, such as patriotism.Footnote 19
Urban/rural influences
An individual's rural or urban background moderates cultural beliefs. Rural people tend to be less affluent and educated, and more likely to have blue-collar occupations. These demographics are generally associated with more conservative values. Rural individuals have been shown to display greater levels of nationalism in other non-Western, post-communist countries.Footnote 20 However, there has been no systematic study of how nationalist values in China might differ. This question is important in understanding the future development of nationalism in light of rapid urbanization resulting from large-scale migration of rural labourers to urban manufacturing centres.Footnote 21 The influx of migrant workers has created significant social tensions, resulting in both official and informal discrimination that has the potential to strengthen regional, rather than national, identities.Footnote 22
Measuring Nationalism
Different empirical approaches to nationalism are used in order to shed light on its demographic correlates. One approach is to measure nationalism at the level of the individual by administering attitude instruments. Political psychologists have identified a number of distinct, but closely related, dimensions of pro-national values and attitudes.Footnote 23 Whereas patriotism reflects positive, emotional attachment to one's own country, nationalism describes feelings of national power and superiority over others. They are distinct but positively correlated constructs. In addition, internationalism (also global openness or cosmopolitanism) extends fellow feeling to others regardless of nationality and is negatively correlated with both nationalism and patriotism. Another value relevant to nationalism, especially in China, is collectivism. This concept involves one's engagement with, and belief in, the interdependence of opinions and welfare within a particular group. Gries argues that, in the Chinese cultural context, nationalistic values involve a Chinese national identity that has similarities to personal identity.Footnote 24 Accordingly, one's identification with the national group is capable of moderating nationalism. There is evidence from other countries that collectivism is positively correlated with patriotism.Footnote 25
Questionnaire responses on subjective issues may be subject to a range of biases.Footnote 26 In the context of nationalism, there may be significant social desirability or self-image bias. Suitable questionnaire items may cause annoyance or, unless distractor questions are used, sensitize the subject to the issue under consideration. Values and attitudes are generally known to have modest predictive ability for behaviour. The expression of nationalism in questionnaires also has few consequences compared to overt behaviour. Boycotting products or publicly protesting against adversary nations involves a real or perceived cost in the loss of foreign goods or the imposition of immigration restrictions.Footnote 27 This is problematic if values are surveyed to provide insight on the kinds of nationalist behaviour policy makers are ultimately concerned with.
One response is to observe nationalistic behaviour directly. However, in field settings, the lack of control of confounding influences may compromise the reliability of observations. As an alternative, one can observe relevant subject decisions under controlled conditions in the experimental laboratory. In particular, incentive compatible tasks can be used, i.e. tasks that carry rewards based on the decisions made to make it in a subject's interest to reveal true behaviour. Nationalistic discrimination has been observed in a range of laboratory experiments with interactions between subjects from different nations. Incentivized laboratory tasks, typically conducted with student subjects, allow control and may reduce response biases. On the other hand, task abstractness and the representativeness of students limit how well experiments of this kind reflect real decision making. As a result, recent experimental work has sought to combine the control laboratory tasks permit with the realism of field studies in more naturalistic subjects, incentives and tasks in ‘field experiments.’Footnote 28
Our purpose is to gauge Chinese nationalism and its demographic correlates using such an alternative approach. First, we used a sampling procedure which generated a demographically mixed and relatively representative group of subjects. Secondly, we designed a field experiment with a naturalistic choice rather than an abstract task, which elicits nationalism in an incentive-compatible way, i.e. by attaching material consequences to particular decisions. We used this new approach to answer specific questions: are our subjects nationalistic according to our task? Which are the more or less nationalistic demographic subgroups among them? In addition, to test the validity of our task measurement, we also explored to what extent it corresponds with the established survey instruments discussed above.
Method
In the experiment, we examined whether the presence or absence of information regarding a foreign and a Chinese charity's nationality affects donation allocations between groups of Chinese subjects in the two respective conditions. The experiment was conducted with adult Chinese subjects in Ningbo 宁波, Zhejiang province, in July 2009.Footnote 29 Ningbo is a relatively affluent coastal city with some six million people, a strong manufacturing base and a large seaport. It has recently become a regional educational hub with several universities and colleges. Chinese project assistants approached every fourth adult passer-by in a city centre shopping location, and stated the reward in general terms. Volunteers were guaranteed anonymity, were fully informed of all aspects of the experiment other than the research objective, and were asked not to discuss it with others. The experiment was conducted in a city centre café, with the cooperation of the owner, and consisted of a task and a survey. Subjects completed these asynchronously upon arrival, with an average completion time of about thirty minutes.
Given the methodological considerations above, we designed a task to elicit nationalistic behaviour in a way that was both unobtrusive and incentive compatible. The task was executed first to avoid potential contamination of subject behaviour from the questionnaire. Conversely, contamination of the latter responses from prior task completion is unlikely as the task was not advertised as such, but merely as a reward choice. We used a between-subjects design and allocated subjects at random to either the control or treatment condition. In both, the subjects were asked to divide (in any proportion) a real ten yuanFootnote 30 donation made on their behalf between the current projects of two genuine charities as a reward for their participation (see appendix for an English translation of the exact information subjects were given in the task). The charities' names were not revealed at any stage of the experiment.
In the control condition, no nationality information was given about the two charities. In the treatment, one was revealed to be Chinese, and the other to be international. In the control condition, subjects therefore chose between a health and an educational charity, based on personal preference given the brief description supplied about each. In the treatment, the only difference provided was in terms of nationality.
We interpreted differing percentages allocated to the two charities between the two conditions as the effect of nationalism. To avoid confounds arising from the reputations of particular charity brands, we avoided providing their names. Similarly, the specific foreign origin was omitted to avoid country-of-origin confounds, i.e. views associated with particular foreign nations. The charities were Médecins Sans Frontières, and the China Youth Development Foundation Hope Project which supports schools in poorer areas of China.Footnote 31
The survey questionnaire was administered after the task. It elicited subject demographics such as age, urban/rural status, marital status, occupation, education and personal income. It also contained items for the collectivism, global openness and patriotism scales to gauge the external validity of our measure (see Table 1).Footnote 32 We used a seven-point Likert scale (from strongly agree to strongly disagree) for all items and presented them in random order (see appendix for items used and their sources). A key indicating the meaning of each point was also included as Chinese subjects are still relatively unfamiliar with questionnaires generally. In order to avoid monotonous and biased responses, we negated some questions. We produced two versions of the questionnaire with the first and second halves reversed in order to spread and permit testing of subject fatigue effects on questions in the last half. No such effects were found. Struggling subjects were administered the questions by the project assistants. At the end of the experiment, subjects were given rewards for their participation in the forms of gifts to the value of 60 yuan. The respective total charity donations were paid to the two organizations concerned after the experiment.
Table 1: Results of the Separate Factor Analyses of the Survey Instruments
Notes:
Cronbach's α- values, descriptives of subjects' scores and numbers of constituent items from original sources given for retained factors.
Results
We obtained valid responses from 447 subjects. The average age of the subjects was 27.4, and 60.1 per cent were female. The relatively young age and high educational achievements of our subjects are typical of shopping mall subjects in large cities, as has been found in other studies. Our first question is to what extent, if at all, subjects exhibit nationalism. Table 2 shows that in the control condition where no nationality information was presented, donation splits between the two charities were 57.5 per cent, close to a 50–50 focal point, but slightly tilted towards the educational charity. When the educational charity was revealed to be Chinese in the treatment, we observe a slight increase towards it of 9.4 per cent. Subjects tended to split the donation equally when charity nationalities were unknown, but their nationalism caused on average only an extra 0.5 yuan out of a maximum of ten to be given to the charity known to be Chinese. The associated swing is modest as its maximum, i.e. had all subjects chosen the Chinese charity in the treatment condition, is 74 per cent. We examined the significance of the nationalism effect using Mann-Whitney tests. The overall finding is support for a modest nationalism: significantly greater shares were given to the Chinese charity in the treatment condition where the charities' nationalities were known (p = 0.038). This can only be explained by nationalistic sentiment as other factors have been controlled for by virtue of the experimental design.
Table 2: Average % Chinese Charity Splits in Control (C) and Treatment (T) Conditions, % Swings between Them and Demographic and Attitudinal Sub-groups
Notes:
The p-values are for one-tailed Mann-Whitney tests for treatment effects. The symbols * and ** denote significance at the 90 and 95% levels, respectively.
To investigate the demographic basis of the nationalism we found, we followed previous studies and dichotomized our sample in terms of each of the demographics and compared the strength of the nationalistic bias in subject choices between the two resulting sub-groups for each. In particular, we used respective median observations to dichotomize the sample in terms of: income (high/low), occupation (white/blue collar), education (university/non-university), urban background (urban/rural), marital status (married/not married) and age (25 and below/above), and tested for treatment effects within these. For each, we performed additional Mann-Whitney tests for differences in charity splits between condition and treatment. The results show that the nationalism we observed in the task varies with all the demographics we elicited. Significant treatment effects towards the Chinese charity were found for female, married and rural subjects (p < 0.05), as well as for older and poorer (p < 0.06) people. We also found this effect in white-collar and university-educated subjects (p < 0.05).
Both the gender and age effect support existing findings. Our effect for marital status, which had not been previously established, could reflect age. Similarly, the result showing poor people to be more nationalistic conforms to empirical literature for other countries. Our results suggest the opposite effect to that found in previous studies both for education and white-collar occupation, a closely correlated demographic. Similarly, university-educated subjects exhibit greater nationalism. The latter two effects are new and suggest that there is a particular educated, professional social group which exhibits nationalism. White collar status is highly related to both university education (2-tail Fisher exact p < 0.001) and urban status (p < 0.01).
We then turned to the attitudinal data collected to validate our measure. We performed a principal components factor analysis for each of the attitude instruments (Table 1). Factors with loadings exceeding 0.5 and cross-loadings less than 0.4 were retained. All eigenvalues of the extracted factors exceeded the conventional threshold of 1.0. In terms of internal consistency, we used the commonly used threshold for Cronbach's α greater than 0.6. Subjects' mean scores for collectivism, global openness and patriotism were a little greater than four out of a maximum of seven. We performed t-tests to examine whether subjects' decisions in the experimental task produced a bias in their subsequent responses to the questionnaire. For all four factors, there were no significant differences between the average responses of treatment and control subjects (all p > 0.32). We concluded that revealing the respective foreign and domestic origin of the two charities did not bias subjects' stated nationalistic sentiments.
We next examined to what extent subjects' incentivized decisions conform to their survey responses as indicative of the validity of the task we designed. To do so, we also dichotomized the attitudinal data using median observations and tested for treatment effects for the resulting sub-samples (Table 2). A bias towards the Chinese charity existed for people with high patriotism (p = 0.014), collectivism (High COL 1 p = 0.041) and low global openness (low GLO 1 and GLO 2 p = 0.042 and 0.029), in line with our expectations from the findings reported in the previous literature discussed above. In general, these results suggest the external validity of our behavioural measure.
We also examined the relationship between subjects' nationalistic sentiments and task behaviour in a multivariate framework. In particular, we regressed individual's Chinese charity allocations on their attitudinal scores using the general-to-specific method.Footnote 33 By experimental design, we could not examine the effects of values on nationalistic behaviour directly. Instead, we tested the interaction between our experimental treatment effect (i) and individual attitudinal scores. The results are presented in Table 3 and mostly confirm our univariate results. In the final model, both PAT and GLO 2 moderated the effect of charity nationality information on behaviour: given this information, people high in patriotism and low in global openness gave relatively less to the foreign charity. Other attitudinal variables as well as a dummy variable for the treatment (i) were not retained due to insignificance. The latter result suggests nationalistic behaviour is manifested only in the presence of certain nationalistic sentiments. However, we have to treat this model with care as it assumes that no relationships exist between subjects' attitudes and preferences for health and education respectively.
Table 3: Regressions for Experimental Treatment Effects on Dependent Variables
Notes:
***, ** and * denote significance at or above the 99, 95 and 90% levels, respectively.
Discussion
This study was designed to contribute empirically to the important debate about the level and nature of nationalism in politically and economically resurgent China. Our field-experimental approach was designed to address methodological issues associated with qualitative, survey and laboratory alternatives. The measure we designed and elicited detected modest but significant levels of nationalism which were found to be externally valid with respect to existing attitudinal measures. We used the measure to identify differences in the expression of nationalism within all of the demographic groups in the study.
Attitudinal studies in political psychology measure subjects' nationalism, globalism, internationalism and patriotism as separate dimensions of their nationalistic sentiment. Such measures can be easily related to subjects' demographic data but may be prone to certain response biases. In contrast, our experimental approach involved incentive-compatible decisions, but did not provide a measurement of nationalism at the individual level. Instead, we gauged nationalism as a behavioural effect between the control and treatment groups' respective decisions. We then tested for the presence of this effect within specific groups by their attitudes or demographics. Our results suggest that the behavioural effects we observed are the result of a combination of a number of particular demographics and nationalist value dimensions which individually exhibit the between-subject effect.
Most of the results we obtained provide support for evidence from case study work in China or empirical work from other contexts. In particular, we found that nationalism tends to be greater in older, rural and lower-income Chinese people. People in this social group are less exposed to foreign media, travel or branded imports that may generate an inclination towards influences from abroad. Conversely, this group may consider their livelihoods particularly threatened by international economic competition. The nexus between gender and nationalism is intriguing, and alternative explanations exist. On a more general level, out-group rejection is known to be associated with less assertive personality traits that are more common in females. In addition, evolutionary psychologists have recently suggested women may discriminate more because of evolutionary adaptations that guard against the possibility of coercion in their reproductive choices.Footnote 34
Our findings relating to education and occupation are interesting and contrary to some existing scholarship. We found that white collar and university-educated subjects exhibit a relatively greater nationalism. As we have seen, opposite arguments have previously been made with respect to education. Our approach does not reveal why university-level educated people are more nationalistic in our sample. Part of the answer may indeed be that university students, with a greater exposure to the global community and awareness of international affairs, see themselves as ambassadors of their nation and have a greater awareness of international rivalries. Another reason may be that universities, especially the elite ones, are training grounds for the country's future elite and are therefore subject to significant compulsory instruction in nationalist ideology.
Appendix
English Translation of Incentive-compatible Task
We would like to make a donation to charity on your behalf. You have a 10 yuan endowment, which you must allocate between the 2 charities listed below. You may give all the money to one of them, or split it in any manner of your choosing. However, the total must sum to 10.
Charity Contribution A
Your money will help build a school in rural China, providing primary education and helping to increase the prospects of children living in several villages.
Amount allocated: ______
Charity Contribution B
Your money will help to provide health care to those suffering all over the world from epidemics, exclusion from health care, malnutrition or natural disasters.
Amount allocated: ______
Items Used, and their Source, for the Survey Questionnaire
Global Openness (Suh and Kwon Reference Suh and Kwon2002)
1. It is necessary to make an effort to understand other cultures' perspectives and to integrate them into my own way of thinking.
2. Living and working in a foreign country may be an influential development experience of my own life.
3. I have a real interest in other cultures or nations.
4. I enjoy being with people from other countries to learn their unique views and approaches.
5. Whenever possible, I avoid buying foreign products.
6. I would feel guilty if I bought a foreign product.
7. Whenever possible, I would prefer to buy foreign products.
8. I like the idea of owning foreign products.
9. I would like to have opportunities to meet people from different countries.
10. It is necessary to make an effort to understand other cultures' perspectives and to integrate them into my own way of thinking.
Collectivism (Hui and Triandis Reference Hui and Triandis1986)
1. It is foolish to try to preserve resources for future generations.
2. People should not be expected to do anything for the community unless they are paid for it.
3. Even if a child won the Nobel Prize, the parents should not feel honoured in any way.
4. I would not let my parents use my car (if I had one), no matter whether they are good drivers or not.
5. I would help within my means if a relative told me s/he is in financial difficulty.
6. I like to live close to my friends.
7. The motto “sharing is both blessing and calamity” is still applicable even if one's friend is clumsy, dumb, and causing a lot of trouble.
8. When my colleagues tell me personal things about themselves, we are drawn closer together.
9. I would not share my ideas and newly acquired knowledge with my parents.
10. Children should not feel honoured even if the father was highly praised and given an award by a government official for his contributions and service to the community.
Patriotism (Kosterman and Feshbach Reference Kosterman and Feshbach1989)
1. I love my country.
2. I am proud to be Chinese.
3. In a sense, I am emotionally attached to my country and emotionally affected by its actions.
4. Although at times I may not agree with the government, my commitment to the Chinese always remains strong.
5. I feel a great pride in that land that is our China.
6. It is not that important for me to serve my country.
7. When I see the Chinese flag flying, I feel great.
8. The fact that I am Chinese is an important part of my identity.
9. It is not constructive for one to develop an emotional attachment to his/her country.
10. In general, I have very little respect for the Chinese people.
11. It bothers me to see children being made to pledge allegiance to the flag or sing the national anthem or otherwise induced to adopt such strong patriotic attitudes.
12. China is really an institution, big and powerful yes, but just an institution.
