The Power of Everyday Networks in Nation-Building: The Case of Inter-Ethnic Friendships in Singapore

Abstract

While scholars commonly see nation-building as a modernist project or a cultural assertion, we suggest that a “third way” is equally important. Analysing data from a representative survey of 2,001 Singaporean residents collected in 2020 during the COVID-19 pandemic, we demonstrate that everyday social networks have been key to creating and maintaining Singaporean nationalism. We make the point that despite ambitious plans at transforming society, modernist projects must rely upon organic, routinised, and quotidian mechanisms such as friendship-making within and between ethnic groups for national cohesion to materialise.

Introduction

In responding to the question “What makes a nation?” (Renan 1882), preeminent scholars have identified two key factors, suggesting “institutions” (Gellner 1983) and “culture” (Smith 2010) explain the rise of nations and nationalism. Concerning the first, scholars argue nations are manufactured political constructions. For example, Gellner asserts it is nationalism “that engenders nations, and not the other way round” (Gellner 1983: 64). He goes on to say that “nations – as a natural, God-given way of classifying men, as an inherent though long delayed political destiny – are a myth” (Gellner 1983: 64). James Scott (1998) concurs, saying nations are products of “high modernism”, buttressed by a strong desire amongst state elites to build the perfect society based on social engineering techniques. Concerning culture, while Gellner and Scott argue nationalism precedes nations (Gellner 1983: 64; Scott 1998), Smith (2010) takes a different view. He asserts nations precede the political projects of nationalism; they appear “much further back in time” (Smith 2010: 105). Furthermore, as Hobsbawm (1992) points out, people sometimes share common characteristics such as history, culture, language, and religion.

In this paper, we propose a “third way”: everyday networks are a significant source of nation-building. We contend routine and quotidian elements within social life, such as ties formed between co-nationals, are powerful building blocks of a sense of national belongingness. By quotidian, drawing on Lefebvre (1971 [1968]), we mean the following:

. . . what is humble and solid, what is taken for granted and that of which all the parts follow each other in such a regular, unvarying succession that those concerned have no call to question their sequence [ . . . ] and it is the ethics underlying routine and the aesthetics of familiar settings. (24)

Invoking this concept of the quotidian, we assert nations are built – often unspectacularly – through everyday associations between inhabitants. In other words, we present a “grassroots version” of nation-building (Ho 2020: 47; Read 2012), seen in how nations arise out of a fundamental desire for friendship and conviviality (Wise and Velayutham 2014).

This paper advocates for the importance of everyday communities, expressed in how daily interactions – taking place within “humble” (Ho 2020: 29) settings such as neighbourhoods (as well as outside them) – are critical to a sense of national belongingness. Following Wise and Velayutham (2014), we endorse the idea of “everyday multiculturalism” enacted in Singapore's public housing sector and beyond, arguing it has served as a “transversal enabler” of “affectively-at-ease-relations of coexistence and accommodation”, alongside norms of “intercultural conviviality” within neighbourhoods (406–407). Empirically, we test this assertion using recently collected survey data from Singapore (collected in 2020 during the COVID-19 pandemic). Specifically, we test the hypothesis that social relationships – including inter-ethnic friendships – have been a key enabler of Singaporeans’ sense of national belongingness.

Concerning inter-ethnic friendships, we follow Portes (2014) to suggest that embeddedness in one's ethnic community, but without ties to other communities, can entrench ethnic segregation. In contrast, having links to people from different communities can increase network diversity and broaden one's perspective from a group orientation to national consciousness. We argue that by connecting people from other parts of the social structure – via inter-ethnic friendships in this case – national unity becomes a lived reality.

In this manner, nations build relationally, one tie at a time, in everyday life's corridors and social spaces. Nowhere can we see such a dynamic than in Singapore's experiment with inter-ethnic ties, where for fifty years, ethnic groups have lived side-by-side in the numerous public housing estates (Chua 2017; Chua et al. 2021).

Institutions, Culture, and Networks as Sources of Nationalism

This section lays out our conceptual foundations, drawing attention to three sources of nationalism: institutions, culture, and networks. We write generally about each, then connect them to the specific context of Singapore, noting how each promotes a sense of belonging to the nation.

First, institutions. Gellner (1983) makes it clear that it is men and women united by political will and purpose who invent nations: “Nationalism is not the awakening of an old, latent, dormant force, (though that is how it does indeed present itself), it is instead the consequence of a new form of social organization, based on deeply internalized, educational-dependent high cultures, each protected by its own state” (Gellner 1983: 63). The literature generally distinguishes between a “civic” and an “ethnic” conception of nationhood as two ideal types (Ariely 2020). Whereas a civic conception characterises the notion of a nation as a community of equal citizens sharing a common set of political principles, values, and duties, an ethnic conception places greater emphasis on national descent and ancestry, history and commonly shared myths, and other sociocultural values (Hofstetter 2023).

In the civic approach, the government is ever-present, whether in framing a specific political culture or establishing the foundations for a particular set of governing principles, ideologies, institutions, and laws. In Singapore, the public housing sector is one primary site of governmental intervention. Since independence, the government has made each neighbourhood a microcosm of society: i.e., residents comprise 75% Chinese, 15% Malay, 8% Indians, and 2% others, as in the general population. As 80% of Singaporeans live in public housing, multicultural experiences permeate everyday life. Each housing estate has the potential to become a laboratory of “social learning” (McFarlane 2011), where residents learn about other groups, reconcile cultural differences, and focus on common identities – such as being “Singaporean” – to unite those cultural differences (Ho and Chua 2018).

Thus, the state mobilises public housing to bring diverse ethnic communities together. They base their conviviality as neighbours on a visual familiarity derived from everyday routines, such as greetings or chatting occasionally (Chua 2017; Ho and Chua 2018). The routinisation is important; by bringing different ethnic groups together in public housing estates, government leaders have replicated “transethnic structures” (Wimmer 2018: 52) throughout the island city-state.

Importantly, this policy of ethnic relations aims at promoting national identity while embracing ethnic diversity. These aims are not mutually exclusive in that strengthening national identity need not entail the suppression of ethnic identity—ethnic diversity is an asset. Of course, ethnic integration does not produce ethnic equality, nor can meritocracy, even with equality of opportunity, lead to ethnic equality. What matters is that citizens see the system as fair and enabling social mobility, thus facilitating ethnic integration and fostering national identity.

Through public housing, the state has sought to create a society of “stakeholders” (Chua 1997). Most Singaporeans are flat owners, giving them a stake in the nation and a sense of belongingness. Residents tie their housing aspirations to a government that has allowed them to use their earnings in mandatory retirement saving accounts to finance their homes, thereby trying welfare (housing) to production (work) (Holliday 2000). The same rules for funding properties through mandatory savings apply to all Singaporeans.

In the realm of national service, the state has also played a hand. Singapore's conscription army has become a basis for forming a national identity by bringing Singaporeans of different classes and racial backgrounds together. National service is often a “great leveller” and a “common denominator” for Singaporeans (da Cunha 1999: 461), based on a “common experience” regardless of ethnicity, religion, or socioeconomic class (Leong et al. 2016: 306).

Second, culture is important. Primordialist explanations of this sort conceive nations in terms of an “ethnic” conception of nationhood, appealing to deeper, historical, ethnic roots and memories, even to nostalgia (Smith 1986). Cultural and community leaders tend to voice these explanations emphasising preserving real and perceived historical essence (Hutchinson 1992; Smith 1989). They differ from modernist elites, who are more concerned with the future development of their nation-state.

Scholars argue that the history of Singapore stretches farther back in time than the official declaration of independence in 1965. Kwa, Heng, Borschberg, and Tan's Seven Hundred Years: A History of Singapore, published by the National Library Board of Singapore in 2019, commemorates that longer history. The authors point to the early rise of Singapore, the Lion City (Singa-pura in Sanskrit), based on an account in the Malay Annals. While they take the story seriously, many historians who have tried to read the Annals for historical facts have found it difficult to reconcile their accounts, thus underscoring the mystery of Singa-pura as folklore and myth. Yet the invocation of Singapore as “Lion City” continues to resonate today, with its expressions found in contemporary national symbols and mascots such as “Singa, the courtesy lion”, unveiled in 1979, “the Merlion”, designed by Fraser Brunner in 1964 as an emblem for the Singapore Tourism Board (STB), and more recently, “Nila”, the lion mascot used in the 2012 Singapore National Games. Thus, invoking history, myth, symbolism, and culture is an attempt to rally modern people to see themselves as a nation with a shared distant past.

Another aspect of culture, but with a more contemporary interpretation, pertains to deliberate attempts to reshape Singaporean culture through a “communitarian ideology”. Here, the work of sociologist Chua Beng Huat is instructive. First, to build a national identity, Singaporean state elites have frequently relied on a specific collective ideology, wherein the emphasis is not on the inherent rights a government owes to an individual – as in the Western model of democracy – but on the inherent responsibilities an individual owes to the community (Chua 1995). In this way, society deems national goals more important than individual ones. In Singapore, the promulgation of nationally shared values such as “society above self”, enshrined in the 1991 “White Paper on Shared Values”, captures the notion that Singaporeans ought to subject their individual interests to national ones and be willing to consider their Singaporean identity as equal to, if not more important than, their ethnic identity (Tan 2012).

The 1980s and 1990s saw the promotion of Confucianism and Asian values (in a majority Chinese Singapore) to eschew the tide of westernisation, which Singapore saw as too insistent on individual rights over collective responsibility (Chua 1995). This communitarian ideology has translated into the social acceptance of large social engineering projects, such as Singapore's experiment with public housing that consistently created opportunities for social mixing across ethnic lines. This project has succeeded mainly in making Singaporeans see themselves as stakeholders in a multi-ethnic society (Chua 2017; Tan 2012). Today, Singaporeans have come to accept “democracy” not as the endowment of individual rights or freedoms but as how to evaluate their government's performance based on principles of progress, unity, and success (Osterberg–Kaufmann and Teo 2022). Thus, the cultural emphasis on collectivism over individualism has created conditions for nurturing a national consciousness that orients people towards shared citizenship (Chua 1995; Tan 2012).

Third, social networks. We suggest interpersonal ties operating at the level of everyday life can be a significant source of a growing sense of national belonging. As argued above, the state has had a hand in social engineering or at least shaping some relationships in domains such as public housing. However, while the government may initiate, the people must respond, and for this to translate into a sense of national belonging, the government cannot coerce this response; the people must volunteer (Ho 2020). A growing number of neighbourhood studies suggest that for state projects to work and maximise their potential, the people must respond to the call for public–private partnerships (Ho and Wong 2023).

Thus, beyond the political construction of nationhood, there is a subterranean level of quotidian realities taking the shape of everyday relationships that are more voluntary than imposed. With such a situation in mind, Anderson (1983) asserts nations are not just “imagined communities”; actual relationships and interactions between people constitute these communities. Relationships at the level of quotidian routines can create numerous opportunities for social formation, including dismantling prejudice between groups due to constant exposure to each other (Pettigrew and Tropp 2006).

That said, some ties are especially conducive to fostering such a sense of nationhood. The literature suggests “bridging ties” (Putnam 2000) are instrumental. Indeed, network scholar Mark Granovetter (1973), in his seminal work “The strength of weak ties”, anticipates such an argument, asserting inter-group ties are more effective for achieving social cohesion than intra-group ties. Intra-group connections are dense and closely knit, making them likely to create silos reinforcing fragmentation. In contrast, weak links have the long-range ability to bridge a range of groups to achieve a more cohesive social structure.

The work of Andreas Wimmer (2018) illustrates such an arrangement in Switzerland, where inter-group ties have been vital to national cohesion. Elites comprise members from all three major linguistic communities: French, Italian, and German. They have coalesced to form an inclusive, “transethnic power structure” at the highest level of government (Wimmer 2018: 52). Wimmer explains this power-sharing as follows:

This transethnic power structure did not emerge because linguistic “minorities” successfully struggled for a balanced representation vis-à-vis a “majority”. Nor did it result from an explicit power-sharing agreement between separate French-, German-, and Italian-speaking elites, as prominent theorists of consociational democracy interpret the Swiss case. Rather, leaders of all linguistic backgrounds formed an encompassing and enduring coalition without any political struggling by minorities and without any negotiating between representatives of the three language groups. (52

Like Switzerland, Singapore is a society of several ethnic groups: Chinese, Malays, Indians, and others, with a government that takes ethnic representation very seriously. Rather than seen as a stumbling block to social cohesion, the government promotes ethnic diversity as an asset, as a means of bringing respectively unique groups together through superordinate goals with which all can identify.

At the level of quotidian realities, Rose Laub Coser (1975) observes that people with ties to different groups learn quickly to relate and adapt to various network members. They develop a sensitivity towards each group and learn to use a language everyone can understand—they become culturally “omnivorous” (Peterson 2005: 257).

Method

1. Sample characteristics

This study's data is from the Making Identity Count in Asia (MICA) study, with information collected between September and November 2020 during the COVID-19 pandemic. To procure the sample, we purchased a nationally representative sampling frame from the Department of Statistics (DOS), a government agency. We informed identified households by letter about the study. An interviewer visited households two weeks before the commencement of fieldwork and chose a participant based on a randomised system to produce variations in the gender and age of the overall sample. We introduced identified prospective respondents to the study and gave them a Participant Information Sheet (PIS). Those who consented to participate completed the questionnaire on a tablet by themselves. The interviewer waited at a distance, ready to provide clarification if required. This approach was also in keeping with the safe distancing measures imposed during the pandemic. Respondents could respond to the survey in any of the four official languages: English, Malay, Mandarin, or Tamil.

A total of 2,001 respondents completed the survey. The study achieved an overall response rate of 67%. The sample was 46% male and 54% female. Respondents ranged in age from 21 to 88: 29% were between 21 and 34; 34% were between 35 and 49; 26% were between 50 and 64; and 12% were 65 years and above. Chinese represented 75% of the sample, Malays 12%, Indians 10.5%, and others 2%, mirroring the national population. The sample included Singapore citizens (89%) and Singapore permanent residents (11%). Those with an ITE (Institute of Technical Education) certificate and below made up 34% of the sample; diploma holders comprised 26%, and graduates made up 40%.

Regarding housing, 26% were residents in one-to-three-room public housing flats, 55% in four-to-five-room public housing flats, and 19% in private housing, including condominiums and landed properties. The sample also comprised respondents at different stages of family formation: 30% were single; 9% were married without children; 33% were married with children under 18; 22% were married with children 18 and above; 5% were divorced or separated; and 2% were widowed.

2. Measurements

Dependent variable

We conceptualised the dependent variable as the sense of belonging to Singapore, measured by the question: “How close do you feel to Singapore?” Although nationhood is a multidimensional belief system, scholars have distilled it into two main dimensions: cognitive and affective (Huddy and Del Ponte 2020). While the set criteria for the cognitive approach is to define national membership and draw the boundaries between national insiders and outsiders (e.g., “Are you a Singaporean citizen?”), an affective understanding underlines such phenomena as an attachment to and feelings about the nation (Hofstetter 2023). Our question “How close do you feel to Singapore?” aimed to capture the latter. The word “feel” emphasises the salience of affection as a marker of nationhood.

The responses were: (1) very close, (2) close, (3) not very close, and (4) not at all close. In the sample, the associated percentages were 55%, 41%, 4%, and less than 1%, respectively. Given these percentages, it made sense for us to recode the dependent variable as three categories: (1) very close, (2) close, and (3) not very close/not at all close. In estimating the determinants of the sense of belonging to Singapore, we used ordinal logistic regression as the most appropriate technique.

Focal Independent Variable

The focal independent variable was the measure of ethnic diversity, based on responses to the following: “Thinking about your close friends, how many of them are: (a) Singaporean Chinese, (b) Singaporean Malay, (c) Singaporean Indian, (d) Singaporean Eurasian, (e) Singaporean from all other racial groups?” For each ethnic group, we assigned “1” as long as respondents said they knew someone as a close friend and “0” otherwise. We summed up the responses across all groups. A respondent with at least one close friend in all ethnic groups received a score of 5. The sample's mean score was 3; thus, the average respondent had ties to contacts from three ethnic groups.

Controls

Our regression model (see Table 1 in “Findings”) included a set of relevant controls: age entered the model as an interval variable. We coded gender as 1 for “female”, with “male” as the reference category. Race entered the model as three dummy variables, “Chinese”, “Malay”, and “Others”, with “Indian” as the reference category. Education entered the model as two dummy variables, “diploma” and “university”, with “ITE and below” as the reference category. Citizenship status included two dummy variables, “Singapore citizen by birth” and “Singaporean by conversion”, with “Singapore permanent resident (PR)” as the reference category. Family formation comprised five dummy variables, “married without children”, “married with children below 18”, “married with children 18 and above”, “divorced and separated”, and “widowed”, with “single” as the reference category.

Table 1. Ordinal Logistic Regression Predicting Sense of Belonging to Singapore

+P<.10, *P<.05, **P<.01, ***P<.001

N=1,936.

Five ethnic groups: Singaporean Chinese, Singaporean Malay, Singaporean Indian, Singaporean Eurasian, Singaporean All other races (e.g., Arab, Filipino).

Model: Ordinal logistic regression—the outcome is a three-category outcome measuring feeling of closeness to Singapore (0 = Not at all close + Not very close, 1 = Close, 2 = Very close).

The proportional odds test (Prob > chi2) is of marginal significance (p = .0887), suggesting we might wish to proceed to estimate a generalised ordered logistic regression.

We included a control for housing, differentiating between public housing (“1”) and private housing (“0”). If public housing represents the strong arm of government in the promotion of ties between Singaporeans, including inter-ethnic ties, then it is necessary to control the effects of public housing to estimate the role of a more voluntary and organic emergence of friendship (including inter-ethnic friendships) on national belonging.

We also included a measure of class diversity, computed as the range of occupational class positionings among contacts known to the respondent. We used 20 occupations, including “nurse” (56), “hawker” (29), “lawyer” (94), “teacher” (68), etc. For each resident, we computed a measure of class diversity by taking the highest prestige score among respondents’ contacts minus the lowest prestige score among respondents’ contacts. The numbers in parentheses (above) indicate the occupation's perceived prestige (based on a survey of occupations conducted by Quah et al. 1991). Calculating class diversity in this manner is standard practice in the social capital literature; it is a good indicator of the number of class resources to which respondents have access (see Lin and Dumin 1986). Controlling the effect of class diversity allows a better estimate of the impact of ethnic diversity on national belonging, as these may be correlated. Assuming ethnic diversity and class diversity of contacts are significantly correlated – for example, people with racially diverse networks also have class diverse networks – controlling for class diversity should produce a more accurate estimate of ethnic diversity.

Finally, we included a measure of values, as we argued these might positively affect the sense of belonging to the nation. The question asked: “How important are these values to you? (a) multiculturalism, (b) meritocracy, (c) tolerance, (d) family, and (e) respect for the law?” Possible answers were (1) not at all, (2) to a small extent, (3) to a moderate extent, and (4) to a great extent. Our model estimated the impact of each kind of value on national belonging.

3. Analytical strategy: Inclusion of a stringent test

In addition to standard regression, we found it appropriate to subject our analysis to further testing to allay possible concerns about causality. For instance: Do inter-ethnic friendships increase national belonging? Or does national belonging increase the propensity to form inter-ethnic friendships? Following the theoretical antecedents above, we wish to establish the former over the latter.

Ideal situation

Ideally, we would have conducted an experimental study where the researcher randomly assigns subjects to treatment and control groups. The treatment group would comprise Singaporean respondents exposed to ethnic diversity (i.e., inter-ethnic friendships), while the control group would have no (or less) exposure. As the assignment of cases is random, the groups would have equivalent compositions on all other characteristics (e.g., gender, race, education, age). Randomisation creates ceteris paribus conditions, where the only thing that varies between groups is the exposure to treatment. After the experiment, the researcher observes if there is a difference in the outcome. A substantial arithmetic difference in national belonging scores would indicate a significant relationship between ethnic diversity and national belonging.

Actual situation

The reality is more challenging. Surveys (our data source) are not experiments but observational studies. Observational studies have no random assignment of cases. Thus, our respondents were likely to have pre-sorted themselves into exposure and non-exposure conditions, creating the problem of confounding. For example, if people of a particular ethnic group Z are more likely to sort themselves into exposure (i.e., ethnic diversity) and are also more likely to have high scores on national belonging, is it their ethnicity per se or the ethnic diversity of their contacts (or both) that determines national belonging? Experimental studies minimise confounding problems because of the random assignment of cases, but they persist in observational studies, such as surveys.

Propensity matching as an innovative response to confounding

Propensity matching is a novel technique that works around the confounding problem (Guo and Fraser 2010). The science is involved, but a brief explanation should suffice for present purposes. Propensity scoring begins with a researcher who builds a model that estimates the propensity of each respondent to enter the treatment group. The factors explaining entry into the treatment group must pre-exist the treatment.

In our study, we chose the factors of age, gender, race, and citizenship status, as these tend to be ascriptive (rather than attained). For example, we decided not to include factors such as education in the pre-treatment model, as inter-ethnic ties (e.g., childhood friends) may precede education completion. In our study, the treatment group refers to respondents with an ethnic diversity score of five (complete diversity). In contrast, the control group refers to all other respondents with an ethnic diversity score of less than five (incomplete diversity). We chose the high bar of complete diversity as our treatment, wishing to test the idea that engagement with various ethnic groups is key to producing a sense of national belonging.

Once the researcher estimates and obtains propensity scores for each respondent, the researcher is ready to find suitable matches for each treated person with a corresponding non-treated person. The matching criterion is as follows: each pair of respondents must share the same (or around the same) propensity-into-treatment score. The computed difference in the outcome between treated and non-treated groups is known as the “effect of the treatment on the treated” (ATET). In our case, ATET represents the causal effect of ethnic diversity on national belonging.

Propensity matching captures the difference in national belonging between two individuals who are similar on a combination of all pre-treatment characteristics; the only difference is that one has the treatment, and the other does not. This situation represents a feasible workaround for the problem of confounding by controlling all pre-treatment differences between pairs of individuals so that each pair is identical, except for the differential amounts of exposure to treatment.

We should briefly mention four established techniques for calculating ATET without deeply exploring technical details. Researchers primarily rely on four matching methods: (1) nearest neighbour matching, (2) radius matching, (3) kernel matching, and (4) stratification matching (Guo and Fraser 2010). In nearest neighbour matching, the researcher most closely matches the propensity scores in the untreated group with those in the treated group. In radius matching, the researcher specifies a radius, and matches qualify if the propensity scores in the untreated group fall into a certain range. In kernel matching, the researcher matches each treated observation with several untreated observations, with weights inversely proportional to the distance between treated and untreated observations. In stratification matching, the researcher bases matches on intervals and blocks of propensity scores. We estimated ATET using all four techniques to ascertain and demonstrate a consistency of findings.

Findings

1. Ethnic diversity and national belonging

Table 1 shows the results of our ordinal logistic regression and offers compelling evidence of a positive relationship between networks and national belonging. The results also suggest that having friends across ethnic groups (e.g., ties to friends from two, three, four, and five ethnic groups) correlates positively with a sense of belonging to Singapore.

The table shows some other interesting results. Age is positively associated with a sense of belonging (β = .034, significant at the .001 level), suggesting older respondents are more likely than younger respondents to say they feel a sense of belonging to Singapore. As builders of and witnesses to Singapore's rise as a nation, older cohorts might feel a stronger connection to this country. Chinese and Malays are less likely than Indians to report feeling a sense of belonging to Singapore. Education is negatively associated with feelings of national belonging; for example, university graduates are less likely to say they feel very close to Singapore than respondents with ITE or lower amounts of education. The mobility of the well-educated could be one reason, as their orientations might be more cosmopolitan. Those married with older children (18 and above) report a stronger sense of closeness to Singapore (β = .421, significant at the .05 level).

Being a citizen of Singapore yields no increased amounts of national belonging over being a permanent resident of Singapore, suggesting cognitive notions of nationhood (e.g., I am objectively a citizen of Singapore) may not necessarily increase the sense of one's affective connection with the country; other mechanisms, apart from citizenship itself, must be mobilised to achieve it.

Importantly for our argument on the value of social networks, those living in public housing report a greater sense of belonging to Singapore (β = .494, significant at the .001 level). This situation is our proxy for the strong hand of government; the coefficient of public housing points to estate living as a critical source of nationalism. It leads us to suggest that the institutions of residents’ committees, provision shops, libraries, and other amenities, which are widespread at the level of estate living, are important sites of activity that inspire feelings of nationalism (Ho and Chua 2018).

It is in the domain of public housing that inter-ethnic friendships more actively forge. Our analysis shows Singaporeans who live in public housing are significantly more likely than those in private housing to name contacts from other ethnic groups. This correlation underscores the need to control for housing type when assessing the role of ethnic diversity on national belonging. Overall, the results suggest two critical pathways. First, public housing increases inter-ethnic ties, and this, in turn, increases national belonging. Second, when we control public housing within the regression, and to the extent public housing represents the strong arm of government, the coefficients related to ethnic network diversity will express the importance of social relationships as organic drivers of nationhood, distinct from the effects of governmental intervention. In summary, the positive coefficients of inter-ethnic friendships suggest such ties have significant potential to strengthen the sense of belonging to Singapore.

Finally, several values are positively associated with the sense of national belonging, in particular, values such as family (β = .456, significant at the .001 level) and respect for law and order (β = .276, significant at the .05 level). Again, to the extent that a strong state has promulgated these values, the fact that we controlled them in regressions suggests relationships have an independent role in shaping feelings of national belonging, apart from values.

Our analysis included a proportional odds test with a null result (p > .05), suggesting we may assume the relationship between each pair of outcome groups is similar. In other words, ordered logistic regression assumes the coefficients that describe the relationship between the lowest and all higher categories of the response variable are the same as those that describe the relationship between the next lowest category and all higher categories (Winship and Mare 1984).

Our p-value is .0887, which only narrowly qualifies it for the proportional odds assumption. Here, some scholars would recommend proceeding to estimate a generalised ordered logistic regression (Agresti 2002). We do this in Table 2. As it turns out, the results are similar to those in Table 1, but we draw particular attention to the results on network diversity.

Table 2. Generalised Ordered Logistic Regression Predicting Sense of Belonging to Singapore

+P < .10, *P < .05, **P < .01, ***P < .001

N = 1,936.

Five ethnic groups: Singaporean Chinese, Singaporean Malay, Singaporean Indian, Singaporean Eurasian, Singaporean All other races (e.g., Arab, Filipino).

The outcome is a three-category outcome that measures feelings of closeness to Singapore (0 = Not at all close + Not very close, 1 = Close, 2 = Very close).

In Table 2, the coefficients on the right-hand side illustrate the impact of factors on the logged odds of feeling “very close” versus “not close and close”. The findings confirm ethnic network diversity has a significant role in increasing national belongingness: for example, the coefficient β = 1.024 for “having close friends from five ethnic groups” is greater than the coefficient β = .883 for “having friends from one ethnic group”. Contrast this with the columns on the left-hand side, where the coefficient β = 1.879 for “having close friends from five ethnic groups” is less than the coefficient β =2.103 for “having close friends from one ethnic group” (this time on the logged odds of feeling “very close and close” versus “not close”). Taking both columns together, we conclude ethnic network diversity is useful for increasing national belonging, particularly in shifting responses into the highest category of “very close”, away from all other lower categories of closeness.

Predictors of ethnic social mixing

Our analysis yielded additional insights into the predictors of ethnic network diversity (Table 3). For example, age is positively associated with selection into ethnic network diversity, with older people having more ethnically diverse networks than younger people (β = .016, significant at the .001 level). Malays, Indians, and Others (minority groups) report greater ethnic diversity in their networks than Chinese (the majority). This result likely reflects group size differences. Since the Chinese comprise the majority (75%), their inter-group contact probability is much smaller than that of minorities. Note that we used this model (Table 3) to calculate propensity scores for selection into treatment.

Table 3. Predictors of (Complete) Racial Network Diversity

+P < .10, *P < .05, **P < .01, ***P < .001

Model: Binary logistic regression – the outcome is dichotomous: complete racial network diversity versus less than complete racial network diversity.

Propensity score analysis

The results of our propensity score matching were very consistent. The ATET coefficients demonstrate a positive relationship between ethnic network diversity and a sense of national belonging (Table 4). The ATET scores range from .065 to .105, depending on the matching method. This finding means having a completely diverse network increases the sense of belonging to Singapore by some 6.5 to 11 percentage points, a substantial amount compared to having a less ethnically diverse network. The t-statistics (between 2 and 4.5) suggest p-values are safely within the customary threshold of .05, hence confirming statistical significance regardless of the matching method.

Table 4. Sensitivity Analysis: Average Treatment Effect on the Treated (ATET) using Four Methods of Propensity Score Matching

Note: All t-statistics are statistically significant, suggesting the positive impact of number of inter-ethnic friendships and sense of belonging to Singapore.

Discussion

While modernists explain national formation in terms of the progress made in elites’ plans and culturalists explain nations in terms of the past, the third way proposes we focus on what is going on right now. The intricate relational networks of ordinary life hold together much of social life. The quotidian speaks of something ordinary but efficacious: nations arise daily, and they connect past, present, and future. Hence, to understand the determinants of a sense of national belonging, we should look to the often-understated role of daily social networks and not limit ourselves to modernist or cultural explanations.

We have proposed that social relationships are beneficial for generating a sense of nationhood. Inter-ethnic ties hold particular promise because they bring together people from different ethnic backgrounds. Under conditions of state management but complemented by voluntaristic friendships, Singaporeans have been able to imbibe a sense of nationhood.

Contrary to observations in other societies where ethnic diversity has accompanied ethnic strife and conflict (Putnam 2007), in Singapore, ethnic diversity has not compromised national cohesion. We suggest three reasons for this. First, public housing has helped create multi-ethnic communities. Second, a communitarian ideology, which eschews the liberal model of rights and liberties and promulgates the sense of one's duty towards community and country, has reshaped culture to inspire an orientation towards the collective, seen in the emphasis on law and order and the family. Third, at the level of social networks, in the Singaporean context, the translation from inter-ethnic friendships to national belongingness operates in a few ways. For one, ethnic diversity inherently implies a need to manage and reconcile differences between cultural communities, and this activates the search for collective superordinate identities, such as being Singaporean. The search for common ground includes the search for a national identity to reconcile and manage differences.

For another, ethnic diversity activates a response towards adaptability: a person with ties to different ethnic communities needs to be able to relate to each of them. The need to be “all things to all people” requires mental agility, a shift in thinking away from cultural silos towards cultural inclusivity. This person imbibes cultural diversity and sees the nation as a container of this diversity. And finally, ethnic diversity finds everyday expression in the national institution of public housing, and it is here that ethnic communities are in a daily association as neighbours. Public housing estates are microcosms of society in that the public housing experience articulates ethnic diversity in proportions that resemble the national profile. Quite simply, the fact that 80% of residents live in public housing makes possible the translation of ethnic diversity on the ground to broader civic effects on a national scale (Ho and Chua 2018). Here, we expect readers may have several questions about our framework for nationalism. In what follows, we anticipate those questions and provide our responses.

First, do network linkages represent a third way, especially if social mixing has become a modernist project applied by state elites through the housing programme? What about the role of propaganda? Is social mixing the result of voluntary action? Or is it traceable to governmental intervention (Sim et al. 2003)? While the Singaporean social mixing is partly programmatic, we suggest its success must depend, at least partially, on organic processes such as the willingness of people to turn structural opportunities into social relationships. In other words, friendships are a choice made by parties who decide it is worth pursuing. Recall that the survey question specifically tapped “friendships”. The voluntary component of friendship suggests one cannot reduce social mixing to social engineering. Instead, people must decide to convert opportunities into friendships in which they willingly invest.

The emerging literature on private–public partnerships in estate living strongly suggests that while the government has the resources to initiate programmes on the ground level, such as building and constructing amenities, residents must be willing participants in government-led efforts to transform their neighbourhoods (Ho 2020). Without the voluntaristic engagement of residents, the government, despite its power and propaganda, ultimately enjoys no real legitimacy. Likewise, while the Singaporean government has engineered numerous opportunities for social mixing in neighbourhoods, residents must choose to convert their neighbourly ties into friendships (Ho and Wong 2023). In short, we acknowledge the strong hand of government in institution-building but also call attention to the role of everyday networks. Statistically, our regression analyses showed that even after controlling for public housing (as representing government intervention), the social network variables remained positively and significantly associated with the outcome of national belonging, suggesting the unique role of voluntary friendship.

Second, do inter-ethnic networks alone contribute to national identity, or do/can other ties of an inter-group nature accomplish the same thing? A 2016 national survey of Singaporean networks suggests the effects of inter-group ties may be more general than specific (Chua et al. 2021). The data demonstrate many instances of a positive relationship between different kinds of network diversity and national identity. In these 2016 data, network diversity includes inter-class, inter-generational, and inter-religious ties. Their positive associations with national identity illustrate that a whole-of-society network structure involving different kinds of network diversities operates to create national cohesion, conforming with Georg Simmel's (1955) view that society is a “web of affiliations”.

Third, we should ask about causality. Do inter-ethnic ties create a sense of national belonging, or does national belonging inspire people to reach across ethnic divides? The theory we have presented here assumes the first: ties significantly shape national consciousness. The propensity score techniques we used, estimating propensity scores based on pre-treatment factors, have enabled us to establish that ethnic diversity can be a causal determinant of the sense of national belonging. Of course, it is not the only determinant.

Fourth, we asked respondents, “How close do you feel to Singapore?” Isn't this question vague, ambiguous, or just plain simplistic? Admittedly, no survey question can ever do justice to or capture the full extent of the meaning of national belongingness. Survey research can be simplistic in this sense. But we did not aim for a profound discourse on national identity. Instead, we wanted to generalise patterns of statistical association from sample to population. Moreover, the measure we used pertains predominantly to affective interpretations of nationhood.

Conclusion

In Habits of the Heart (1985), Robert Bellah writes about America as a society caught between two values. On the one hand, it is a “nation of joiners”, where people belong to clubs, societies, and associations (also see Curtis et al. 2001), something recognised very early on by Tocqueville in his travels to America (Damrosch 2010). On the other hand, America is a highly individualistic society. How, then, can Americans form a society at all? Bellah proposes the idea of a “second language”, where people function in the realm that pertains to themselves (and their immediate communities) and also in the realm of a “second language”, where they experience other cultures and a wider public life:

Yet the public realm still survives, even though with difficulty, as an enduring association of the different. In the civic republican tradition, public life is built upon the second languages and practices of commitment that shape character. These languages and practices establish a web of interconnection by creating trust, joining people to families, friends, communities, and churches, and making each individual aware of his reliance on the larger society. They form those habits of the heart that are the matrix of a moral ecology, the connecting tissue of a body politic. (Bellah et al. 1985: 251)

Echoing Bellah et al., we argue for the coexistence of bonding and bridging forms of social capital, where ethnic groups preserve their ways, customs, and traditions and connect to a broader set of ethnic communities through friendship ties to other groups. Ethnic network diversity can constitute the “connecting tissue” that binds the different ethnic segments of society together, thus forming the “transethnic (network) structure” about which Wimmer (2018) writes.

Although studies show ethnically diverse neighbourhoods are sometimes associated with lower levels of social trust (Anderson and Paskeviciute 2006; Delhey and Newton 2003; Knack and Keefer 1997), we argue for the more optimistic case that if constructive social mixing brings together ethnic groups, the overall social fabric can become stronger. Of course, each society has its history and contemporary circumstances. In Singapore, the separation from Malaysia on 9 August 1965 drove home the message of the importance of racial harmony as key to Singapore's survival. Fifty years later, our data suggest ethnic diversity under a mixture of government intervention and voluntaristic friendship-making has brought a nation closer together, not torn it apart.