Interpersonal Trust across Six Asia-Pacific Countries: Testing and Extending the ‘High Trust Society’ and ‘Low Trust Society’ Theory

Paul R. Ward; Loreen Mamerow; Samantha B. Meyer

doi:10.1371/journal.pone.0095555

Abstract

Background

Trust is regarded as a necessary component for the smooth running of society, although societal and political modernising processes have been linked to an increase in mistrust, potentially signalling social and economic problems. Fukuyama developed the notion of ‘high trust’ and ‘low trust’ societies, as a way of understanding trust within different societies. The purpose of this paper is to empirically test and extend Fukuyama’s theory utilising data on interpersonal trust in Taiwan, Hong Kong, South Korea, Japan, Australia and Thailand. This paper focuses on trust in family, neighbours, strangers, foreigners and people with a different religion.

Methods

Cross-sectional surveys were undertaken in 2009–10, with an overall sample of 6331. Analyses of differences in overall levels of trust between countries were undertaken using Chi square analyses. Multivariate binomial logistic regression analysis was undertaken to identify socio-demographic predictors of trust in each country.

Results

Our data indicate a tripartite trust model: ‘high trust’ in Australia and Hong Kong; ‘medium trust’ in Japan and Taiwan; and ‘low trust’ in South Korea and Thailand. Trust in family and neighbours were very high across all countries, although trust in people with a different religion, trust in strangers and trust in foreigners varied considerably between countries. The regression models found a consistent group of subpopulations with low trust across the countries: people on low incomes, younger people and people with poor self-rated health. The results were conflicting for gender: females had lower trust in Thailand and Hong Kong, although in Australia, males had lower trust in strangers, whereas females had lower trust in foreigners.

Conclusion

This paper identifies high, medium and low trust societies, in addition to high and low trusting population subgroups. Our analyses extend the seminal work of Fukuyama, providing both corroboration and refutation for his theory.

Citation: Ward PR, Mamerow L, Meyer SB (2014) Interpersonal Trust across Six Asia-Pacific Countries: Testing and Extending the ‘High Trust Society’ and ‘Low Trust Society’ Theory. PLoS ONE 9(4): e95555. https://doi.org/10.1371/journal.pone.0095555

Editor: K. Brad Wray, State University of New York, Oswego, United States of America

Received: December 10, 2013; Accepted: March 27, 2014; Published: April 23, 2014

Copyright: © 2014 Ward et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: The study was funded through an internal research grant from Flinders University. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. They simply provided financial resources to collect and analyse data.

Competing interests: The authors have declared that no competing interests exist.

Background

There have been a large number of scholars who have attempted to define, conceptualise and operationalise trust within the disciplines of sociology [1], [2], [3], [4], [5], political science [6], [7], and philosophy [8], [9], [10]. There is also scholarship on trust in other academic disciplines such as economics, social policy and management, but they are outside of the realm of this paper. Within sociology, there exist a number of useful conceptual reviews of trust [11], [12], [13], [14], [15] which reduce the need to provide a full and complete review here. A generally accepted definition of trust is ‘the mutual confidence that no party to an exchange will exploit the other’s vulnerability’ [16], with a trustworthy person having both good intentions and reasonable competence [9]. Nevertheless, we provide a brief overview in order to set the context and conceptual framework for this paper.

Trust is regarded as a necessary component for the smooth running of society [17] and the glue that holds society together [1]. Fukuyama [6] uses economic theory to argue for the necessity of trust, “it is very difficult to conceive of modern economic life in the absence of a minimum level of informal trust” (p.151). Without trust, individuals would be constantly trying to make complex decisions on how to live their lives, deciding on a case-by-case basis whether to trust their neighbour to look after their house whilst they are on holiday, a teacher to look after their children at school or a stranger they talk to at the bus stop to be courteous and not malevolent. These decisions and the time and energy taken to make them, are known as externalities in economic theory [6] which increase the efficiency of the system, be it a social system or economic system. However, there is a concern with the decline in trust in several democratic countries [7] and evidence suggests that modern social developments have led to the erosion of trust in these countries.

Luhmann argues that the act of trusting (or indeed not trusting) reduces complexity in society, allowing individuals to get on with other aspects in their lives [1]. Therefore, understanding the nature and extent of trust in society becomes extremely important – if we can understand which groups are distrusted or which groups are more likely to distrust others, we can begin to develop social policy and maybe social marketing to reduce the distrust, and in Giddens’ view, make society more smooth functioning.

Giddens’ argues that the ‘consequences of modernity’ have been profound on the nature and extent of trust [17] – people have become reflexive actors who use available information and also personal experience in order to make decisions about whether or not to trust another person [3]. Govier [9] similarly argues this when she says “Trust is not faith; we need not trust blindly” (p.7). This reflexivity or questioning of trust has led some researchers to suggest that Western society exhibits a generalized manifestation of distrust, akin to conspiracy theory [18]. This also reflects Hobbesian ideas that human relationships are built on suspicion, greed and competition, which lead to a default position of distrust. However, Govier [9] argues philosophically for trust to be the default position, although she also recognises that ‘blind trust’ can lead to complacency, “the fact that things “seem all right” or that we “have a good feeling about that person” is simply not an adequate basis for trust” (p.9). Govier cites Ghandi’s philosophy which was based on a person being ‘trustworthy unless they proved otherwise’ – if a stranger has not shown themselves to be untrustworthy, we should trust them.

The notion of the ‘default position of trust’ is critical to this paper, since we analyse trust in people that are known (e.g. family, neighbours) but also people that are unknown (e.g. strangers, foreigners) to our respondents. Trust in strangers cannot necessarily be built on ‘rational’ evidence or prior experience like Giddens suggests, unless an individual has had very bad experiences with lots of strangers. Therefore, for a person to decide whether or not to trust a stranger or foreigner, there must be some social or cultural norms underpinning the decision, often based on a constructed characterisation or stylised notion of them [9]. Indeed, Fukuyama [6] argues that “trust arises when a community shares a set of moral values in such a way as to create expectations of regular and honest behaviour” (p.153).

The work of Fukuyama is extremely important within this paper, since he brings culture and social norms into the mix. Fukuyama argues for the centrality of trust in economic development, along with sociologists who argue for its centrality in social development. He argues that the family and civil society create durable social institutions which are resilient to legislative and economic change – trust becomes the glue in these forms of social order. Fukuyama makes this abundantly clear when he says “If the institutions of democracy and capitalism are to work properly, they much coexist within certain premodern cultural habits that ensure their proper functioning” (p.11). He goes on to say “Law, contract, and economic rationality and prosperity…. must as well be leavened with reciprocity, moral obligation, duty toward community, and trust…. The latter are not anachronisms in a modern society but rather the sine qua non of the latter’s success” (p.11). Given this centrality of trust, he then attempts to understand why some societies have lower or higher trust.

Fukuyama develops a hypothesis about ‘low trust societies’ and ‘high trust societies’ which is interrogated within this paper. He argues that within low trust societies, social relations/connections are primarily within the family – familial piety – and that these societies will be less trusting of people outside of the family [6]. His work examines in great depth both Confucianist and non-Confucianist countries in order to develop and argue his thesis. In terms of social networks or social capital, Fukuyama argues that “communities with the strongest internal ties will have the weakest bonds with the outside” (p.154), and he links the internal ties to higher trust within groups and lower trust outside of those groups. In an Asian context, he links trust in the family, as the most important social unit, to Confucianism which regards the family as the most important social unit. He provides extensive evidence of the ways in which the family is still the central unit in Confucianist societies and of the impact this has on reduced trust in people and institutions outside the family. However, he argues that in having less trust outside the family, people are not necessarily individualistic (in an economic rationalist sense), but view the family as the central and most important social unit. Nevertheless, Fukuyama argues that “the lack of trust outside the family makes it hard for unrelated people to form groups or organizations” (p.73).

Fukuyama shows how businesses have developed as family owned/run organisations more in Chinese-centred countries like Taiwan and Hong Kong, but less so in ‘less Chinese’ countries like Japan – with South Korea having parts of both Confucianist philosophy but also a more open attitude to trusting others due to State interventions in attempts at ‘globalising’ South Korea. Essentially, Fukuyama presents the case for Taiwan, Hong Kong and to a slightly lesser degree South Korea being ‘low trust societies’ – having high trust in the family, but low trust outside of the family. He presents the case for Japan being a high trust society, alongside countries like the US, the UK and Germany. Whilst he does not explicitly talk about trust in Australia, we would suggest that it is closer to the US and UK, and thus may be hypothesised as a high trust society. In this paper we also present data on interpersonal trust in Thailand, although Fukuyama does not specifically focus on it. He mentions (p.71) that the small enclaves of Chinese people and culture in Thailand may make it similar to the Chinese influenced countries like Hong Kong and Taiwan, although he also shows that Thailand, like Singapore and Malaysia, accepts and encourages foreign investment, thereby reducing the focus on the family, and potentially increasing trust.

There have been theoretical [19], [20] and empirical [21], [22], [23] critiques of Fukuyama’s work on trust. The theoretical critiques tend to focus on the possibility of being able to have generalised notions of trust, as opposed to context-dependent and locally situated definitions. The empirical critiques have tended to examine the levels and predictors of trust between two countries. For example, Yamagishi & Yamagishi [23] explored the differences in trust between Japan and America, finding some similarities to Fukuyama – trust was generally higher in America than Japan. However, they found different social relations, with Japanese having more durable exchange relations than Americans. Further research comparing Japan and America found little support for differences in trust on the basis of cultural differences between the countries. [21] However, these studies only compared Japan and America, both of which according to Fukuyama are high trust societies, partly due to them not being based on Confuciansim. Therefore, further research is required to examine social trust between Confucianist and non-Confucianist societies.

Fukuyama’s seminal work sets up a number of hypotheses which we test within this paper. The first major hypothesis relates to there being higher trust in non-Confucianist societies as compared to Confucianist societies. The second major hypothesis relates to trust in families being higher in Confucianist societies as compared to non-Confucianist societies. The third major hypothesis relates to trust in social groups other than families being lower in Confucianist socieites as compared to non-Confucianist societies. In order to empirically tests Fukuyama’s hypothesess, we present data on interpersonal trust (trust in individuals rather than individuals/systems) in Taiwan, Hong Kong, South Korea, Japan, Australia and Thailand. Corroboration of Fukuyama’s hypothesese would see Australia, Japan and South Korea (non-Confucianist societies) as ‘high trust societies’ and Taiwan and Hong Kong (Confucianist societies)as ‘low trust societies’. The ‘high trust’ societies would have lower trust in families and higher trust in non-family social groups. Given Fukuyama’s comments on Thailand, one may also regard it a ‘low trust society’, although this was not explicitly stated by Fukuyama.

In addition to empirically testing the above hypotheses, the paper extends Fukuyama’s work by identifying the population groups within each country who have higher or lower trust, thereby not assuming homogeneity within each country. Fukuyama’s work was seminal in providing a sociological analysis of between-country trust, but he did not examine within-country trust, which is understandable given the size of the task and the research aims of his work. In this way, our paper both operationalizes and extends the sociology of trust.

Methods and Dataset

Ethics Statement

Appropriate approvals were obtained within each country to undertake the individual surveys. The authors were granted ethics approval from Flinders University Social and Behavioural Research Ethics Committee to obtain and use the collected data for secondary analysis (project number 5221).

Data Collection

The data presented in this paper come from a larger survey across six Asia-Pacific counties: Australia (Flinders University), Hong Kong (Chinese University Hong Kong), Japan (Chiba University), South Korea (Seoul National University), Taiwan (National Taiwan University), Thailand (King Prajadhipok Institute). The details and critique of the survey methods across the six countries have been published elsewhere [24], [25], although a brief synopsis is provided here for readers to understand the results presented in the paper. Details of method used within the study countries are also published elsewhere [26], [27], [28], [29], [30].

We were cognisant of the various methodological issues related to cross-country research, which could potentially lead to difficulties in interpreting the data. In order to respond to these issues, we undertook a number of strategies to make each country-specific survey as comparable as possible. Initially, we used questions from pre-validated questionnaires, including the World Values Survey [31] and the General Social Survey [32], to develop the questionnaire. The English version of the questionnaire was then validated [33] and subsequently translated into the language of the host country. Consultations were then undertaken with academics from each of the collaborating universities in order to further refine the questionnaires. Pilot studies were then undertaken in each country to validate the questionnaires and to make sure that respondents understood the questions and response categories, reducing the potential for systematic bias. Our decision to include the same questions and response categories across all countries (albeit in different languages) fits in with recognised standards for cross-country survey research [34], [35] Nevertheless, given that trust is both socially and culturally mediated, a potential limitation of the study may reside in slightly different interpretations of ‘trust’ both within and between countries. In addition, research has identified that respondents in different countries may interpret response categories slightly differently [36]. The potential for slightly different interpretations of both the question and the response categories has been termed ‘living with imperfect comparisons’ [37], which remains a potential limitation for all cross-country comparative research.

Data were collected in each country between 2009 and 2010, using either face-to-face or postal survey techniques. The total sample size was 6331∶681 in Hong Kong; approximately 1000 in Australia and Japan; and 1200 in each of South Korea, Taiwan and Thailand.

This paper presents an analysis of questions on inter-personal trust. The question appeared in the questionnaire as: How much do you trust various groups of people? (emphasis in original). There were four response categories: Trust them completely; Trust them a little; do not trust them very much; do not trust them at all. The survey provided a list of 11 different groups against which respondents were asked to rate their level of trust. For the purpose of this paper, we focus on trust in five groups: family, neighbours, strangers, people with a different religion and foreigners. We have chosen family and neighbours because they represent groups ‘known’ to the respondents, whereas the other groups are ‘unknown’. This allows us to examine trust on the basis of family ties and in-depth knowledge in addition to trust without these bases, thereby allowing us to further examine Fukuyama’s hypothesis across six countries. The questionnaire also included 11 questions on socio-demographics such as income, gender and age.

Data Analysis

After the surveys had been undertaken, merging and cleaning of the dataset was conducted by academics at Seoul National University. In addition, once we had a cleaned SPSS dataset, data were weighted for each country on the basis of age and sex, to mitigate potential bias in age-sex response rates. Responses to surveys were generally under-represented by males and younger respondents, and therefore the age-sex weighting created a nationally representative dataset for analysis.

To achieve suitable levels for analysis and a basis for comparison, responses to survey items enquiring about individual’s trust in various groups of people were dichotomised. Retaining the original four categories did not yield enough cases in each category to allow multivariate analysis across each country. The new dichotomised variable does not have the fine grained differences in trust as the original variable, and this may represent a potential limitation in interpreting the results of the paper. For all countries, ‘trust them completely’ and ‘trust them a little’ were combined to form a categorical variable labelled ‘medium/high trust’, whilst responses in the form of ‘do not trust them very much’ and ‘do not trust them at all’ were combined into ‘low/no trust’. Beyond purely descriptive statistics, we were also interested in whether (socio) demographic attributes of respondents hold predictive qualities for trust invested in different groups of people. For this purpose, a pool of categorical predictor variables was generated comprising respondents’ sex, age group, marital status, work status, household income, as well as subjective health status and whether respondents reported a chronic health condition.

Chi-Square analyses were conducted to examine the relationship between country and trust in various groups of individuals, including self-reported levels of trust in family, neighbours, strangers, foreigners, and people with a different religion. All chi-square analyses revealed a significant association between different countries and level of trust placed in the separate groups of individuals (all chi-square analyses significant at p<.001). In order to develop a better understanding of potential mechanisms driving these associations, we examined the standardised residuals obtained for each cell. This allowed us to identify cell counts, which made significant contributions to the overall significance of the particular association under investigation. To facilitate the interpretation of z-scores, the reader is reminded that z-scores with a magnitude equal to or higher than ±1.96 are associated with a two-tailed significance level of p = 0.05, z-scores of ±2.575 refer to a two-tailed probability of p = 0.01 and z-scores of ±3.3 are associated with two-tailed significance at p = 0.001.

For the regression analyses, the predicted level of the dichotomised outcome variables (i.e. level of trust) was specified in the direction of the majority of responses. For instance, for a group of people respondents indicated to trust more frequently than not, the regression model was specified to predict trust. An outcome variable for which more low/no trust responses were observed, low/no trust was predicted. This procedure does not change the analysis for it does not impact on the relationships between variables, but facilitates interpretation of the regression results. Data were analysed using the SPSS version 21.0 (SPSS Inc., Chicago, IL, USA). Binomial logistic regression models were used to investigate associations for all six countries (Hosmer & Lemeshow, 2000). All demographic predictor variables were entered into the analysis as categorical variables. Bivariate logistic regression analyses were performed to examine the relationship between the individual demographic predictors and the various recipients of trust. Only items showing an association at the p<0.25 level were entered into multiple binary logistic regression analyses (Hosmer and Lemeshow, 2000). Following suggestions by Field (2009), for the purposes of the present investigation the method of choice for conducting regression analyses was to enter relevant predictor variables in one block rather than stepwise procedures. Predictor variables that were entered into the model but returned as not significant were in turn tested against models containing only significant predictor variables. This process allowed for the comparison of several models, resulting in a final model containing only variables, which significantly contributed to the model fit. For each outcome variable, predictor variables included in the regression model were checked for multicollinearity. Given the relatively high number of regression analyses presented in this paper, there is a possibility of some spurious regression models, and therefore caution needs to be taken with p values close to 0.5.

Results

General Levels of Trust

Table 1 outlines the proportions of respondents indicating ‘trust a little’ or ‘trust a lot’ (called ‘higher trust’ throughout the paper) in each of the countries. The general trend in Australia was for respondents to report higher levels of trust than in other countries. Across all categories of groups and individuals, 81% of respondents indicated higher trust. Family was the group which respondents indicated a higher level of trust (95%), followed by neighbours (86%), foreigners (85%) and people with a different religion (84%). Only 57% of respondents indicated higher trust in strangers. Overall, we may categorise Australia as a ‘high trust society’.

Download:

Table 1. Proportion of respondents indicating higher trust in various groups or individuals.

https://doi.org/10.1371/journal.pone.0095555.t001

Hong Kong had the second highest levels of trust, with 63% of respondents indicating higher trust across the various groups and individuals. Family received higher trust by 99% of respondents, followed by neighbours (75%), people with a different religion (65%), and foreigners (59%). Strangers were the only group in Hong Kong with relatively low trust, with only 17% of respondents indicating higher trust in this group. In Taiwan, the overall level of trust across the groups and individuals was 60%, making it similar to Hong Kong. The group most frequently trusted in Taiwan was family (99%), followed by neighbours (77%) and people with a different religion (62%). Trust was much lower for foreigners (44%) and even lower for strangers (17%). Other than trust in foreigners, which was higher in Hong Kong than Taiwan (50% and 44% respectively), Hong Kong and Taiwan had similar levels of trust across the groups and individuals, and relative to other countries, we may therefore categorise them as ‘medium level trust societies’, a categorization not identified in Fukuyama’s thesis.

Trust in Japan was lower than the previous three countries (54%). Whereas family (98%) and neighbours (80%) were highly trusted, lower trust was found in people with a different religion (30%) and foreigners (28%). However, trust in strangers (33%) was higher than in either Hong Kong or Taiwan, which both had 17% of respondents indicating higher trust. In South Korea a pattern similar to that observed for Japan was found, and a similar overall level of trust across the individuals and groups (51%). There were high levels of trust in family (99%) and neighbours (79%), but lower levels of trust in people with a different religion (38%), foreigners (24%) and strangers (17%). Again, trust in strangers is the outlier here, with the very low levels of trust in South Korea being similar to Hong Kong and Taiwan, but much lower than Japan. Thailand had the lowest overall level of trust across the individuals and groups, with 49% of respondents indicating higher trust, making it similar to Japan and South Korea (54% and 51% respectively). As with all other countries, trust was highest for family (99%) followed by neighbours (92%). However, there was much less trust in people with a different religion (23%), foreigners (21%) and strangers (10%). Compared to Japan and South Korea, Thailand had higher trust in neighbours, but much lower trust in people with a different religion and strangers. Notwithstanding these differences, we categorise Japan, South Korea and Thailand as ‘lower trust societies’.

In order to further analyse the similarities and differences outlined above, Table 2 presents the Chi-square output and z-scores for each of the countries across each of the individuals and groups. In terms of trust in neighbours (χ² (5, 5941) = 145.1, p<.001), Australia and Thailand had significantly fewer respondents indicating low trust (z = −2.8 and z = −8.4, respectively), whereas in all other countries z-scores with a magnitude of 2.0 or over were observed, indicating higher numbers of respondents having lower trust in neighbours. Overall, Australia and Thailand had lower levels of trust in neighbours significantly less frequently than the other countries.

Download:

Table 2. Chi-square and z-scores for high and low trust across the countries.

https://doi.org/10.1371/journal.pone.0095555.t002

Examining z-scores for the associations for medium/high trust in strangers (χ² (5, 5845) = 747.51, p<.001), cell counts were significantly higher than expected in Australia (z = 19.5) and Japan (z = 5.8), but significantly lower than expected in Hong Kong (z = −3.8), South Korea (z = −4.7), Taiwan (z = −5.1) and Thailand (z = −9.6). In other words, Australia and Japan had higher levels of trust in strangers significantly more frequently than the other countries.

In terms of trust in foreigners (χ² (5, 5424) = 1102.40, p<.001), significantly higher than expected cell counts for medium/high level of trust in foreigners were observed for Australia (z = 19.6) and Hong Kong (z = 7.1), whilst significantly lower than expected cell counts were obtained for Japan (z = −6.3), South Korea (z = −7.5) and Thailand (z = −10.8). Observed cell counts in Taiwan did not contribute to the association. Therefore, Australia and Hong Kong had higher levels of trust in foreigners significantly more frequently than the other countries.

For the associations for trust in people with a different religion (χ² (5, 5406) = 1014.65, p<.001), cell counts were for medium/high trust were significantly higher than expected for Australia (z = 15.9), Hong Kong (z = 6.4) and Taiwan (z = 6.6), whereas for Japan (z = −8.1), South Korea (z = −4.1) and Thailand (z = −12.1) had significantly lower than expected observed cell frequencies. These analyses show that Australia, Hong Kong and Taiwan had higher levels of trust in people with a different religion significantly more frequently than the other countries.

Trust within Countries

The rest of the paper focuses on the regression models which examine the main sub-populations with either low or high trust in each country. This analysis extends the work of Fukuyama and facilitates appropriate policy responses to increase trust within particular groups in each country. Given the overwhelmingly high levels of trust in the family, it was not possible to generate regression models due to the lack of variability. Therefore, we only present data on trust in neighbours, people of a different religion, strangers and foreigners. Within the following sections, we use the term higher trust to refer to people who responded to either ‘trust them completely’ or ‘trust them a little’ and the term lower trust to refer to people who responded to either ‘do not trust them very much’ or ‘do not trust them at all’.

Trust in Neighbours

Australia.

Overall, 86% of respondents had higher trust in neighbours. Four independent variables were associated with trust in neighbours: age, income, work status and subjective health (see Table 3). Younger people were much less likely to trust neighbours than older people (aged 60 years or over), with people aged less than 20 years being over 99% less likely to trust (OR 0.01, 95% CI 0.00–0.05) and people aged 20–29 and 30–39 being less than 90% likely to trust neighbours (OR 0.06, 95% CI 0.02–0.25; and OR 0.08, 95% CI 0.02–0.31 respectively). Compared to people working full-time, part-time employed people were over 5 times more likely to trust neighbours (OR 5.39, 95% CI 2.16–13.47) and people working without pay were over 2.5 times more likely to trust neighbours (OR 2.62, 95% CI 1.14–6.03). Income was strongly associated with trust in neighbours, with people on higher incomes having higher trust. Compared to people earning less than $30,000, people earning either $60,000-$89,999 or $90,000-$119,999 were around 3 times more likely to trust neighbours (0R 2.87, 95% CI 1.20–6.88, and OR 3.14, 95% CI 1.12–8.82 respectively) and people earning $120,000-$149,999 were over 6 times more likely to trust (OR 6.20, 95% CI 1.79–21.45). People with worse subjective health were approximately 50% less likely to trust neighbours than people with better subjective health (OR 0.47, 95% CI 0.28–0.79).

Download:

Table 3. Regression model for trust in neighbours in Australia.

https://doi.org/10.1371/journal.pone.0095555.t003

Hong Kong.

All predictors were entered into the regression model but it was not significant and therefore no model is presented here.

Japan.

In Japan, 79.5% of respondents had higher trust in neighbours. Four independent variables were associated with trust in neighbours: age, income, marital status and subjective health (see Table 4). Compared to individuals aged 60 years and above, all other (younger) age groups were significantly less likely to indicate higher trust in neighbours. Indeed, as age group gets younger, trust in neighbours becomes lower in a step-wise fashion, with people aged 50–59 being 59% less likely to trust (OR 0.41, 95% CI 0.24–0.72), people aged 40–49 being 63% less likely to trust (OR 0.37, 95% CI 0.21–0.67), people aged 30–39 being 73% less likely to trust (OR 0.23, 95% CI 0.13–0.39) and people aged 20–29 being 87% less likely to trust neighbours (OR 0.13, 95% CI 0.07–0.24). Compared to married/cohabiting respondents, people who were separated, divorced or widowed were almost 50% less likely to report higher trust in neighbours (OR 0.51, 95% CI 0.30–0.87). In terms of income, people in the highest income group were over 70% more likely to have higher trust in neighbours than people in the lowest income group (OR 1.71, 95% CI 1.04–2.81). People with worse subjective health were approximately 40% less likely to trust neighbours than people with better subjective health (OR 0.63, 95% CI 0.43–0.90).

Download:

Table 4. Regression model for trust in neighbours in Japan.

https://doi.org/10.1371/journal.pone.0095555.t004

South Korea.

In South Korea, 79% reported higher trust in neighbours. Two independent variables were associated with trust in neighbours: age and subjective health (see Table 5). Respondents under the age of 20, between 20–29 and between 30–39 years were significantly less likely to have higher trust in neighbours than people aged 60 years or over (OR 0.13, 95% CI 0.06–0.30; OR 0.30, 95% CI 0.17–0.51; OR 0.53, 95% CI 0.31–0.91 respectively). People with worse subjective health were almost 50% less likely to trust neighbours than people with better subjective health (OR 0.51, 95% CI 0.37–0.72).

Download:

Table 5. Regression model for trust in neighbours in South Korea.

https://doi.org/10.1371/journal.pone.0095555.t005

Taiwan.

In Taiwan, 77% of respondents reported high trust in neighbours. Three independent variables were associated with trust in neighbours: work status, marital status and income (see Table 6). Compared to married or cohabitating, respondents who were separated/divorced/widowed or never married were around 50% less likely to report higher trust in neighbours (OR 0.55, 95% CI 0.33–0.93; OR 0.44, 95% CI 0.31–0.63 respectively). Compared to people reporting full-time employment, people in part-time employment were 50% less likely have higher trust in neighbours (OR 0.49, 95% CI 0.28–0.86) and retired people were twice as likely to report higher trust (OR 2.02, 95% CI 1.05–3.89). In terms of income, people in the highest income group were over 50% more likely to have higher trust in neighbours than people in the lowest income group (OR 1.52, 95% CI 1.04–2.22).

Download:

Table 6. Regression model for trust in neighbours in Taiwan.

https://doi.org/10.1371/journal.pone.0095555.t006

Thailand.

In Thailand, 92.4% of respondents had higher trust in neighbours. Only one independent variable was included in the regression model, namely subjective health status (see Table 7). People with worse subjective health were almost 50% less likely to trust neighbours than people with better subjective health (OR 0.49, 95% CI 0.32–0.76).

Download:

Table 7. Regression model for trust in neighbours in Thailand.

https://doi.org/10.1371/journal.pone.0095555.t007