Introduction

Democratic policing, characterized by accountability, transparency, and respect for human rights within a democratic society, ensures that law enforcement agencies adhere to democratic values such as the rule of law, protection of civil liberties, and maintenance of order for the public (Bradford & Quinton, 2014). This approach aims to establish and maintain policing practices consistent with democratic principles. Conversely, public confidence in the police, indicative of trust, belief, and perception in the effectiveness, fairness, and integrity of the police force (Cao, 2015), is a set of attitudes toward the police as an institution and a measure of the trustworthiness and satisfaction the public has with law enforcement agencies (Bradford & Jackson, 2010; Cao, 2015). It reflects the extent to which individuals believe the police can maintain law and order impartially. Therefore, while democratic policing is a concept and approach to policing, public confidence in the police is an outcome or result of how well democratic policing principles are implemented and perceived by the public. In mature democracies, the quality of democratic policing serves as a pivotal indicator of professionalism, nonpartisanship, and efficacy within the framework of democratic governance, leading to long-standing public confidence in the police (Almond & Verba, 1963; Bradford & Quinton, 2014; Cao et al., 2012).

A bulk of research in the West has underscored the correlation between stable democratic societies and elevated levels of public confidence in police over the past decades (please refer to Cao, 2015 for further review), whereas the investigations into public confidence in the police have burgeoningly proliferated in Asian democratic societies (Cao et al., 2012; Goldsmith, 2005), including China (Sun et al., 2017; Wu & Liu, 2023), Japan (Kobayashi, 2013 for theoretical review; Tsushima & Hamai, 2015), South Korea (Jang & Hwang, 2014; Kwak & McNeeley, 2019; Lim & Kwak, 2022; Park et al., 2021), and Taiwan (Lai & Zhao, 2018; Sun et al., 2014; Wang & Sun, 2020). Additional studies have attempted to conduct this topic from a comparative perspective between some East Asian and Western societies (e.g., Boateng et al., 2016; Cao & Dai, 2006; Cao & Stack, 2005; Cao et al., 1998) and among those societies in East Asia (e.g., Jang and Hwang, 2022; Lai et al., 2010; Wu et al., 2012).

In the literature concerning the relationship between democracy and public confidence in policing in Asian societies, three notable limitations deserve scholars’ attention. First, Japan, South Korea, and Taiwan, as democratic societies in East Asia, underwent significant political, social, and economic transitions from World War II to the present day but have not conducted a comparative analysis in this respect (Jang & Hwang, 2022). Therefore, the similar political shifts (e.g., from imperial systems to democracy), societal values (e.g., collectivism and harmony), cultural influences (e.g., Confucian philosophy), and religious backgrounds (e.g., Buddhism) in these three democratic societies underscore the necessity of investigating public confidence in the police separately from both Western societies and authoritarian regimes.

Second, Almond and Verba (1963) posited a positive association between public confidence in the police and the level of democracy in a nation (see also, Cao et al., 2012). In line with this argument, based on the Economist Intelligence Unit's 2022 Democracy Index (2023), while these three countries have transitioned to democracy, public confidence in the police should follow a sequence of Taiwan (ranking 10th on full democracy index), Japan (ranking 16th on full democracy index), and South Korea (ranking 24th on full democracy index). Moreover, using data drawn from the 2017–2020 World Values Surveys (WVSs), Jang and Hwang (2022) found that Taiwanese reported relatively higher confidence in the police, followed by Japanese and South Koreans, respectively. It is important to verify whether this pattern of public confidence in the police among these three East Asian democratic societies holds true.

Third, the extant studies examining the factors influencing the public assessment of police relied upon data from the World Values Surveys (WVS) (e.g., Boateng et al., 2016; Cao & Dai, 2006; Cao et al., 1998; Lai et al., 2010). While the multiple iterations of the WVS have permitted a plethora of informative comparative studies for three decades (now in its 7th iteration), it does not feature a specific focus on public confidence attitude formation in democratic policing. As a result, it is not possible to use WVS data to assess the theoretical frameworks typically featured in the policing literature pertaining to citizen confidence in the police. It is necessary, therefore, to collect original data on this question as opposed to placing continued reliance upon a secondary analysis of WVS data.

Four, many studies examining public confidence in the police employed one or sometimes two of three leading models (i.e., instrumental, expressive, or normative models) as theoretical frameworks for explaining the empirical relationship in an individual society (e.g., Bradford & Myhill, 2015; Jackson & Bradford, 2009; Jang & Hwang, 2014; Lim & Kwak, 2022; Park et al., 2021; Sargeant & Kochel, 2018; Sun et al., 2014); unfortunately, very few studies have assessed the three models as predictors in the explanation of public confidence in the police while conducting a comparative study. Hence, how those theoretical models and their affiliated factors impact public confidence in the police in these three East Asian democratic societies remains unknown.

Relying on convenience sampling-based online survey data collected across these three countries in 2022, this study examines the following: (1) the extent of confidence Japanese, South Koreans, and Taiwanese citizens have in their police; specifically, who is more likely to report the highest levels of confidence in their police; (2) how instrumental, expressive, and normative models influence citizens of these three democratic societies and of each country after controlling for demographics; and (3) which factors drawn from these three theoretical models impact all three societies' citizens and each country's citizens, or uniquely influence citizens within a specific country.

Historical Overview and Literature Review

Compared with many Western societies, East Asian societies tend to be homogeneous—that is, they are composed of one dominant race, using one language, and carrying out public affairs by one set of broadly shared collective norms (Jang & Hwang, 2022). In this regard, East Asian communities have been strongly influenced by Confucianism for more than two thousand years, a philosophical orientation emphasizing the prioritization of collective interests, respect for authority, hierarchical social relations, and reliance upon informal social control forces to promote a harmonious society (Cao et al., 1998; Jang & Hwang, 2022; Woo et al., 2018). While the degree of influence of Confucianism varies from country to country, all of East Asia reflects the strong presence of Confucianism. In accordance with the Confucian historical legacy, individuals are required to tolerate and sacrifice his/her values of personal autonomy, and, instead be compliant with the directives of hierarchical authorities, respect the elderly, have trust in others, and develop a sense of shared destiny with others in one’s own society (Cao & Hou, 2001; Cao et al., 1998; Hwang et al., 2005; Jang & Hwang, 2022; Komiya, 1999; Woo et al., 2018). More specifically, while informal social controls play a significant role in maintaining an ordered society, the police play a pivotal role in a Confucian society as the guardians of societal harmony (Cao et al., 2015). Accordingly, East Asian countries have uniformly established and maintained a highly centralized police structure.

Some scholars have suggested that, as a general rule, people in East Asian societies have been taught to respect and obey police directives at home, in schools, and in the workplace (Boateng et al., 2016; Cao et al., 1998; Lai et al., 2010). Moreover, among the three societies selected for a more detailed comparison, Japan began its peaceful yet slow transition to democratic governance as early as the 1950s, whereas the process of democratic transition for South Korea and Taiwan dates from the 1980s (Cao & Dai, 2006; Cao et al., 1998). Drawing from the work of Almond and Verba (1963) and Cao et al. (2012), arguing that public confidence in the police is highly correlated with the degree of democracy present in a nation, we anticipate that citizens in East Asia will express relatively high levels of confidence in their respective country’s police force.

Although these three East Asian countries share many similarities, the differences in their developmental paths to democracy may contribute to variations in citizens’ reported levels of confidence in their respective police forces.

First, before 1945, Japan was a monarchy, and its police were tasked with tight control over civilians in service to the Royal Emperor, employing various means to achieve ubiquitous surveillance (Kushner, 2006). Specifically, the Legal Authority of the police to control and manage the residents, leading to perceptions of police as “ubiquitous, officious, and arrogant” (Perry, 1980: 147).

During the US occupation of Japan (1945–1952), significant changes occurred. The monarchy was redefined, with the emperor symbolized as the spiritual leader rather than the head of state, and governance shifted to a cabinet system based on popular voting (Aldous & Leishman, 1997). The police system underwent extensive reform under US influence, including democratization training, decentralization of command, emphasis on citizen rights protection, and building relationships with communities. A new police law was enacted in 1947, followed by significant amendments in 1954, which re-centralized prefectural police departments under the National Public Safety Commission’s control (Chwialkowski, 1998).

Community-Oriented Policing (COP) was introduced in Japan in the 1980s to strengthen the rapport between police and neighborhoods (Bayley, 1991). Utilizing urban Kobans and rural Chuzaishos, police officers stationed in these substations maintain close ties with their assigned communities. They conduct regular door-to-door visits, providing crime prevention tips and addressing safety concerns, while also engaging in service-oriented activities such as distributing information and assisting with directions and lost items (Suzuki et al., 2023). Scholars attribute Japan’s low crime rate since the 1980s to the strong relationships between police, community leaders, and residents, fostering high levels of public cooperation (Ames, 1981; Bayley, 1991; Komiya, 1999; Parker, 1984).

Despite expectations of high confidence in the police, a decline in trust has been observed. A 2008 survey by Yomiuri Shimbun revealed a decrease from 87% in 1995 to 65% in 2004 (Tsushima & Hamai, 2015). Public trust further eroded due to widely publicized police scandals and rising crime rates in the early 2000s (Tsushima & Hamai, 2015). In response, the National Public Safety Commission and the National Police Agency formulated the “Guideline for Police Reform” in 2000, aiming to improve information disclosure, handle complaints of police misconduct, enhance consultation functions, increase police officer identifiability, and establish police station advisory councils (Yoshida & Leishman, 2006). Since then, Japanese police have sought to regain public confidence by adhering closely to these reform guidelines, highlighting the importance of understanding current perceptions of the police among Japanese citizens.

Second, during Japan’s 35-year colonization of the Korean Peninsula (1910–1945), the police were tasked with maintaining public order, suppressing religion and liberation movements, and expelling political opponents and independent fighters (Woo et al., 2018). After World War II, South Korea (the Republic of Korea) was established in 1948 and soon faced a national war initiated by North Korea in 1950. The war ended in 1953 with an armistice involving US and UN forces assisting South Korea. However, South Korea’s governance from 1948 to 1993 faced criticism for being illegitimate, coercive, and undemocratic, led by military personnel and dictators who seized power through coups and election fraud. Consequently, the police, prosecutors, and courts were politicized, corrupted, and used to suppress citizens, particularly students involved in anti-governmental activities. Throughout the 1970s and 1980s, amidst economic growth, social movements and riots were frequent, with police responding with violence, brutality, and deadly force, leading to citizens’ distrust in law enforcement (Moon et al., 2005; Woo et al., 2018).

In 1993, South Korea established its first civilian government, marking the initial steps toward democratization. During this period, the police-initiated reforms to improve their relationship with citizens and enhance their public image (Moon, 2004). Subsequently, from 1998 to 2003, the government implemented a “Police Grand Reform” aimed at increasing citizen support and trust in law enforcement (Moon et al., 2005). Despite these efforts, negative attitudes toward the police persist among many South Koreans, rooted in historical tensions and mutual suspicion between law enforcement and the public (Moon & Morash, 2008; Woo et al., 2018). Research consistently shows that South Koreans have lower levels of trust and confidence in the police compared to citizens in developed countries (Boateng et al., 2016; Cao & Dai, 2006; Jang, 2014).

Third, Taiwan (the Republic of China in Taiwan), separated from Mainland China in 1949, emerged as a de facto independent political entity after its government’s defeat in the Chinese civil war. The Nationalist Government (Koumintang or KMT) relocated its party from China to Taiwan, and Chiang Kai-shek resumed his presidency of the Republic of China (ROC) in 1950. Since then, two separate and independent entities have faced each other across the Taiwan Strait (Lai et al., 2010). Over the past seven decades, Taiwan has navigated a challenging journey toward democracy. Three waves of political reform shaped this transition, including a period of marital law (1950–1988), a period of widespread civil democratization (1988–2000), and ultimately consistent peaceful transition of presidential power (2000–2020) (Jang & Hwang, 2022; Lai et al., 2010).

During this transition, the role of the police in Taiwan underwent a complete transformation. They evolved from being an extension of military forces, focused on control and surveillance, to practicing democratic policing. This shift emphasized order maintenance, protection of citizens’ rights, and a wide range of services for the community (Cao & Dai, 2006; Lai & Zhao, 2018; Lai et al., 2010). The introduction of Community-Oriented Policing (COP) strategies further enhanced professional police services, fostering a mission of “connecting to the local community” since the early 1990s (Cao et al., 2014).

Concrete COP strategies in Taiwan encompass various initiatives: police-led charitable activities, procedural justice improvements in police-citizen interactions, digital communication via social media, recruitment of volunteers for community patrol teams, promotion of citizen cooperation in preventing disorders, and establishment of citizen advisory councils (Chang et al., 2011). Despite facing police scandals, corruption, and student-led movements, Taiwanese confidence in the police has remained strong over the past two decades (Lai & Zhao, 2018; Sun et al., 2014). A recent poll by the Crime Research Center at National Chung Cheng University found that 84% of respondents expressed satisfaction with police performance in maintaining order and safety (Yang et al., 2023).

Overall, three observations should be highlighted in this brief historical overview section. First, while Japan was the first among the three nations to become a democracy in the 1950s (Chwialkowski, 1998), both South Korea and Taiwan experienced a challenging political transition from authoritarianism to democracy over the period from 1950s to the 1980s (Cao & Dai, 2006). In this regard, the first direct presidential election in South Korea took place in 1988. In Taiwan, Martial law was officially lifted in 1987 permitting nationwide elections to take place (Lai et al., 2010). Today, according to the Economist Intelligence Unit's 2022 Democracy Index (2023), the three East Asian societies are classified as full democracies among the 167 countries and entities surveyed.

Hypothesis 1. All respondents in these three East Asian societies, as well as respondents within each country, report relatively higher levels of confidence in the police.

Second, surveying the levels of public attitudes toward police is an essential task for all levels of professional police administrators in democratic societies (Cao et al., 2012). Lai et al. (2010) have noted that public confidence in the police may suffer in the short run during the democratization process, but in the long run confidence in the police increases as democratic policing becomes widespread. According to the Economist Intelligence Unit's 2022 Democracy Index (2023), among the 165 countries and entities surveyed, Taiwan ranks #10, followed by Japan at #16, and South Korea at #24. Meanwhile, using data from the WVS wave fielded in 2017–2020, Jang and Hwang (2022) found that Taiwanese and Japanese respondents expressed higher levels of confidence in the police (with 82.5% and 78.6%, respectively) than did South Korean citizens (with 56.6%).

Hypothesis 2. The ranking of the public confidence in the police would be highest in Taiwan, followed by Japan, with South Korea exhibiting the lowest level.

Theoretical Models of Confidence in the Police

The extant studies examining public attitudes toward the police is generally conceived of within three leading theoretical models in the exploration of empirical determinants: instrumental model, expressive model, and normative model (Lim & Kwak, 2022; Park et al., 2021; Sargeant & Kochel, 2018). Notably, to the best of our knowledge, this is the first study to integrate three theoretical models simultaneously to examine the predictive potential of public confidence attitudes toward the police in these three East Asian democracies. These three models are briefly discussed below.

The instrumental model focuses on tangible outcomes and instrumental benefits citizens derive from law enforcement actions, linking confidence in the police to perceptions of their effectiveness in achieving specific goals, such as preventing victimization and reducing crime incidents (Jackson & Bradford, 2009; Jang & Hwang, 2014; Park et al., 2021; Sun et al., 2014). If perceived as ineffective, public support and confidence in the police decrease (Lim & Kwak, 2022). Sunshine and Tyler (2003) argue that public acceptance of police is tied to their ability to create credible sanctioning threats for rule-breakers (risk) and effectively control crime (performance). Factors affecting confidence in the police include crime rates, victimization experiences, fear of crime, public safety perceptions, likelihood of sanctions for deviant behaviors, and police effectiveness (Cao et al., 2012; Lai & Zhao, 2018; Pryce, 2019; Tsushima & Hamai, 2015). These factors impact confidence across various countries, including the UK, South Korea, Taiwan, Japan, Jamaica, and the US (Jackson & Bradford, 2009; Jang & Hwang, 2014; Lim & Kwak, 2022; Park et al., 2021; Pryce & Grant, 2021; Sun et al., 2014; Tsushima & Hamai, 2015). Based on the existing research findings, we hypothesize the following:

Hypothesis 3. Worry about crime produces a significant negative influence on confidence in the police in all respondents across three surveyed countries.

Hypothesis 4. The possibility of being sanctioned for some deviant behaviors produces a significant positive influence on confidence in the police in all respondents across three surveyed countries.

Hypothesis 5. The perceived police effectiveness in neighborhoods produces a significant positive influence on confidence in the police in all respondents across three surveyed countries.

The expressive model focuses on assessing cohesion, social control, civility, and moral consensus, reflecting concerns about societal breakdown. When individuals perceive their community’s moral order as intact, they feel less fear of crime and have greater confidence in the police (Jackson & Bradford, 2009). Jackson and Sunshine (2007) suggest that the police serve as guardians of social order, symbolically representing the community. This symbolic role is reflected when citizens, feeling they reside in a “good neighborhood,” exhibit higher confidence in the police (Sargeant & Kochel, 2018). Studies have used police visibility and residential stability as indicators to gauge confidence in the police within this model (e.g., Kwak & McNeeley, 2019; Sindall et al., 2017; Tsushima & Hamai, 2015; Wang & Sun, 2020). Notably, while citizens perceive police effectiveness as a tangible outcome, police visibility underscores their symbolic role as guardians of social order in communities. Based on the previous studies, we hypothesize the following:

Hypothesis 6. Feeling of safety produces a significant negative influence on confidence in the police in all respondents across three surveyed countries.

Hypothesis 7. The perceived police visibility in neighborhoods produces a significant positive influence on confidence in the police in all respondents across three surveyed countries.

Hypothesis 8. The residential stability produces a significant positive impact on confidence in the police in all respondents across three surveyed countries.

The normative concern underscores how citizens’ confidence in the police is influenced by the perceived fairness of police conduct (Gau, 2014; Tankebe, 2013). Tyler (2006) presents two main elements of procedural justice: quality of decisions making and quality of interpersonal treatment. Specifically, when individuals believe that the police treat them with good quality—exhibiting traits such as fairness, respect, courtesy, and attentiveness to citizens’ concerns—they are more likely to trust the police (Murphy et al., 2014). This trust stems from both direct encounters with police and indirect experiences via media consumption or social interactions, shaping perceptions of how others are treated. In existing research procedural justice is operationalized as the way in which the police treat citizens and the “fairness” of the decisions made in police-citizen encounters (Reisig et al., 2007; Sunshine & Tyler, 2003), whereas distributive justice refers to perceptions regarding the “fairness” of the distribution of police services and activities between different communities, groups, and individuals (Jonathan-Zamir and Weisburd, 2009: 7). Put differently, procedural justice and distributive justice are key components of this model, with Tyler’s work emphasizing procedural justice’s role in shaping attitudes toward legal authorities. Recent studies in Pan-Asian societies confirming its positive impact on the public trust in the police (Lim & Kwak, 2022; Park et al., 2021; Sargeant & Kochel, 2018; Woo et al., 2018).

Distributive justice, often considered under the instrumental model (Sunshine & Tyler, 2003), is here incorporated into the normative model due to its integral role in discussions of fairness and legitimacy in policing. Tyler and Huo (2002) stress the importance of fair outcome distribution in shaping perceptions of authority legitimacy, while Skogan (2006) highlights fair treatment and resource allocation’s role in fostering trust. Despite previous emphasis on distributive justice’s impact on police legitimacy (e.g., Bradford & Jackson, 2008; Sunshine & Tyler, 2003; Tyler and Huo, 2002, recent studies like Pryce and Grant’s (2021) findings in Jamaica underscore its significant influence on police trust (also see Nix et al., 2015). Based on the existing research findings, we hypothesize the following:

Hypothesis 9. The perceived procedural justice produces a significant positive influence on confidence in the police in all respondents across three surveyed countries.

Hypothesis 10. The perceived distributive justice produces a significant positive influence on confidence in the police in all respondents across three surveyed countries.

Methods

Research Setting

The research reported upon here was conducted during the period ranging from early January to late February in 2022. Among the three countries, Japan is the largest with 124.95 million in population in 2022 living in a geographic territory of 377,962 square kilometers. The population density is about 338 people per square kilometer (Statistics Bureau of Japan, 2023). The Japanese police system is centralized, and the headquarters of the police system is the National Police Agency (JNPA) in Tokyo overseen by the National Public Safety Commission. In addition, the 47 prefectural police headquarters have responsibility for all regular police duties in their geographic area of responsibility. Currently, the Japanese police system maintains 6000 urban Kobans and more than 6000 Chuzaishos staffed by 296,400 police personnel, among whom 262,500 were commissioned police officers in 2020. The citizen/police ratio is 477, the highest ratio among the three societies under study. Japan has a very low rate of crime; the total crime rate was 487 cases per 100,000 habitants in 2021 (Statista, 2023).

The population of South Korea is approximately 51,611,400, and the geographic area is about 100,413 square kilometers. The population density is about 507 people per square kilometer. The South Korean police system is highly centralized as well, and the headquarters is the Korean National Police Agency (KNPA) located in Seoul. The headquarters directly commands all local police organizations and special units nationwide. Currently, the KNPA is regionally subdivided, covering 18 metropolitan cities and provinces, maintaining 255 stations, 518 precincts and 1433 police boxes under a network of metropolitan and provincial police agencies. In 2020, 126,227 police officers were deployed, constituting a citizen/police ratio of 409. In 2021, the crime rate was 1774 cases per 100,000 habitants, which was down 12% from a year earlier (Statista, 2023).

Finally, the population of Taiwan is about 23,109,034, and the territory is about 35,887 square kilometers. The population density is about 644 people per square kilometer, which is the highest of the three countries. The police system in Taiwan is highly centralized as well, and headquarters is Taiwan National Police Agency (TNPA) located in Taipei. As of 2022, the TNPA employs approximately 73,700 law enforcement personnel (among them 70,000 police officers) and operates 20 departments, 160 precincts, and 1600 police field stations (including sub-district stations) located in 20 cities and counties (TNPA statistics, 2023). The police/citizen ratio was 330, the lowest among the three countries. In 2021, a total of 1036 crime cases per 100,000 inhabitants were reported in Taiwan (TNPA statistics, 2022). Table 1 displays some demographic, police resource, and crime statistics associated with each of three research sites.

Table 1 Demographics and police resource and crime statistics
Full size table

Sampling Procedure

The survey data employed here were derived partly from a major five-nation study of citizen attitudes toward police in Japan, Taiwan, South Korea, the UK, and the USA. This project involved the collection of survey data from 10,000 respondents. The five-nation study project was approved by the Research Ethics Committee (IRB) at a Japanese University in Japan. The survey questionnaire was developed by participating scholars from Japan, Taiwan, and the UK in an online teleconference in the winter of 2021. As is customary in such survey work, attitude scales from previous published studies were integrated into the survey questionnaire. The consensually agreed upon final study questionnaire was then put through a process of double-back translation so that equivalent versions were available in Japanese, Chinese, Korean, and English. A pre-test was conducted, and some minor revisions were carried out based upon pre-test feedback.

This particular study (one of several ongoing) utilizes survey data collected in Japan, Taiwan, and South Korea during the first two months of 2022. Respondents for the survey were recruited from Internet users between the ages of 20 and 69 who registered with LUCID, a marketing research firm that supplied samples to the researchers. The researchers used the quota method to collect up to the same number of respondents stratified by sex and 10-year age group in each of Japan, Korea, and Taiwan. However, due to the small number of monitors supplied by the company, in Taiwan, the number of respondents aged 60 or older was 7.6% of all respondents.Footnote 1

The population ratio of Japan, South Korea, and Taiwan is approximately 1:0.4:0.2. In this study, we oversampled respondents in South Korea and Taiwan up to the upper limit of sample size that LUCID could provide, resulting in the sample size of 2314 in Japan, 1622 in South Korea, and 1370 in Taiwan (1:0.7:0.6).

The survey was conducted via Qualtrics. At the beginning of the survey, respondents were asked to answer several personal background questions pertaining to income, educational background, and other characteristics necessary for appropriate data analysis, followed by a three-question Directed Questions Scale to eliminate respondents who lacked sufficient interest in the subject matter of policing. After data screening and the removal of some incomplete or otherwise deficient questionnaires, a total of 947 respondents in Japan, 750 respondents in South Korea, and 551 respondents in Taiwan are available for analysis; this represents an estimated response rate of 42.4%.

Measures

Dependent Variable

The dependent variable of confidence in the police was measured by a single item—namely, “To what extent do you have confidence in the police?” Responses range from 1 = none at all, 2 = not very much, 3 = neither, 4 = quite a lot, and 5 = a great deal. The dependent variable represents an ordered categorical variable. Ordinal logistic regression constitutes the most appropriate type of statistical analysis when dependent variables are of the ordered categorical type (Lai et al., 2010; Wu & Sun, 2009). While use of a single item to measure confidence in the police is an acknowledged weakness, this measure’s use in previous studies indicates that it is a stable and reliable indicator of citizen sentiment. Moreover, single-item studies are particularly appropriate when conducting comparative studies within different language settings where equivalence of meaning issues dictate use of tried-and-true measures of proven utility in cross-cultural applications (Cao & Hou, 2001; Frank et al., 1996; Jang et al., 2010).

Instrumental Variables

Three variables were included in the instrumental model to capture the levels of citizen perceptions of police function and services in a local community—namely, worry about crime, likelihood of sanctions, and police effectiveness. The worry about crime is measured by four items. Respondents were asked how much they worried about being a victim of the following set of crimes over the course of the past year: (1) property crimes (theft, burglary, or fraud), violent crimes (robbery, assault, or attack), traffic accidents, and terrorism. Response options for the four items were 1 = not at all, 2 = not so worried, 3 = neither, 4 = worried, and 5 = very worried. The worry about crime scale has a robust Cronbach’s α of 0.86, with an eigenvalue of 2.825.

Additionally, the likelihood of sanctions was a scale featuring three items prefaced by this lead-in: “What is the likelihood of being sanctioned for engaging in any of the following behaviors?” (1) Make an exaggerated or false insurance claim, (2) buy something you thought might be stolen, and (3) commit a traffic offense such as speeding or running a red light. The responses range from 1 = not probable at all to 4 = highly probable. Higher scores on the scale indicated that respondents perceive a higher possibility of being sanctioned. The reliability of the scale has a modest Cronbach’s α of 0.68, with an eigenvalue of 1.863.

The police effectiveness was measured by five items as follows: (1) the police are successful at preventing crimes where violence is used or threatened; (2) the police are successful at catching people who commit house burglaries; (3) the police respond to calls for service quickly; (4) the police give good advice to the public; and (5) the police maintain a good relationship with the public. Response options for the five items were arrayed along a Likert-type continuum ranging from 1 = strongly disagree to 5 = strongly agree. The police effectiveness scale combining the five items has a robust Cronbach’s α of 0.86, with an eigenvalue of 3.226.

Expressive Variables

Four variables are used to represent the expressive model. The feeling of safety is measure by two items. Respondents were asked how safe he/she felt while in their neighborhood and in the whole country over the past year. Response options for the two items were 1 = very unsafe, 2 = unsafe, 3 = neither, 4 = safe, and 5 = very safe. Two survey items were used to measure police visibility: (1) how often do you see the police officers on foot or bike patrol; (2) how often do you see the police patrolling in cars? Response options for the two frequency of observation items were 1 = about once a month or less, 2 = several times a month, 3 = about once a week, 4 = about once a day, and 5 = more than once a day. Drawing from Lee et al. (2022) work, two single items as two variables are employed to capture the capture of residential stability, namely, house ownership and length of residence. House ownership was a dummy variable where 0 = renter and 1 = owner. The length of residence was represented as an ordinal variable by asking the respondents, “How long have you been living at your current residence?” Response options for this item were (1) less than 1 year, (2) 2–5 years, (3) 6–10 years, (4) 11–20 years, and (5) 21 years or more.Footnote 2

Normative Variables

Two variables are used to represent the normative model. Four survey items were employed to capture the concept of procedural justice being present in police-citizen encounters: (1) the police treat people fairly and equally, (2) the police are honest, (3) the police generally treat people with respect, and (4) when dealing with people, the police generally explain their decisions and actions when asked to do so. Likert scale-like response options of the four items were arrayed along a continuum ranging from 1 = strongly disagree to 5 = strongly agree. The scale combining the four items has a robust 0.90 Cronbach’s α, with an eigenvalue of 3.055.

Four survey items were used to capture the concept of distributive justice. Survey participants in Japan, South Korea, and Taiwan were asked: “Do you think the police are equitable in their work in the following respects”: (1) a victim’s wealth/income; (2) a victim’s race or ethnicity, (3) a suspect’s wealth/income, and (4) a suspect’s race or ethnicity. Response options of the four items were 1 = treated unequally to 5 = treated equally. The reliability of distributive justice scale has a robust Cronbach’s α of 0.92, with an eigenvalue of 3.240.

Control Variables

Four variables were treated as conventional control variables. Gender was coded as a dummy variable where 0 = male and 1 = female. Age was captured as a five-category variable with age groupings of 1 = 20–29, 2 = 30–39, 3 = 40–49, 4 = 50–59, and 5 = 60 and above. Educational attainment is documented in a five-category variable, with 1 = primary school; 2 = secondary school up to 16 years; 3 = higher or secondary or further education (A-levels, BTEC, etc.); 4 = college or university; and 5 = post-graduate degree. Marital status was recoded as a dummy variable where 0 = single/divorced/widow and 1 = married and cohabitant. House ownership was a dummy variable where 0 = renter and 1 = owner.

To avoid severe multicollinearity problem, conventional diagnostic tests were performed on correlation matrices featuring these independent and control variables. Results from these diagnostic tests revealed no collinearity problems for any of the independent variables used in this study. All variance inflation factor [VIF] scores were well below the conventional standard of 4 (Tabachnick & Fidell, 1996).

Results

Mean Comparisons

Table 2 sets forth the results of a mean comparison of all variables included in this study. The mean score of confidence in the police was 3.37 with a standard deviation of 0.97, suggesting that respondents express a modest level of confidence in the police overall. Hypothesis 1 has been supported. Specifically, the mean score (3.45) of confidence in the police in Taiwan is significantly higher than other counterparts (F = 3.565*), followed by the mean score of 3.37 in Japan, and the mean score of 3.31 in South Korea (please refer to Fig. 1). This finding supports Hypothesis 2 that the ranking of the public confidence in the police would be highest in Taiwan, followed by Japan, with South Korea exhibiting the lowest level and confirms Jang and Hwang’s (2022) recent research.

Table 2 Mean comparisons and all study variables in three East Asian Nations (n = 2248)
Full size table
Fig. 1
figure 1

Mean scores of confidence in Police in Japan, South Korea, and Taiwan

Full size image

In terms of respondents’ demographics, males accounted for more than half of all respondents, with no difference between the three countries in this regard. The mean age category was 2.88, with Japanese and South Korean respondents being significantly older than their Taiwanese counterparts. The education level of South Koreans was the highest, followed by that of the Taiwanese and the Japanese. Most respondents were married or cohabitants, of which the number of married or cohabitant people in Japan and South Korea was significantly higher than that in Taiwan. While most respondents reported that they were homeowners, South Korean and Taiwanese respondents were considerably more numerous than was the case among the Japanese survey participants.

With respect to scale-type variables, while the respondents reported modest to robust levels of perceived worry about crime, feeling of safety, police visibility, police effectiveness, likelihood of sanctions, distributive justice, and procedural justice ranging from 2.39 to 3.66, Taiwanese respondents reported the highest levels on all of those seven perceived scales.

Multivariate Analyses

A set of ordinal logistic regression (OLR) models was estimated to find whether instrumental, expressive, and normative models were predictors of confidence in the police for all respondents, as well as across the three countries. Table 3 displays the results of the regression for all respondents and for respondents from each of the three countries. In Column 1, two variables—likelihood of sanctions and police effectiveness representing the instrumental model—were each significant and positive predictors of all respondents’ confidence in the police. At the same, house ownership and feeling of safety representing the expressive model produced significantly positive influences on all respondents’ confidence in the police. Additional, procedural justice and distributive justice representing the normative model significantly increased the levels of confidence in the police. Among the control variables, female individuals, young individuals, and married/cohabiting individuals reported significantly higher levels of confidence in the police. The most robust predictors of police confidence in the pooled sample were procedural justice, police effectiveness, and Taiwanese nationality. The pseudo-R square showed that the model explained 42.4% of the total variance in confidence in the police for the pooled, three-nation dataset.

Table 3 OLR results of confidence in the police in three East Asian democratic societies
Full size table

Column 2 displays the results of explanatory variables in predicting Japanese respondents’ confidence in the police. Among the instrumental variables, only police effectiveness and likelihood of sanctions produced significant and positive associations with confidence in the police. At the same time, drawing from expressive model, only house ownership produced significant and positive influence on confidence in the Japanese police. In addition, the procedural justice representing the normative model significantly increased the levels of confidence in the police. Females, among the control variables, reported higher levels of confidence in the Japanese police. The most robust predictors were procedural justice, police effectiveness, and female. The pseudo-R square showed that the model explained 44.0% of the total variance in confidence in the police.

Column 3 presents the results of explanatory variables in predicting South Korean respondents’ confidence in the police. Among instrumental variables, only police effectiveness produced a statistically significant positive association with confidence in the police. Meanwhile, house ownership representing the expressive model produced significantly positive influence on confidence in the police. In addition, procedural justice representing the normative model significantly increased the level of confidence in the police. Of note, the control variables produced insignificant influence on confidence in the police. The most robust predictors in South Korea were procedural justice, police effectiveness, and house ownership. The pseudo-R square showed that the model explained 42.4% of the total variance in confidence in the police in South Korea.

The final column sets forth the results for Taiwanese respondents. The police effectiveness representing the instrumental model was a statistically significant and positive predictor of confidence in the police. Additionally, feeling of safety and house ownership representing expressive model produced significant and positive influences on confidence in the police. Moreover, the procedural justice and distributive justice representing the normative model significantly increased the levels of confidence in the police among Taiwanese survey participants. Among the control variables, only married/cohabitating individuals significantly reported higher levels of confidence in the police. The most robust predictors were procedural justice, police effectiveness, and feeling of safety. The pseudo-R square showed that the model explained 42.5% of the total variance in confidence in the police.

Of note, among those explanatory variables, police effectiveness, house ownership, and procedural justice have significantly and positively increased the levels of confidence in the police across all respondents and all three countries. Put differently, Hypotheses 5, 7, and 9 have been supported by all respondents and the respondents in each surveyed country. Additionally, while the likelihood of sanctions solely predicted the confidence in the police among the Japanese respondents, feeling of safety and distributive justice significantly increased the levels of confidence in the police among those Taiwanese respondents. Namely, Hypothesis 4 has been confirmed by all respondents and Japan respondents, whereas Hypotheses 6 and 10 have been verified by all respondents and Taiwanese respondents. Hypotheses 3 and 7 failed to be supported in this study.

Discussion and Conclusion

Discussion

To extend our study of collective knowledge of residents’ attitudes toward the police, the current study investigated the public confidence in three East Asian democratic societies. The investigation entailed an inquiry into the extent to which factors drawn from instrumental, expressive, and normative analytical models influence public confidence in the police after controlling for conventional demographic variables. Noteworthy findings can be derived from three-nation inquiry.

First, these three East Asian democratic societies have long shared common values, such as Confucianism, collectivism, respect for authority, and an emphasis on harmony. They have also experienced a political transformation from authoritarianism to democracy after World War II, resulting in a shift in the role of the police from controlling people to serving them (Cao et al., 1998; Jang & Hwang, 2022). Consequently, similar to mature democratic societies in the West, respondents from these East Asian nations express higher levels of confidence in their police. Specifically, Taiwanese respondents reported the highest level of confidence in the police, followed by the Japanese and then the South Koreans. This finding aligns with existing research (Jang & Hwang, 2022). Over the past decade, the total crime incidents and crime rates for various types of offenses have dropped by more than 23% in Taiwan (Agency and [TNPA], 2023). Moreover, the evolutions in Taiwanese police departments should be highlighted.

Over the past three decades, Taiwanese police have prioritized enhancing police-civilian relations. For instance, during the police-citizen confrontations like the student-led Sunflower Movement in 2014, police have refined their enforcement techniques, becoming exemplars in training courses (Lai, 2016). They swiftly rectify procedural injustices and evidence mishandling in criminal investigations to ensure fairness. Additionally, when scandals like corruption or drunk driving arise, authorities promptly investigate or suspend implicated officers, earning public support. Consequently, the public’s satisfaction with police performance has increased, leading to significant confidence in the police (Yang et al., 2023). Sun et al. (2021) argue that Taiwan has transitioned into a “low power distance society,” marked by fully democratized policing principles of accountability, transparency, and human rights respect.

While the similar great efforts have been done to cultivate the police-citizen relations in other counterparts, Tsushima and Hamai (2015) noted that some police corruption scandals and accusations of brutality and higher crime rates have adversely impacted the levels of trust in the police among Japanese in the 2000s. Boateng et al. have also opined that South Koreans have relative less confidence in the police as compared to other East Asian societies at least in part because of the negative image of high policing developed during the Japanese occupation and authoritarian presidential periods before 1993. Although those many violations of human rights and police brutality were pervasive during the authoritarian regimes that occurred more than forty years ago (Pyo, 2001), research consistently reveals that such deeply negative historical experiences have remained implemented in the citizens’ collective memory (Moon, 2004; Moon and Morash, 2008; Woo et al., 2018).

Second, regarding the theoretical model, police effectiveness drawing from the instrumental model, house ownership representing the expressive model, and procedural justice representing the normative model each produced a significant impact on public confidence in the police, for all respondents combined and for each country. This finding is consistent with the extant research (Lim and Kwak, 2022; Park et al., 2021). Among the tree significant factors, police effectiveness had the strongest effect on public confidence in the police, which is consistent with the previous studies conducted regarding in West or in those non-Western societies (e.g., Jackson & Bradford, 2009; Lim & Kwak, 2022; Park et al., 2021; Pryce, 2019; Sunshine & Tyler, 2003). According to the instrumental theoretical framework, citizens’ judgement of police effectiveness depends on the capability of the police to fight crime, to reduce fear of crime, and to enhance safety and security in a neighborhood (Jackson & Bradford, 2009; Jang & Hwang, 2014; Park et al., 2021; Sun et al., 2014). Thus, the police should continuously commit themselves to implement their basic tasks of preventing and suppressing crime to win the citizens’ support they served (Lim & Kwak, 2022).

Compared to renters, homeowners were more likely to express confidence in the police among all respondents and across the three surveyed countries. This finding diverges somewhat from previous research. For instance, Lim and Kwak (2022) found in their study in South Korea that homeowners were less likely to trust the police (see also Pryce and Grant, 2021). Similarly, Lai and Zhao (2018), using data collected from metropolitan Taipei in Taiwan, discovered that the length of residence (closely related to homeownership) was negatively associated with satisfaction with the police. We speculate that homeowners who have resided at their current address for a long time (r = 0.29, p < 0.01), representing residential stability, are satisfied with police effectiveness (r = 0.42, p < 0.01), consequently leading to higher levels of confidence in the police (Kwak & McNeeley, 2019; Sindall et al., 2017; Wang & Sun, 2020). However, additional studies would be necessary to confirm the presence of these linkages. Heeding this finding, the police should make more efforts to build up the community or neighborhood as a “good quality residence” to gain support from the citizens they serve. Heeding this finding, the police should make more efforts to build up the community or neighborhood as a “good quality residence” to gain support from the citizens they serve (Sun et al., 2014).

Procedural justice, often considered a precursor to police legitimacy (Mazerolle et al., 2013), has been identified as a crucial factor determining confidence in the police in East Asian societies, consistent with previous studies in the West (Hinds & Murphy, 2007; Pryce, 2019; Sargeant and Kochel, 2018; Sunshine & Tyler, 2003; Tankebe, 2013). This finding supports the overarching thesis that public perceptions of procedural justice influence confidence in the police globally, including in non-Western societies (Lim & Kwak, 2022; Park et al., 2021; Woo et al., 2018). As noted by Tyler (2005), the fundamental components of procedural justice include voice, respect, neutrality, and trust. When police officers adhere to these norms in their interactions with citizens, compliance with police directives is more likely to follow. To enhance public confidence in the police, officers should prioritize procedural fairness and treat citizens with respect and patience. Furthermore, police should afford residents the opportunity to understand decision-making processes, allow them to lodge complaints, and maintain neutrality and courtesy (Lim & Kwak, 2022).

Finally, when investigating how these theoretical models and their composite variables varied among pooled respondents and in each country, feeling of safety and distributive justice significantly impacted all respondents and Taiwanese survey participants, whereas likelihood of sanctions only influenced the Japanese. Sun et al. (2014) found that the concerns about safety were significantly linked to public trust in the police in Taiwan, suggesting this study confirmed this linkage in Taiwan and extended it to other East Asian societies. In addition, this study confirmed that distributive justice matters greatly in shaping public confidence in the police across BOTH East Asian and Western Societies (Nix et al., 2015). Pryce and Grant (2021) have noted that citizens universally expect that police should distribute their services fairly to all community members and everyone, regardless of their station in life, deserves equitable treatment from the police.

The variable of likelihood of sanctions impacted all respondents, especially the Japanese respondents. As noted, East Asian societies have rather similar cultural background, among which informal social controls leading to mutual sanctioning and shaming among one’s compatriots characterize this important cultural context (Cao et al., 1998). As a means of informal social control, self-shaming for wrongdoers sanctioned by societal authorities represents a successful social mechanism promoting order maintenance in Japanese society (Braithwaite, 1989). It can be speculated there that the stronger the sense of self-shaming the higher leveler the trust in the police (Pryce et al., 2017).

Conclusion

This study expands the applicability of three theoretical frameworks developed in the West to East Asian democratic societies. The findings confirm the assumption that Taiwanese respondents report the highest levels of confidence in the police, followed by the Japanese and then the South Koreans. Moreover, police effectiveness, homeownership as a proxy for residential stability, and procedural justice significantly influence public confidence in the police in all three democratic countries. Specifically, police effectiveness and procedural justice are likely universal factors shaping citizens’ attitudes toward the police (Lim & Kwak, 2022). Our study also identifies noteworthy differences among the three East Asian democratic societies. Taiwanese citizens prioritize feelings of safety and distributive fairness when expressing confidence in the police, while the likelihood of sanctions impacts Japanese confidence levels as expected. Overall, this study successfully adds to our collective understanding of public attitude formation on police and how citizens in three East Asian countries which have undergone transitions from authoritarianism to democracy see their respective police officers.

Limitations

First, this study employed convenience sampling via an online survey system (LUCID), and as such, it does not represent a truly random sampling of citizens in Japan, South Korea, and Taiwan, suggesting that the capability of generalizability is problematic. Second, some measures developed from the survey questionnaire may be problematic. For example, the measurement of the feeling of safety and procedural justice, which is not an appropriate scale, should be improved. Additionally, attention should be paid to using single items to capture the gauge of confidence in the police. Some important variables, such as victimization representing the instrumental model, and some items seeking to capture the expressive model (e.g., informal social control, social cohesion, and collective efficacy) should be included in such comparative studies in the future.