Neighborhood-Level Effects on Trust in the Police

Abstract

From a multilevel perspective, the present study explores theoretical explanations for the causes of variations in neighborhoods’ trust in the police. The study is designed to answer these questions: Why do some neighborhoods trust the police more than other neighborhoods? What makes a person living in a specific neighborhood to have more or less trust in the police, controlling for the person’s personal views? To address these questions, the study analyzed survey responses from 1,024 residents selected from 25 communities across 5 regions in Ghana. Results revealed significant neighborhood variations in trust in the police in Ghana. Furthermore, a hierarchical linear modeling analysis revealed that the variations among the neighborhoods could be explained by their levels of disorder, income, and education. Findings from this study have both theoretical and practical implications and provide important insights for the police to improve upon their trustworthiness.

Keywords

neighborhood police trust legitimacy Ghana

Introduction

Understanding how residents’ attitudes toward the police are shaped has become an important endeavor in recent years, since police institutions constantly face pressure to develop strong relationships with the public. Investigations into public attitudes have produced four distinct models explaining how citizens form their opinions about the police. The first is the experience with police model, which relates favorable attitudes to the quality of contacts citizens have with the police (Reisig & Parks, 2000). The second model, quality of life, presumes a connection between individuals’ perceptions of their communities and how they think about the police. The demographic model relies on individual characteristics—age, sex, income, education, and race—to explain trusting relationships between the police and the public (Hurst & Frank, 2000; Weitzer & Tuch, 2002). Finally, the neighborhood context model links neighborhood conditions to individual assessment of the police (Reisig & Parks, 2000). Although the neighborhood context model makes a significant contribution to our understanding of how attitudes are developed, it is the least tested among the four models in attitudinal research.

The few studies that have tested neighborhood context model have examined the relationship between contextual variables and public attitudes toward the police and have mostly found negative effects on trust or confidence in the police (Hennigan, Maxson, Sloane, & Ranney, 2002; Reisig & Parks, 2000). Residents of communities characterized by poverty and inequality have been found to exhibit lower confidence in the police (Reisig & Parks, 2000). Moreover, a community’s level of crime also predicts whether a person will have favorable or unfavorable perceptions of the police (Blumstein & Wallman, 2000; Hennigan et al., 2002; Jesilow, Meyer, & Namazzi, 1995; Maxson, Hennigan, & Sloane, 2003; Reisig & Parks, 2000; Sampson, Raudenbush, & Earls, 1997). These studies have consistently found a negative relationship between high crime rates and attitudes toward the police. Similarly, residential location significantly predicts an individual’s level of satisfaction with police service (Schafer, Huebner, & Bynum, 2003).

Most of these studies suffer from a serious methodological issue that raises concern. Previous research adopted a microlevel approach in examining the effects of neighborhood variables on attitudes toward the police (see Ackerman et al., 2001; Macdonald & Stokes, 2006). The micro-level approach involved the use of multivariate regression techniques such as ordinary least square to ascertain the effects of neighborhood variables. This method is problematic because one is likely to violate the assumption of independence of errors, since individuals who live in the same neighborhood are often correlated with one another in other ways. To address this issue, this study uses a multilevel modeling approach to determine the influence of neighborhood-level factors on citizens’ views about the police.

The main purpose of the current study is to further the discussion about the neighborhood context model. Using an aggregated survey of citizens to test the effects of neighborhood variables on trust in the police, the study attempts to answer these questions “Why do some neighborhoods trust the police more than other neighborhoods? What makes a person living in a specific neighborhood to have more or less trust in the police, controlling for the person’s personal views?” Providing answers to these questions will facilitate intellectual discussion about effective ways of enhancing the citizen relationship with the police from a macro perspective. Fieldwork for this study was conducted in 2014 in 25 neighborhoods selected across 5 regions of Ghana, using a multistage sampling procedure. Overall, 1,024 residents participated in a study that lasted for 4 months.

Neighborhood Context Explanations

Studies examining the effects of neighborhood variables on citizens’ perceptions of the police have considered the importance of neighborhood disorder in influencing perceptions. Neighborhood disorder has been found to reduce confidence in the police (Cao, Frank, & Cullen, 1996; Covington & Taylor, 1991; Dowler, 2002; Dowler & Sparks, 2008; Maxson et al., 2003; Sprott & Dood, 2009). Sprott and Dood (2009) compared the influence of perception of disorder on citizens’ attitudes toward the police and the court system. They found that perception of disorder has a greater influence on citizens’ attitudes toward the police than toward the courts. This means that if people consider disorder to be high in the community, they tend to have more negative attitudes toward the police than toward the courts. A plausible explanation could be that people consider crime and disorder prevention to be the work of the police and not the courts.

Maxson, Hennigan, and Sloane (2003) also observed that residents who perceived disorder in their neighborhoods expressed less approval of the police. Consistent with this finding is the observation that individuals’ perception of quality of life in the neighborhood greatly influences their perception of the police (Dowler & Sparks, 2008). These authors further noted that the impact of disorder is greater than the impact of race on citizens’ attitudes toward the police, suggesting that people formulate their attitudes about the police based on the level of quality of life in the community, which is affected by the level of disorder in the neighborhood (Skogan, 1990; Wilson & Kelling, 1982). Neighborhood disorder can cause serious crime if unchecked (Wilson & Kelling, 1982) as well as community deterioration (Skogan, 1990). Furthermore, these scholars have argued that if disorderly behavior is unchecked, members of the neighborhood become fearful and accordingly lose trust and confidence in the police for failing to prevent or stop disorder. Therefore, responding to neighborhood disorder is one way of preventing crime, reducing feelings of insecurity, and subsequently improving public attitudes toward the police.

In relation to the effect of fear of crime in a neighborhood on citizens’ perception of the police, there has been limited effort, and the few studies have yielded largely inconsistent findings. Some studies have demonstrated that fear of crime has a negative relationship with attitudes toward the police (Cao et al., 1996; Cao, Stack, & Sun, 1998; Kaariainen, 2008; Reisig & Giacomazzi, 1998; Reisig & Parks, 2000; Reynolds, Semukhina, & Demidov, 2008; Sampson & Bartusch, 1998; Zhao, Scheiden & Thurman, 2002; Weitzer & Tuch, 2005). The more people fear crime in their neighborhoods, the less confidence they express in the police. Zhao, Scheiden and Thurman (2002) have argued that a decline in fear of crime ultimately leads to an increase in confidence in the police.

In addition, research on fear of crime has shown that citizens’ fear of crime in the neighborhood has a greater influence on attitudes toward the police than do demographic variables (Cao et al., 1996). Similarly, Reynolds, Semukhina, and Demidov (2008) analyzed raw longitudinal data collected annually from 1998 to 2005 to examine the influence of fear of crime on trust in criminal justice institutions. Using structural equation modeling, the authors found that an increase in the level of fear of crime results in a decrease in the variance of trust in the criminal justice system. This further suggests that respondents who scored high on the fear of crime index reported low criminal justice trust. The result is consistent with the findings of Kaariainen (2008), who observed an inverse relationship between insecurity and trust in the police, suggesting that a feeling of insecurity in someone’s neighborhood lowers the level of trust the person has in the police.

Research on fear of crime has discussed two basic types: generalized fear of crime and specific fear of crime (Clemente & Kleiman, 1976; Ferraro & LeGrange, 1987; Hale, 1996). Generalized fear denotes the fear that individuals have in general about their safety in the community and is normally measured by a single item asking respondents how worried they are about walking alone at night. Specific fear of crime denotes the fear of becoming a victim of specific types of crimes. Police researchers examining the influence of fear of crime on public attitudes toward the police have made attempts to investigate the specific impact of these two components and have obtained remarkable results. For example, Kautt (2011) examined the factors that influence citizens’ attitudes toward the police and included two distinct measures of fear of crime. The first measure of fear of crime was a composite scale with five questions asking respondents how worried they were about becoming victims of crimes such as robbery, home break-ins, and sexual assaults. The second measure of fear of crime used a single item, asking respondents how safe they felt when walking alone after dark. These two measures of fear of crime, respectively, denoted specific fear and generalized fear. The author found that respondents who had higher scores on the fear of specific crime index rated the police excellent, while those who were worried about walking alone after dark were less confident in the police. These findings imply that fear of being robbed or sexually assaulted results in higher confidence in the police, because such individuals consider the police to be their protectors. This argument is in line with Block’s (1970) argument that people who were fearful of crime would give the police extensive authority to stop and question suspected individuals. However, being generally afraid is linked to the inability of the police to protect individuals, hence results in lower confidence in the police. Kautt (2011) demonstrated that the effect of fear of crime on perceptions of the police largely depends on whether the fear is specific or general.

Contrary to studies, that have observed effects for the influence of fear of crime on public attitudes toward the police, some researchers claimed that fear of crime is unrelated to attitudes toward the police (Zevitz & Rettammel, 1990). These authors argued that no matter how fearful a person may be, the attitudes of that person toward the police will not be affected. This is surprising, considering the number of studies that have observed significant effects of fear of crime on citizens’ attitudes toward the police. The inconsistencies among prior studies regarding the effect of fear of crime on perceptions are due to measurement and operationalization issues. For example, some studies used a single item to capture residents’ general fear of crime in their neighborhoods. Using a single item, according to fear of crime researchers, is insufficient for capturing individuals’ general fear of crime (Dubow, McCabe, & Kaplan, 1979; Ferraro & LeGrange, 1987). This study addresses this limitation using a scale of 5 items to measure residents’ aggregate fear of crime on attitudes toward the police.

The above reviews illustrate the extent to which neighborhood variables can influence citizens’ attitudes toward the police. To supplement these efforts, the current study examines the influence of the aforementioned contextual variables on Ghanaians’ trust in the police by testing the following hypotheses:

Hypothesis 1: Citizens living in neighborhoods where there is higher rate of disorder will demonstrate lower levels of trust in the police.

Hypothesis 2: Neighborhood’s level of fear of crime will negatively influence citizens’ trust in the police. It is expected that individuals staying in neighborhoods where there are higher rates of fear of crime will report lower levels of trust in the police.

Effects of Other Variables

In addition to the effects of neighborhood factors, several other variables, mostly individual-level characteristics, have been found to predict perceptions of the police. For example, most studies have concluded that individuals’ perception of how well they are treated by the police influences their subjective evaluation of the police (Gau, Corsaro, Stewart, & Brunson, 2012; Hinds, 2007; Hinds & Murphy, 2007; Mazerolle, Antrobus, Bennett, & Tyler, 2013; Mazerolle, Bennett, Antrobus, & Eggins, 2012; Mazerolle, Bennett, Davis, Sargeant, & Manning, 2013; Murphy, 2009; Reisig & Lloyd, 2009; Tyler, 2011; Tyler & Wakslak, 2004). Specifically, Gau, Corsaro, Stewart, and Brunson (2012) used hierarchical linear modeling (HLM) to examine the effects of macro-level factors on procedural justice and police legitimacy and found that, though macro-level factors such as concentrated disadvantage influenced perceptions of procedural justice, procedural justice remained the strongest predictor of legitimacy. This finding is consistent with the conclusion made by Hind and Murphy (2007) using an Australian sample.

Media exposure has also been observed to influence citizens’ evaluation of the police. The public’s negative ratings of the police as well as their dissatisfaction with the police have been linked to high media exposure and news consumption. Studies have shown that extensive media coverage of antisocial police behaviors such as use of force, brutality, and corrupt acts decreases people’s confidence in the police (Dowler & Zawilski, 2007; Eschholz, Blackwell, Gertz, & Chiricos, 2002). The media therefore play significant role in shaping attitudes toward the police. Regarding the effect of ethnicity, previous studies have found that ethnicity is unrelated to perception of the police (Boateng, 2012).

Furthermore, studies have argued that individuals of lower socioeconomic status tend to have less favorable attitudes toward the police than those of higher socioeconomic status (Cao et al., 1996; Huang & Vaughn, 1996; Wu, Sun, & Tripplett, 2009). For instance, Wu, Sun, and Tripplett (2009) observed that lower class individuals were most likely to express lack of satisfaction with the police. However, some studies have also found contrarily that wealthy and highly educated individuals perceive the police less favorably than lower income and less educated persons (Murphy & Worrall, 1999; Weitzer & Tuch, 1999). In a recent comparative study that utilized both US and South Korean samples, Boateng, Lee, & Abess (2016) found that the less educated in both countries had higher confidence in their respective police institutions than highly educated individuals. Specifically, the authors observed that Americans with a high school education were more likely to have higher confidence in the police compared to those with more than a high school education. Likewise, South Koreans who possessed less than a high school education tended to have higher confidence in the police than those with high school education or higher.

To complicate matters further, other studies have argued that socioeconomic status and education are unrelated to attitudes toward the police (Correia, Reisig, & Lovrich, 1996; Parker, Onyekwuluje, & Murty, 1995; Priest & Carter, 1999; Ren, Cao, Lovrich, & Gaffney, 2005; Sims, Hooper, & Peterson, 2002). Correia, Reisig, and Lovrich (1996) observed that level of education did not significantly predict a person’s trust in the police. Similarly, Priest and Carter (1999) found that income had no impact on perceptions of the police.

Despite the conflicting results of previous studies regarding the effects of socioeconomic status and education, no study has examined the effects of a neighborhood’s average income and educational level on individuals’ attitudes toward the police. To fill this void in the literature, the present study tests the following hypotheses:

Hypothesis 3: Citizens living in neighborhoods with lower average income will express lower levels of trust in the police.

Hypothesis 4: Citizens living in neighborhoods where there is higher percentage or number of residents possessing higher education will express higher levels of trust in the police.

Method

Study Population and Participants

The target population for this study included all individuals aged 18 years and above living in urban areas¹ of Ghana at the time of the survey administration. The survey collected information on people’s opinions about the Ghana police and utilized a face-to-face interviewing technique to obtain relevant information from respondents selected randomly from five communities in the capital cities of five administrative regions. Each respondent participating in the study was selected from one household. Survey administration in all the five regions lasted approximately 4 months, from March to June 2014. Specifically, in each region, fieldwork, including training that was offered to research assistants, lasted 3 weeks. The remaining 4 weeks were spent on training of data entry personnel and the entry of data in Excel format.

Sampling Techniques and Procedure

To select respondents for participation in this study, a multistage cluster sampling method was adopted to ensure that a representative sample was obtained. The selection involved five stages. First, 5 regions were purposively selected from the 10 administrative regions in Ghana. It needs to be mentioned that some regions in Ghana are not diverse in terms of population. Some are highly homogeneous, whereas some are diverse because of tremendous economic and industrial activities. To obtain a representative sample, five regions with highly heterogeneous populations were selected. Based on this criterion, the following regions were selected for fieldwork: Ashanti, Greater Accra, Central, Eastern, and Volta. Second, the capital city of each region was purposively selected for the study. There are two main reasons for focusing the study on the capital cities. The first is that regional capitals have more heterogeneous populations than other cities, due to economic and other commercial activities that are widespread in the capital cities. People migrate from other cities to the regional capitals for work or school. Second, police activities are presumed to be higher in the regional capitals than in the rest of the cities, mainly because of the widespread economic activities that attract people from different parts of the regions. Due to the intensity of enforcement efforts, it is believed that people in the regional capitals will experience the police more frequently than those in the other cities. Subsequently, such people are assumed to be in the position to evaluate the police fairly.

The third selection stage involved selecting five communities purposively from each regional capital in the five regions. Several reasons explain the purposive selection of the five communities. First, certain areas in Ghana are typically for commercial activities and are not residential areas. Hence, they are not useful for reaching populations. Second, some communities are reserved for specific government officials such as police officers, correctional officers, fire officials, and ministers. Including such areas in the study may bias the results, since respondents may not offer genuine responses to the questions asked. Third, some communities are inhabited solely by the rich or by the poor, and by including any of those communities in the study, one is bound to obtain partial views of the police, since the rich and the poor hold differential views of the police because of the differential treatment they receive. These reasons preclude random selection of the communities.

In the fourth stage, 250 households were randomly selected from the five communities (50 households from each community) in each capital city. According to Hailand (2003), a household constitutes one or more people who live in the same house and share meals or living accommodation. For the purpose of data collection, any single house where people did not share meals or living space was considered to constitute multiple households. In selecting the households, first, the researcher and the research assistants walked around the selected communities to count the number of households within a particular community. The idea was to prepare a list of all households in the community to facilitate random selection of the households. Once the list was prepared, a quota of 50 was used, and households were randomly selected until 50 were met. This system of selection was to ensure that every household in the community was given an equal chance of inclusion.

The final selection involved the selection of individual respondent from each household. In this selection, the use of the birthday methods—which involve selecting the individual from a household with the most recent or next upcoming birthday—was highly desirable, as they are quick, easy, and less intrusive as well as maximizing cooperation rates (Gaziano, 2005; Oldendick, Bishop, Sorenson, & Tuchfarber, 1988). Both the last-birthday and the next-birthday methods were used randomly in selecting the respondents. The birthday methods have been criticized for not being truly random; however, by applying both the last-birthday and the next-birthday selection methods in the same survey, the selection becomes truly random (Battaglia, Link, Frankel, Osborn, & Mokdad, 2008). The last-birthday method was used to select an individual who was at least 18 years old and was the last to celebrate his or her birthday in the household at the time of the survey administration. The next-birthday method was used to select an individual whose birthday was nearest to the date of the survey administration and was 18 years or older. Overall, 250 respondents were selected from each of the five administrative regions mentioned above, making the total sample 1,250 respondents.

Although the nature of the study did not necessarily require authorities’ permission, as a matter of courtesy, I sought permission from the central police administration and the local councils in various regions where the study was conducted. This approach enabled authorities to be aware that the study was being conducted in the country. Questionnaires were administered to the selected individuals and were collected at the time of administration. This procedure ensured that the sampled individuals were the ones who actually filled out the questionnaire. Overall, 1,024 questionnaires out of the 1,250 were completed and collected, for a response rate of about 82%.

Measures

Dependent variable

Trust in the police was an index variable, which was created from 5 items in the survey. Each item was measured using a 5-item Likert-type scale. These items were adopted from Sunshine and Tyler’s (2003) questions with modifications to fit into the context of the present study. All 5 items had the same lead-in question, “For the following items, kindly indicate whether you agree or disagree: Overall, I trust the police in my neighborhood to protect lives and property; the police can be trusted to make decisions that are right for the people in your neighborhood; the police in your neighborhood are generally honest; I have absolute confidence that the police can do their job well; and the police care about the well-being of everyone they deal with.” Response categories were (1) strongly disagree, (2), disagree, (3) undecided, (4) agree, and (5) strongly agree. A factor analysis with a maximum likelihood estimator indicated that all these items measured the same underlying construct (see Table 1). Therefore, the responses for all the items were summed to form an additive trust in the police scale. The scale had a mean of 14.67 (SD = 4.46) and an α value of .79, suggesting good internal reliability.

Table 1.

Descriptive Statistics of Study Variables.^a

Variables	Mean	SD	Minimum	Maximum
Outcome variable
Trust in the police	14.67	4.46	5.00	25.00
Independent variables
Individual-level predictors
Education—more than senior high school	0.40	0.49	0.00	1.00
Ethnicity—Akan	0.50	0.50	0.00	1.00
Perception of fairness	16.78	4.53	6.00	30.00
Media exposure	3.44	1.04	1.00	5.00
Neighborhood-level predictors
Aggregate fear of crime	12.19	1.36	10.13	15.16
Aggregate disorder	18.50	1.54	16.00	22.53
Aggregate income	0.36	0.11	0.18	0.60
Aggregate more than senior high school	0.41	0.17	0.16	0.85

^aTotal sample size is 1,024 residents and 25 neighborhoods.

Independent variables

Four neighborhood variables were anticipated to influence trust in the police across neighborhoods. These included fear of crime, neighborhood disorder, aggregate level of education, and aggregate income. Descriptive statistics for the variables used in this study are presented in Table 1.

Fear of crime was measured using a 4-item Likert-type scale asking respondents to indicate whether they agreed or disagreed with the following statements: “I am afraid to walk in my neighborhood in the daytime,” “I am afraid to walk in my neighborhood at nighttime by myself,” “The level of security in my neighborhood is very low,” and “Overall, I am afraid to be attacked in my neighborhood.” The response categories were (1) strongly disagree, (2) disagree, (3) undecided, (4) agree, and (5) strongly agree. A factor analysis with a maximum likelihood estimator indicated that all these items measured the same underlying construct. As a result, the responses for the 4 items were summed together to form an additive fear of crime scale. The scale had a mean of 12.08 (SD = 3.52) and an α value of .64, suggesting acceptable internal consistency.

Neighborhood disorder measured the extent of neighborhood disorderly problems perceived by the respondents. It was measured using 8 items, 6 of which were used by Reisig and Parks (2000) to measure perceived incivility in their hierarchical analysis of satisfaction with police study. The8 items measured the extent of neighborhood problems: litter/trash, hanging around, vandalism, abandoned buildings, dirty gutters, gangs, unrepaired streetlights, and drug dealing. Each of these items had a 3-point response set (1 = not a problem; 2 = minor problem; and 3 = major problem). A factor analysis indicated that all these items measured the same underlying construct. Therefore, the responses were summed to form an additive neighborhood disorder scale. The scale had a mean of 18.50 (SD = 1.54) and an α value of .79, suggesting good internal reliability.

Neighborhood level of education was measured as an aggregation of respondents’ level of education (see individual-level variables for details). The aggregate educational variable had a mean of 0.41 with an SD of 0.17.

Neighborhood average income was measured as an aggregation of respondents’ annual household income. Respondents were asked to indicate their household’s income per year (1 = less than Ghana Cedis (GHC) 5,000; 2 = 5,000–10,000; 3 = 10,001–15,000; and 4 = more than 15,000). These categories were later combined to form a dichotomous measure with 0 = GHC 10,000 or less (included initial Categories 1 and 2) and 1 = more than GHC 10,000 (included initial Categories 3 and 4). Any respondent who earned GHC 10,000 or below was considered a low-income earner.² The variable had a mean of 0.36 and an SD of 0.11.

Individual-level variables

Four individual-level variables were included in the HLM analysis to examine their effects on citizens’ trust in the police. These included respondent’s educational status, perception of fairness, media exposure, and ethnicity.

Level of education was measured by asking respondents to indicate their level of educational attainment at the time of survey administration: 1 = no formal education; 2 = junior high school ; 3 = general education development or senior high school (SHS); 4 = higher national diploma; 5 = bachelor’s degree; and 6 = graduate/professional degree. These categories were later combined to form a dichotomous measure with 0 = SHS or below (included initial Categories 1, 2, and 3) and 1 = more than SHS (included initial Categories 4, 5, and 6).

Procedural fairness was measured using 6 Likert-type items adopted from Sunshine and Tyler (2003) and Tankebe (2009). The items, which were modified, asked respondents to indicate the frequency with which the police engaged in behavior consistent with procedural fairness in their neighborhood using the following response categories: (1) never, (2) almost never, (3) sometimes, (4) almost always, and (5) always. The items included “The police make decisions about how to handle problems in fair ways,” “The police treat people fairly,” “The police treat everyone in your neighborhood equally,” “The police accurately understand and apply the law,” “The police make decisions based upon facts, not their personal biases or opinions,” and “The police give honest explanations for their actions to the people they deal with.” A factor analysis revealed that all these items measured the same underlying construct. Therefore, the responses were summed to form a procedural fairness scale, which had a mean of 16.78 (SD = 4.53) and an α value of .77.

A single 4-point Likert-type item asking respondents to indicate the extent to which they heard news about the Ghana police through the mass media measured media exposure. The response categories included (1) never, (2) almost never, (3) sometimes, (4) almost always, and (5) always.

Ethnicity was initially measured as a categorical variable with 1 = Akan; 2 = Ewe; 3 = Ga; 4 = Mole-Dagbon; 5 = others. However, for the purpose of comparison, response categories were recoded as 0 = other (Ewe, Ga, Mole-Dagbon, and others) and 1 = Akan.

Plan of Analysis

HLM was the main modeling approach used in this study.³ This approach allowed the researcher to examine the relative effects of individual-level (Level 1) and neighborhood-level (Level 2) variables on citizens’ trust in the police. Four models were conducted, starting with the null (intercept only) model with no predictors, followed by the total effect model, which included only Level-1 individual variables with no Level-2 variable. The third model, the contextual effect model, included both Level-1 and Level-2 variables, which were grand mean centered. Finally, the fourth model was the between effect model, which included Level-1 group mean-centered variables and Level-2 grand mean-centered variables to determine the between-neighborhood effect on trust. However, before running these models, an analysis of variance (ANOVA) was conducted to determine whether the neighborhoods being studied varied by their level of trust in the police as well as by level of fear of crime, education, and income.

Results

Exploring Significant Differences Among Neighborhoods

To explore significant differences among the 25 neighborhoods on the 5 dimensions, ANOVA and effect size calculations were conducted (see Table 2). The first column represents the results for neighborhoods’ aggregate trust in the police. The test revealed significant group differences (F = 9.610, df = 24, p < .001).⁴ A post hoc Bonferroni comparison was conducted to determine each group impact. This analysis indicated significant mean differences among the neighborhoods, suggesting that the neighborhoods did differ in terms of their levels of trust in the police. The effect size calculations (η² = .193) indicate that the neighborhood variable explains 19% of the variance in trust in the police.

Table 2.

Neighborhood-Level Variations in Aggregate Trust in the Police, Fear of Crime, Disorder, More Than High School, and Income.

Neighborhood	Trust	Fear of Crime	Disorder	More Than High School	Income
A	14.88 (3.67)	11.31 (3.57)	18.39 (4.22)	0.21 (0.42)	0.18 (0.39)
B	14.44 (4.04)	12.05 (2.99)	16.44 (3.55)	0.16 (0.37)	0.22 (0.42)
C	13.22 (5.90)	11.07 (3.19)	18.19 (4.23)	0.25 (0.44)	0.32 (0.47)
D	16.98 (3.70)	10.34 (3.61)	16.00 (4.60)	0.20 (0.41)	0.42 (0.50)
E	13.00 (5.06)	11.67 (3.14)	18.26 (3.82)	0.24 (0.43)	0.32 (0.48)
F	15.57 (4.99)	13.26 (3.83)	17.85 (4.28)	0.56 (0.50)	0.26 (0.44)
G	13.90 (4.77)	11.48 (3.83)	17.59 (3.29)	0.32 (0.47)	0.24 (0.43)
H	14.12 (4.30)	11.85 (3.95)	19.20 (3.12)	0.42 (0.50)	0.31 (0.47)
I	14.74 (4.08)	12.72 (3.80)	18.94 (3.86)	0.85 (0.37)	0.60 (0.50)
J	15.22 (3.59)	11.48 (3.97)	17.17 (3.68)	0.69 (0.47)	0.22 (0.42)
K	17.47 (3.91)	11.58 (2.72)	16.87 (3.65)	0.35 (0.48)	0.58 (0.50)
L	12.63 (4.06)	10.13 (3.43)	21.07 (3.23)	0.38 (0.49)	0.41 (0.50)
M	11.40 (3.65)	10.74 (3.18)	22.53 (2.61)	0.51 (0.51)	0.38 (0.49)
N	11.04 (2.90)	12.94 (3.27)	20.72 (3.07)	0.21 (0.41)	0.34 (0.48)
O	13.12 (3.56)	12.17 (3.16)	21.53 (3.21)	0.37 (0.49)	0.32 (0.47)
P	12.92 (4.76)	10.47 (3.62)	17.51 (3.44)	0.31 (0.47)	0.35 (0.49)
Q	14.80 (4.61)	12.00 (3.66)	18.13 (3.92)	0.35 (0.49)	0.44 (0.51)
R	13.86 (4.30)	11.83 (3.46)	18.75 (4.11)	0.50 (0.51)	0.34 (0.48)
S	15.57 (3.59)	12.18 (3.06)	18.61 (2.87)	0.52 (0.51)	0.40 (0.50)
T	15.81 (4.23)	11.71 (3.61)	18.26 (3.17)	0.24 (0.44)	0.46 (0.51)
U	18.20 (2.80)	15.16 (1.62)	17.51 (2.44)	0.48 (0.51)	0.57 (0.53)
Z	16.70 (3.09)	14.58 (2.68)	18.50 (3.55)	0.53 (0.50)	0.50 (0.51)
W	16.86 (3.40)	13.36 (2.88)	18.00 (2.88)	0.40 (0. 50)	0.36 (0.50)
X	18.00 (4.08)	13.80 (3.03)	18.45 (3.71)	0.61 (0.50)	0.33 (0.49)
Y	16.00 (2.22)	14.88 (2.00)	17.93 (2.17)	0.56 (0.51)	0.25 (0. 50)
F-score	9.61	6.25	8.13	5.35	1.81
p-Value	<001	<001	<001	<001	<001
η²	.193	.135	.180	.118	.056

Note. Actual names of the neighborhoods have been replaced with alphabetical letters due to confidentiality reasons. Standard deviations are in parentheses. η² = effect size.

The second column presents the results for the neighborhoods’ aggregate levels of fear of crime. With an effect size of .135, neighborhood explains about 14% of the variance in fear of crime. The ANOVA test revealed significant group differences (F = 6.25, df = 24, p < .001). Since the F-test was significant, a post hoc Bonferroni comparison was conducted to determine each group impact. The analysis revealed significant mean differences, indicating that the neighborhoods differed in terms of their aggregate levels of fear of crime.

The third column presents the results for the neighborhoods’ aggregate levels of neighborhood disorder. The test revealed significant group differences (F = 8.13, df = 24, p < .001). This suggests that the 25 neighborhoods did differ in terms of their aggregate levels of disorder. A post hoc Bonferroni comparison revealed significant mean differences among the groups. For example, the analysis demonstrates that neighborhoods A and M, A and O, B and L, B and M, B and N, B and O, C and M, C and O, D and L, D and M, and D and N significantly differ. The effect size calculations (η² = .180) indicate that the neighborhood variable explains 18% of the variance in disorder.

Similarly, ANOVA of the neighborhoods’ aggregate rates of residents possessing more than high school education (Column 4) revealed significant group differences (F = 5.35, df = 24, p < .001). A post hoc Bonferroni comparison conducted revealed significant mean differences among the neighborhoods, suggesting that the neighborhoods differed in terms of average number of individuals who had attained more than high school education. Neighborhood explains about 12% of the variance in more than high school education.

Finally, Column 5 presents the results for the neighborhoods’ average income. The test revealed significant group differences (F = 1.81, df = 24, p < .001). A post hoc Bonferroni comparison conducted revealed significant mean differences among the groups. The effect size calculations (η² = .056) suggest that the neighborhood variable explains 6% of the variance in aggregate income.

Overall, the ANOVA on trust in the police indicates significant neighborhood variations. The question left to be answered is could these variations be due to the variation that exists among the neighborhoods in terms of their aggregate rates of fear of crime, disorder, education, and income? This question is explored in the HLM analysis.

Exploring Contextual Effects on Trust in the Police

To examine contextual effects on citizens’ trust in the police, a multilevel modeling technique was employed using HLM software Version 6. Four models were created and the results are presented in Table 3. Model 1 was the null model, which included only the grand mean-centered outcome variable (trust in the police). The results revealed that the group mean level of trust in the police was positive and higher than the grand mean (14.80 vs. 14.67). The model was significant (t = 38.63, p < .001), indicating that there was a significant variation in trust in the police at Level 2.⁵

Table 3.

Hierarchical Linear Model Analysis of Trust in the GPS.^a

Variables	Model 1	Model 2	Model 3	Model 4
Variables	Null	Total Effect (Level 1 Grand)	Contextual Effect (Levels 1 and 2 Grand)	Between Effect (Level 1 Group and Level 2 Grand)
Intercept	14.80 (0.38) [38.63]***	14.69 (0.23) [63.21]***	14.70 (0.17) [86.56]***	14.77(0.22) [67.42]***
Level 2: Neighborhood
Mean fear of crime	—	—	0.02 (0.13) [0.18]	0.50 (0.18) [2.82]**
Mean disorder	—	—	−0.40 (0.10) [−3.96]***	−0.79 (0.15) [−5.39]***
Mean income	—	—	1.31 (1.84) [0.71]	4.11 (1.96) [2.10]*
Mean more than SHS	—	—	3.06 (1.29) [2.73]*	1.79 (1.83) [0.98]
Level 1: Individual
Akan	—	0.63 (0.28) [2.37]*	0.55 (0.27) [2.02]*	0.56 (0.30) [1.88]*
More than SHS	—	0.39 (0.17) [2.24]*	0.30 (0.18) [1.63]	0.30 (.18) [1.61]
Procedural fairness	—	0.57 (0.03) [18.42]***	0.56 (0.03) [17.97]***	0.60 (0.03 [17.73]***
Media exposure	—	−0.26 (0.14) [−1.90]*	−0.23 (0.14) [−1.73]*	−0.26 (0.14) [−1.89]*
Variance components
Level 1 (r)	16.43	10.98	10.97	10.97
Level 2 (u ₀)	3.38	1.09	0.62	1.19
Pseudo R ²
Level 1	—	33%	33%	33%
Level 2	—	68%	82%	65%
Model fit statistics
Deviance statistics	5,632.802	4,699.294	4,684.193	4,693.678
No. of Est.	2	2	2	2

Note. GPS = Ghana Police Service. SHS =Senior High School. Standard errors in parenthesis, t-ratios in bracket.

^aModel Specification: Model 1: TRUST _ij = γ₀₀ + u _0j+ r_ij ; Model 2: TRUST _ij = γ₀₀ + γ₁₀*AKAN _ij + γ₂₀*POSTSENI _ij + γ₃₀*PROCEDUR _ij + γ₄₀*MEDEX _ij + u _0j+ r_ij ; Model 3: TRUSTSCA _ij = γ₀₀ + γ₀₁*FEARCRIM _j + γ₀₂*DISORDER _j + γ₀₃*INCOME_M _j + γ₀₄*POSTSENI _j + γ₁₀*AKAN _ij + γ₂₀*POSTSENI _ij + γ₃₀*PROCEDUR _ij + γ₄₀*QJ12 _ij + u _0j+ r_ij ; Model 4: TRUST _ij = γ₀₀ + γ₀₁*FEARCRIM _j + γ₀₂*DISORDER _j + γ₀₃*INCOME_M _j + γ₀₄*POSTSENI _j + γ₁₀*AKAN _ij + γ₂₀*POSTSENI _ij + γ₃₀*PROCEDUR _ij + γ₄₀*MEDEX _ij + u _0j+ r_ij .

*p < .05. **p < .01. ***p < .001.

In Model 2, four Level-1 grand mean-centered variables—Akan ethnicity, more than high school education, perception of procedural fairness, and media exposure—were added to determine their total effect on citizens’ trust in the police. The intercept indicates that the levels of trust possessed by citizens who were Akans, had attained more than SHS education, perceived the police to be fair, and had less frequent media exposure were slightly above the group mean (14.69***). The Akan variable was significant (t = 2.37, p < .05), and with a coefficient of 0.63, indicating that citizens who belonged to the Akan ethnic group had more trust in the police than those who belonged to other groups. Similarly, more than SHS was significant (t = 2.24, p < .05) and with a positive coefficient, citizens who possessed more than SHS education had more trust in the police than citizens who possessed SHS education or less.

In addition, procedural fairness had a significant influence on trust in the police (t = 18.42, p < .001), and had a positive coefficient of 0.57, meaning a unit increase in perception of procedural fairness resulted in a 0.57 increase in trust in the police. Finally, media exposure was equally significant (t = −1.90, p < .05), and the negative coefficient (−0.26) indicated that citizens who frequently experienced the police through the media had lower trust in the police.

Model 2, which contained only individual-level variables, explained a decent amount of variability in trust in the police at both Level 1 and Level 2. Specifically, the model explained 33% of the variance in trust in the police at Level 1 and 68% at Level 2.

To determine the Level 2 contextual effect and the Level 1 within-neighborhood effect, four grand mean-centered variables—mean (aggregate) fear of crime, mean (aggregate) disorder, mean (aggregate) income, and mean (aggregate) more than high school—were added in Model 3. The intercept indicates that citizens who were Akans; perceived the police to be fair; had less frequent media exposure; and lived in neighborhoods with the grand mean levels of fear of crime, disorder, income, and more than high school education possessed levels of trust in the police that were slightly above the group mean (14.70). Mean (aggregate) disorder was significant (t = −3.96, p < .001) and the coefficient was negative (−0.40). The negative coefficient suggests that irrespective of individual perceptions of disorder, citizens who lived in neighborhoods with greater aggregates of disorderly conduct tended to have lower trust in the police compared to those who lived in neighborhoods with lower aggregates of disorderly behavior. Similarly, mean (aggregate) more than SHS education was significant (t = 2.73, p < .05), suggesting that, regardless of individual educational background, citizens who lived in neighborhoods with a greater proportion of residents possessing more than SHS education tended to have higher trust in the police than those who lived in neighborhoods with few residents having more than SHS education.

Moreover, three Level-1 grand mean-centered variables remained significant in Model 3 after controlling for Level 2 contextual variables. The Akan variable was significant (t = 2.02, p < .05), and the coefficient of 0.55 indicates that citizens who belonged to the Akan ethnic group had higher trust in the police than those who belonged to other groups. Furthermore, procedural fairness had a significant influence on trust in the police (t = 17.97, p < .001), with a positive coefficient of 0.56; a 1-unit increase in perception of procedural fairness resulted in a 0.56 increase in trust in the police. In addition, media exposure was significant (t = −1.89, p < .05), indicating that citizens who frequently experienced the police through the media had lower trust in the police. This model also explained a fair amount of the variability in trust in the police at both levels. At Level 1, the model explained 33% of the variance in trust in the police, and at Level 2, it significantly explained 82%.

To examine between-neighborhood effects, four group-mean-centered Level-1 variables and four grand mean-centered Level 2 variables were added in Model 4. The intercept indicates that citizens who were Akans; perceived the police to be fair; had less frequent media exposure; and lived in neighborhoods with the grand mean levels of fear of crime, disorder, income, and more than high school education possessed levels of trust in the police that were slightly above the group mean (14.77). Three grand mean-centered Level 2 variables were found to influence trust in the police significantly. Mean (aggregate) fear of crime was significant (t = 2.82, p < .01) and the coefficient was positive (0.50). The positive coefficient suggests that, irrespective of individual perceptions of fear of crime, citizens who lived in neighborhoods with greater aggregates of fear of crime tended to have higher trust in the police compared to those who lived in neighborhoods with lower aggregate fear of crime.

Mean (aggregate) disorder was significant (t = −5.39, p < .001) and the coefficient was negative (−0.79), implying that, irrespective of individual perceptions of disorder, citizens who lived in neighborhoods with a greater aggregate level of disorderly conduct tended to have lower trust in the police compared to those who lived in neighborhoods with lower aggregates of disorderly behavior. Similarly, mean (aggregate) income was significant (t = 2.10, p < .05), suggesting that, regardless of individual income levels, citizens who lived in neighborhoods with high average income (or earnings) tended to have higher trust in the police than those who lived in neighborhoods with low average income.

Likewise, three Level 1 grand mean-centered variables remained significant in Model 4 after controlling for Level 2 contextual variables. The Akan variable was significant (t = 1.88, p < .05), and the coefficient of 0.56 shows that citizens who belonged to the Akan ethnic group had higher trust in the police than those who belonged to other groups. Furthermore, procedural fairness had a significant influence on trust in the police (t = 17.73, p < .001) and had a positive coefficient of 0.60; a 1-unit increase in perception of procedural fairness resulted in a 0.60 increase in trust in the police. In addition, media exposure was significant (t = −1.89, p < .05), indicating that citizens who frequently experienced the police through the media had lower trust in the police. This model explained 33% of variance in trust at the individual level and 65% at the neighborhood level.

Discussion

The purpose of the present study was to explore contextual explanations for the varied levels of trust among different neighborhoods. The study aimed to answer these questions: Why do some neighborhoods trust the police more than other neighborhoods? What makes a person living in a specific neighborhood place more or less trust in the police, controlling for the person’s personal views? The findings revealed significant effects of community characteristics in explaining variations in levels of trust across neighborhoods. The 25 neighborhoods studied differed significantly in their levels of trust in the police. Some neighborhoods had higher trust in the police, while others had lower trust in the police.

Using advanced multilevel modeling, the study addressed these questions by observing the effects of three contextual variables on neighborhoods’ confidence in the police. Community rate of disorder exerted a significant negative influence on the level of trust of people living in the area, controlling for their individual characteristics, such as ethnic background, perception of fairness, and exposure to the media. This result implies that Ghanaians who live in neighborhoods with higher rates of disorder will express lower trust and confidence in the police irrespective of their perceptions of the police. Citizens’ levels of trust, therefore, vary based on the area where they live. The observation is consistent with prior studies that found similar results in other social contexts (Dowler & Sparks, 2008; Maxon et al., 2003; Sprott & Dood, 2009). Maxon et al. (2003) found that residents who believed disorder was on the increase in their neighborhoods expressed disapproval of the police and were more critical of how the police performed their duties.

Additionally, community levels of income and education were also found to affect neighborhood level of trust in the police. The effect of aggregate community income was positive, indicating that neighborhoods possessing higher average income had greater trust in the police compared to those with lower average income. At the individual level, citizens who reside in higher average income neighborhoods demonstrated favorable attitudes toward the police regardless of their opinions of the police. Similarly, neighborhood level of education influenced how citizens perceived the police. People who lived in neighborhoods where the majority of the residents were highly educated were more likely to show greater confidence in the police.

Possible explanations for these results could be seen from the assumptions of the conflict theory (see Chambliss & Seidman, 1971). Supporters of this theory have argued that the interests of the dominant class—which in this case would be the highly educated and high-income earners in the society—are protected by the police, whereas the lower class individuals—the less educated and lower income earners—are continuously monitored by law enforcement officials (Chambliss & Seidman, 1971; Das, 1983). It is, therefore, obvious that communities where the highly educated and high-income citizens live will receive favorable treatment from the police, which might lead to a favorable trusting relationship between the two groups. Conversely, areas where lower class citizens are concentrated may experience biased treatment and aggressive enforcement strategies that might deteriorate the relationship between the police and such communities. According to researchers (Gabbidon & Jordan, 2013; Weitzer & Tuch, 2006; Wu et al., 2009), this deteriorated relationship may lead to the expression of negative views about the police. This line of reasoning also fits well into the ecological contamination explanation of police action on the street (Alpert & MacDonald, 2001; Kane, 2002). These authors have argued that police tailor their behavior based on the specific neighborhoods they serve. Therefore, if police officers work in dangerous neighborhoods, they are tempted to use force more often.

The last neighborhood variable that had a significant influence on community trust in the police was average fear of crime in the neighborhood. Interestingly, the effect was positive, indicating that neighborhoods with high rates of fear of crime had higher confidence in the police. Stated differently, residents who lived in neighborhoods where fear of crime was high tended to have greater confidence in the police irrespective of their personal opinions of the police. This finding is surprising, given the abundant studies that have observed a negative relationship between fear of crime and attitudes toward the police at the individual level (Cao et al., 1996; Kaariainen, 2008; Zhao et al., 2002). These authors believed that an increase in fear of crime would result in a decrease in attitudes. However, one might also be tempted to think that, at the community level, if citizens perceive fear of crime to be high in communities where they reside, they may tend to see the police as the only source of security. To ensure maximum security and protection, citizens inadvertently develop greater trust in the police.

Understanding the Effects of the Neighborhood Variables in Ghana

The observed effects of the community variables in the Ghanaian context can be better understood with an exploration of the social development prevailing within specific neighborhoods in Ghana. Social change—modernization, industrialization, and urbanization—has led to a complex division of labor, migration, and production and distribution of goods and services (Amuzu & Leitmann, 1991; Asiama, 1984) as well as altering the patterns of people’s routine activities and lifestyles (Appiahene-Gyamfi, 2002). Moreover, the process of social change has widened the gap between the rich and the poor and has created a phenomenon known as persistent differentiated neighborhoods in modern Ghana. The rich and the poor live in different locations with different levels of development. It can be argued that differentiated neighborhoods existed during the colonial days, but the problem today is worse than what existed in the past. This phenomenon has serious implications for policing, since police officers tend to treat people differently based on where they stay or work. Residents of affluent neighborhoods, by virtue of their influence and power, are accorded much respect and offered friendly treatment by the police. In contrast, those who live in poor neighborhoods—popularly known as ghettos or slums—are treated harshly and sometimes with no respect (Appiahene-Gyamfi, 2002).

Residential neighborhoods in Ghana are broadly categorized as low-income, middle-income, and high-income areas. These categories are differentiated primarily by factors such as housing conditions, availability of facilities, and level of security. For example, in the Accra metropolitan area, housing conditions in the low-income neighborhoods are depressed with social and engineering infrastructure. Buildings in these neighborhoods are usually dilapidated and often made of poor quality materials: mud, untreated timber, and zinc roofing sheets for walling (see Accra Metropolitan Assembly website: www.ama.ghanadistricts.gov.gh).

According to the Assembly, about 58% of the city’s population reside in the low-income areas, such as Osu, Nungua, Chorkor, Jamestown, Nima, Abeka, and Sukura, to mention a few. These areas in the city are notoriously characterized by inadequate housing infrastructure, poor drainage systems, erosion, and high population concentration. In addition, layouts are haphazard and there are no proper streets running through the neighborhoods. Houses and buildings are scattered everywhere, and there is no room for easy maneuvering. These kinds of layouts create space for illicit activities to flourish but do not support police activities such as bike, vehicular, and foot patrols (Appiahene-Gyamfi, 2003). The majority (if not all) of the residents in the poor neighborhoods do not have access to telephones at home, and as Appiahene-Gyamfi (2003) noted, the lack of access to telephones by residents prevents crime reporting to the police. To complicate matters, most of the poor neighborhoods in Ghana are located far from police stations, further limiting residents’ ability and desire to physically walk in to make complaints or report to the police.

Comparatively, the middle-income areas are generally planned, and buildings are made with quality materials. Housing conditions are much better than those in the low-income neighborhoods, and residents in these neighborhoods make up 32% of the city’s population. The middle-income neighborhoods include areas such as Kanda Estates, Abelempke, Tesano, and Dansoman Estates. Most buildings in these areas are government-owned and are mainly occupied by government workers. The areas are semiplanned with streets that sometimes support police operations.

The high-income areas, also called the affluent neighborhoods, are well planned and structured and have well-developed infrastructure. Residents in these neighborhoods form 10% of the city’s population and are mostly the wealthiest people in the society—academics, businessmen and businesswomen, politicians, foreigners, diplomats, and those who have lived in either the United States or Europe. Because of their wealth, residents in these neighborhoods enjoy maximum security, often provided either by the police security apparatus or by private security agency. Appiahene-Gyamfi (2003) observed that homes with police or private security guards experienced fewer burglary incidents. Furthermore, the patterns of layout in the rich neighborhoods support police patrolling activities, and as a result, there is an increased police presence at all times. Additionally, residents have access to telephones, which makes it easy for them to call the police to report criminal incidents.

Situations in all three categories of neighborhoods differ and have different effects on residents’ relationship with the police. For instance, the prevailing circumstances in the low-income neighborhoods—no police presence and not being able to reach the police, coupled with ill treatment from the police—significantly and negatively affect residents’ relationship with and attitudes toward the police. It is, therefore, not surprising that residents in such neighborhoods will express negative feelings and distrust in the police service. Conversely, the rich, who enjoy maximum security, are recognized with respect and can easily reach the police, have different, mostly favorable attitudes toward the police in Ghana.

The present study is not without limitations, and these must be acknowledged. First, the study examined contextual factors influencing trust in the police and attempted to generalize its findings to the entire population of Ghana. However, the study excluded the opinions and attitudes of individuals living in the rural areas of Ghana, hindering the study’s ability to generalize its findings to the entire population. In the light of these limitations, it is recommended that future research be conducted to examine the opinions of individuals living in rural areas of Ghana about the Ghana police. Second, the study failed to examine all the relevant neighborhood variables that have been found by research to influence trust in the police. As a result, the effects of variables such as neighborhood crime rates, unemployment rates, and proximity of police stations on trust in the police are still not known. Furthermore, important individual-level factors such as gender, age, and race were not included in the current analysis, given their influence on trust in the police. It is therefore suggested that future research should be completed to take into consideration these and other factors not examined in this study.

Despite these limitations, the findings of this study help to answer crucial questions pertaining to individual and community attitudes toward the police. The findings offer both theoretical and practical implications. Theoretically, the findings offer empirical justification for the use of neighborhood-level variables in explaining citizens’ attitudes toward the police. Specifically, the findings help to address questions such as why attitudes toward the police vary from neighborhood to neighborhood and why individuals with the same perceptions of the police might have different levels of trust and confidence in the police.

Practically, the findings can offer meaningful indications for the police to develop better policies directed at establishing cordial relationship with the community they serve. Analysis based on the study’s data indicates that people who live in communities with low educational attainment, low annual income, and high disorder rates will inadvertently develop negative attitudes toward the police irrespective of their individual perceptions. Although the police cannot address issues pertaining to low income and educational attainment, they can certainly control disorderly conducts in the neighborhoods in two ways. First, the police must be empowered to enforce city ordinances so that minor disorderly conducts that are not illegal could be punished. Second, police administrators must increase the presence of the police in areas where disorder is high and consider fighting all forms of disorderly behaviors as operational priority. Zero-tolerance policy needs to be implemented to fight neighborhood disorder effectively. In addition to reducing disorder, each community must have access to educational facilities to boost the education needs of its residents.

The present study represents one of the few efforts to use mixed model approach to understand how citizens develop their perceptions about their local police and recommends that police departments and the communities need to continue efforts to improve police–citizen relationship. Reducing disorderly conduct and its associated fear not only enhance the image of the police in the community but also foster an environment that promotes social cohesion among community members.

Footnotes

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Ackerman

Anderson

Jensen

Ludwig

Montero

Plante

Yanez

(2001). Crime rates and confidence in the police: America’s changing attitudes toward crime and police, 1972-1999. Journal of Sociology and Social Welfare, 28, 43–54.

Alpert

G. P.

MacDonald

J. M.

(2001). Police use of force: An analysis of organizational characteristics. Justice Quarterly, 18, 393–409.

Amuzu

A. T.

Leitmann

(1991). Urban management and environment: Environmental profile of Accra: Case study. Accra, Ghana: UNDP.

Appiahene-Gyamfi

(2002). An analysis of the broad crime trends in Ghana. Journal of Criminal Justice, 30, 229–243.

Appiahene-Gyamfi

(2003). Urban crime trends and patterns in Ghana: The case of Accra. Journal of Criminal Justice, 31, 13–23.

Asiama

S. O.

(1984). The land factor in housing for low income urban settlers: The example of Madina, Ghana. Third World Planning Review, 6, 171–184.

Battaglia

M. P.

Link

M. W.

Frankel

M. R.

Osborn

Mokdad

A. H.

(2008). An evaluation of respondent selection methods for household mail surveys. Public Opinion Quarterly, 72, 459–469.

Block

R. L.

(1970). Support for civil liberties and support for the police. American Behavioral Scientist, 13, 781–796.

Blumstein

Wallman

(2000). The crime drop in America. New York, NY: Cambridge University Press.

10.

Boateng

F. D.

(2012). Public trust in the police: Identifying factors that shape trust in the Ghanaian Police (IPES Working Paper Series, No. 42). Retrieved from www.ipes.info/WPS/WPS_No_42.pdf

11.

Boateng

F. D.

Lee

H. D.

Abess

(2016, Forthcoming,). Analyzing citizens’ reported levels of confidence in the police: A cross-national study of public attitudes toward the police in the United States and South Korea. Asian Journal of Criminology, DOI: 10.1007/s11417-016-9231-8.

12.

Browne

W. J.

Draper

(2000). Implementation and performance issues in the Bayesian fitting of multilevel models. Computational Statistics, 15, 391–420.

13.

Cao

Frank

Cullen

F.T.

(1996). Race, community context and confidence in police. American Journal of Police, 15, 3–22.

14.

Cao

Stack

Sun

(1998). Public confidence in the police: A comparative study between Japan and America. Journal of Criminal Justice, 26, 279–289.

15.

Chambliss

Seidman

(1971). Law, order, and power. Reading, MA: Addison-Wesley.

16.

Clemente

Kleiman

M. B.

(1976). Fear of crime in the United States: A multivariate analysis. Social Forces, 56, 519–31.

17.

Correia

Reisig

Lovrich

(1996). Public perceptions of state police: An analysis of individual-level and contextual variables. Journal of Criminal Justice, 24, 17–28.

18.

Covington

Taylor

R. B.

(1991). Fear of crime in urban residential neighborhoods: Implications of between-and-within-neighborhood sources for current models. Sociological Quarterly, 32, 231–249.

19.

Das

(1983). Conflict views on policing: An evaluation. American Journal of Police, 3, 51–81.

20.

Dowler

(2002). Media influence on citizens attitudes toward police effectiveness. Policing & Society, 12, 227–238.

21.

Dowler

Sparks

(2008). Victimization, contact with police and neighborhood conditions: Reconsidering African American and Hispanic attitudes toward the police. Police Practice and Research: An International Journal, 9, 1–20.

22.

Dowler

Zawilski

(2007). Public perceptions of police misconduct and discrimination: Examining the impact of media consumption. Journal of Criminal Justice, 35, 197–203.

23.

Dubow

McCabe

Kaplan

(1979). Reactions to crime: A critical review of the literature. Washington, DC: National Institute of Law Enforcement and Criminal Justice, U.S. Government Printing Office.

24.

Eschholz

Blackwell

B. S

Gertz

Chiricos

(2002). Race and attitudes toward the police: Assessing the effects of watching ‘‘reality’’ police programs. Journal of Criminal Justice, 30, 328–341.

25.

Ferraro

K. F.

LaGrange

(1987). The measurement of fear of crime. Sociological Inquiry, 57, 70–101.

26.

Gabbidon

S. L.

Jordan

K. L.

(2013). Public opinion on the killing of Trayvon Martin: A test of the racial gradient thesis. Journal of Crime and Justice, 36, 283–298.

27.

Gau

J. M.

Corsaro

Stewart

E. A.

Brunson

R. K.

(2012). Examining macro-level impacts on procedural justice and police legitimacy. Journal of Criminal Justice, 40, 333–343.

28.

Gaziano

(2005). Comparative analysis of within-household respondent selection techniques. Public Opinion Quarterly, 69, 124–157.

29.

Hale

(1996). Fear of crime: A review of the literature. International Review of Victimology, 4, 79–150.

30.

Hailand

W. A.

(2003). Anthropology. Belmont, CA: Wadsworth.

31.

Hennigan

Maxson

Sloane

Ranney

(2002). Community views on crime and policing: Survey mode effects on bias in community surveys. Justice Quarterly, 19, 565–587.

32.

Hinds

(2007). Building police-youth relationships: The importance of procedural justice. Youth Justice, 7, 195–209.

33.

Hinds

Murphy

(2007). Public satisfaction with police: Using procedural justice to improve police legitimacy. Australian and New Zealand Journal of Criminology, 40, 27–42.

34.

Huang

Vaughn

(1996). Support and confidence: Public attitudes toward the police. In Flanagan

Longmire

(Eds.), American view crime and justice: A national public opinion survey (pp. 31–45). Thousand Oaks, CA: Sage.

35.

Hurst

Y. G.

Frank

(2000). How kids view cops: The nature of juvenile attitudes toward the police. Journal of Criminal Justice, 28, 189–202.

36.

Jesilow

Meyer

J. A.

Namazzi

(1995). Public attitudes towards the police. American Journal of Police, 14, 67–88.

37.

Kaariainen

(2008). Why do the Finns trust the police? Journal of Scandinavian Studies in Criminology and Crime Prevention, 9, 141–159.

38.

Kane

R. J.

(2002). The social ecology of police misconduct. Criminology, 40, 867–896.

39.

Kautt

(2011). Public confidence in the British police: Negotiating the signals from Anglo-American research. International Criminal Justice Review, 21, 353–382.

40.

Macdonald

Stokes

J. R.

(2006). Race, social capital, and trust in the police. Urban Affairs Review, 41, 358–375.

41.

Maxson

Hennigan

Sloane

D. C.

(2003). Factors that influence public opinion of the police. Washington, DC: United States Department of Justice, National Institute of Justice.

42.

Mazerolle

Bennett

Antrobus

Eggins

(2012). Procedural justice, routine encounters and citizen perceptions of police: Main findings from the Queensland Community Engagement Trial (QCET). Journal of Experimental Criminology, 8, 343–367. doi:10.1007/s11292-012-9160-1

43.

Mazerolle

Bennett

Davis

Sargeant

Manning

(2013). Procedural justice and police legitimacy: A systematic review of the research evidence. Journal of Experimental Criminology, 9, 245–274. doi:10.1007/s11292-013-9175-2

44.

Mazerolle

Antrobus

Bennett

Tyler

(2013). Shaping citizen perceptions of police legitimacy: A randomized field trail of procedural justice. Criminology, 51, 33–63.

45.

Murphy

(2009). Public satisfaction with police: The importance of procedural justice and police performance in police-citizen encounters. Australian & New Zealand Journal of Criminology, 42, 159–178.

46.

Murphy

D. W.

Worrall

J. L.

(1999). Residency requirements and public perceptions of the police in large municipalities. Policing: An International Journal of Police Strategies & Management, 22, 327–342.

47.

Oldendick

R. W.

Bishop

F. G.

Sorenson

B. S.

Tuchfarber

J. A.

(1988). A comparison of the Kish and last birthday methods of respondent selection in telephone surveys. Journal of Official Statistics, 4, 307–318.

48.

Parker

K. D.

Onyekwuluje

A. B.

Murty

K. S.

(1995). African Americans’ attitudes toward the local police: A multivariate analysis. Journal of Black Studies, 25, 396–409.

49.

Priest

T. B.

Carter

D. B.

(1999). Evaluations of police performance in an African American sample. Journal of Criminal Justice, 27, 457–465.

50.

Reisig

Giacomazzi

L. A.

(1998). Citizen perceptions of community. Policing: Are attitudes toward police important? Policing: An International Journal of Police Strategies and Management, 21, 547–561.

51.

Reisig

M. D.

Parks

R. B.

(2000). Experience, quality of life and neighborhood context: A hierarchical analysis of satisfaction with police. Justice Quarterly, 17, 607–630.

52.

Reisig

M. D.

Lloyd

(2009). Procedural justice, police legitimacy, and helping the police fight crime: Results from a survey of Jamaican adolescents. Police Quarterly, 12, 42–62.

53.

Ren

Cao

Lovrich

Gaffney

(2005). Linking confidence in police with the performance of the police: Community policing can make a difference. Journal of Criminal Justice, 33, 55–66.

54.

Reynolds

K. M.

Semukhina

O. B.

Demidov

N. N.

(2008). A longitudinal analysis of public satisfaction with the police in the Volgograd region of Russia 1998-2005. International Criminal Justice Review, 18, 158–189.

55.

Sampson

R. J.

Raudenbush

S. W.

Earls

(1997). Neighborhood and violent crime: A multilevel study of collective efficacy. Science, 277, 918–924.

56.

Sampson

R. J.

Jeglum-Bartusch

(1998). Legal cynicism and (sub cultural) tolerance of deviance: The neighborhood context of racial differences. Law & Society Review, 32, 777–803.

57.

Schafer

J. A.

Huebner

M. B.

Bynum

S. T.

(2003). Citizen perceptions of police services: Race, neighborhood context, and community policing. Police Quarterly, 6, 440–468.

58.

Sims

Hooper

Peterson

(2002). Determinants of citizens’ attitudes toward police—Results of the Harrisburg Citizen survey—1999. Policing: An International Journal of Police Strategies & Management, 25, 457–471.

59.

Skogan

W. G.

(1990). Disorder and incline: Crime and the spiral of decay in American neighborhoods. Berkeley, CA: University of California Press.

60.

Sprott

J. B.

Doob

A. N.

(2009). The effect of urban neighborhood disorder on evaluations of the police and courts. Crime and Delinquency, 55, 339–362.

61.

Sunshine

Tyler

R. T.

(2003). The role of procedural justice and legitimacy in shaping public support for policing. Law & Society Review, 37, 513–548.

62.

Tankebe

(2009). Public cooperation with the police in Ghana: Does procedural fairness matter? Criminology, 47, 1265–1293.

63.

Tyler

R. T.

(2011). Trust and legitimacy: Policing in the USA and Europe. European Journal of Criminology, 8, 254–266.

64.

Tyler

T. R.

Wakslak

C. J.

(2004). Profiling and police legitimacy: Procedural justice, attributions of motive, and acceptance of police authority. Criminology, 42, 253–281.

65.

Weitzer

Tuch

S. A.

(1999). Race, class, and perceptions of discrimination by the police. Crime and Delinquency, 45, 494–507.

66.

Weitzer

Tuch

S. A.

(2002). Perceptions of racial profiling: Race, class, and personal experience. Criminology, 40, 435–456.

67.

Weitzer

Tuch

S. A.

(2005). Determinants of public satisfaction with the police. Police Quarterly, 8, 279–297.

68.

Weitzer

R. W.

Tuch

S. A.

(2006). Race and policing in America: Conflict and reform. Cambridge, NY: Cambridge University Press.

69.

Wilson

J. Q.

Kelling

G. E.

(1982). Broken windows: The police and neighborhood safety. Atlantic Monthly, 249, 29–38.

70.

Sun

I. Y.

Triplett

R. A.

(2009). Race, class, or neighborhood context: Which matters more in measuring satisfaction with police? Justice Quarterly, 26, 125–156.

71.

Zevitz

Rettammel

(1990). Elderly attitudes about police service. American Journal of Police, 9, 25–39.

72.

Zhao

Scheider

Thurman

(2002). The effect of police presence on public fear reduction and satisfaction: A review of the literature. The Justice Professional, 15, 273–299.