1035306

research-article2021
EUC0010.1177/14773708211035306European Journal of CriminologyYesberg et al.

Article

European Journal of Criminology

Police visibility, trust in

1­–26
© The Author(s) 2021

police fairness, and collective Article reuse guidelines:

efficacy: A multilevel sagepub.com/journals-permissions
https://doi.org/10.1177/14773708211035306
DOI: 10.1177/14773708211035306
Structural Equation Model journals.sagepub.com/home/euc

Julia Yesberg
University College London, UK

Ian Brunton-Smith
University of Surrey, UK

Ben Bradford
University College London, UK

Abstract
Areas high in collective efficacy – where residents know and trust one another and are willing
to intervene to solve neighbourhood problems – tend to experience less crime. Policing is
thought to be one antecedent to collective efficacy, but little empirical research has explored
this question. Using three waves of survey data collected from London residents over three
consecutive years, and multilevel Structural Equation Modelling, this study tested the impact of
police visibility and police–community engagement on collective efficacy. We explored direct
effects as well as indirect effects through trust in police. The findings showed levels of police
visibility predicted trust in police. Trust in police fairness, in turn, predicted collective efficacy.
There was a small indirect relationship between police visibility and collective efficacy, through
trust in police fairness. In other words, police presence in neighbourhoods was associated with
more positive views about officer behaviour, which in turn was associated with collective efficacy.
The findings have important implications for policies designed to build stronger, more resilient
communities.

Keywords
Collective efficacy, community policing, police presence, procedural justice

Corresponding author:
Julia Yesberg, Jill Dando Institute for Security and Crime Science, University College London, 35 Tavistock
Square, London, WC1H 9EZ, UK.
Email: [email protected]
2 European Journal of Criminology 00(0)

Neighbourhoods in which residents know and trust one another and are motivated to take
collective action experience fewer crime problems. Decades of research across different
contexts has confirmed this finding: when neighbourhoods are higher in collective effi-
cacy – a construct that relates to social ties among neighbours, combined with a willing-
ness to intervene to solve local problems – crime tends to be lower (Armstrong et al.,
2015; Burchfield and Silver, 2013; Gerell and Kronkvist, 2017; Mazerolle et al., 2010;
Sampson et al., 1997; Sampson and Wikström, 2008; Weisburd et al., 2020). Collectively
efficacious neighbourhoods also experience other benefits, such as lower levels of fear
of victimization and perceived disorder (Brunton-Smith et al., 2014), better overall
health (Browning and Cagney, 2002), a lower frequency of bullying in schools (Williams
and Guerra, 2011), and decreased levels of partner violence (Wright and Benson, 2011).
But, despite these benefits, little research has examined how collective efficacy is gener-
ated and sustained over time (Hipp and Wickes, 2017; Wickes et al., 2013).
Policing is thought to be one factor shaping collective efficacy within neighbour-
hoods, but how and why police activity and officer behaviour are expected to facilitate
and encourage collective efficacy is not well understood (Kochel and Gau, 2019;
Sargeant, 2017; Sargeant et al., 2013). Three mutually compatible mechanisms have
been proposed in the literature: (1) that trust in police fosters collective efficacy
(Drakulich and Crutchfield, 2013; Kubrin and Weitzer, 2003; Silver and Miller, 2004);
(2) that police legitimacy encourages collective efficacy (Kochel, 2012; LaFree, 1998);
and (3) that certain types of policing strategies help build collective efficacy (Kochel and
Weisburd, 2019; Renauer, 2007; Sargeant et al., 2013; Scott, 2002). However, there is a
lack of empirical research – in particular, research that uses a longitudinal design – test-
ing these mechanisms. Understanding the mechanisms through which policing might
facilitate collective efficacy would go some way to helping design policies to aid neigh-
bourhoods in developing better defences against crime.
In this study, we use three waves of repeated cross-sectional survey data from neigh-
bourhoods in London, UK. Although not a true longitudinal design, by focusing attention
at the neighbourhood level (with different residents providing assessments of the same
areas each year) we are able to examine the dynamic relationship between police activity
and collective efficacy. We use multilevel Structural Equation Modelling to explore the
relationship between police visibility, police–community engagement and collective
efficacy, and potential indirect effects through trust in police. The article proceeds as fol-
lows. First, we outline the concept of collective efficacy, and then we discuss the theo-
retical and empirical evidence base linking policing and collective efficacy. We next
introduce the rationale and research questions guiding the current study before proceed-
ing with methods, results and discussion.

Collective efficacy
Within the criminological literature, collective efficacy is defined as ‘the process of acti-
vating or converting social ties among neighbourhood residents in order to achieve col-
lective goals, such as public order or the control of crime’ (Sampson, 2010: 802).
Collective efficacy is a combination of social cohesion and trust among neighbourhood
residents, along with shared expectations for social control. Neighbourhoods are thought
Yesberg et al. 3

to be high in collective efficacy when residents know and trust one another and are will-
ing to act to address local problems – for example, to break up fights or intervene if
children are skipping school. Research findings across a range of contexts have shown
that, when collective efficacy is high in neighbourhoods, crime tends to be lower
(Armstrong et al., 2015; Burchfield and Silver, 2013; Gerell and Kronkvist, 2017;
Mazerolle et al., 2010; Weisburd et al., 2020; Wikström et al., 2012). Collective efficacy
has also been shown to mediate the relationship between concentrated disadvantage and
crime (for example, Browning et al., 2004).
Collective efficacy is considered to be an important factor explaining why some
neighbourhoods with predisposing structural characteristics, such as concentrated
disadvantage, ethnic heterogeneity and residential mobility, experience high crime
rates, whereas others do not. According to social disorganization theory, crime occurs
when regulatory mechanisms – for example social organization and neighbourhood
attachment – collapse (Shaw and McKay, 1942). When neighbours do not know or
trust one another, or are unwilling to intervene when incidents occur, crime increases
(Sampson et al., 1997). However, when neighbours have opportunities to interact with
one another, trust and solidarity increase and neighbours develop and conform to
shared norms for behaviour. Neighbourhood residents increasingly act to defend these
norms, including intervening to solve neighbourhood problems, ultimately leading to
a reduction in crime and disorder (Sampson et al., 1997).
Collective efficacy is considered to be an attribute of neighbourhoods rather than of
individuals. Yet the construct is typically measured using individual ratings by survey
respondents, which are then aggregated to the neighbourhood level. In the literature,
there are numerous debates over the measurement and conceptualization of collective
efficacy. For example, although Sampson and colleagues (1997) originally conceptual-
ized collective efficacy as a single construct, many studies treat social cohesion and
informal social control as separate constructs in analysis. Some evidence has shown the
two are not always highly correlated (Horne, 2004), they do not consistently load onto a
single factor (Armstrong et al., 2015; Gau, 2014; Wickes et al., 2013), and the causal
ordering of the constructs, and downstream variables such as crime rates, might vary
across contexts (Rhineberger-Dunn and Carlson, 2011). There is also debate over the
measurement of informal social control: for example, whether it should be measured
using perceptions or actions; whether it should relate to the actions individuals report
they would undertake as opposed to what their neighbours would do; and whether these
actions are direct (for example, intervening themselves to solve the problem) or indirect
(for example, seeking help from outside agencies) (Gau, 2014; Warner, 2007).

Policing and collective efficacy

Most of the research on collective efficacy has focused on its outcomes rather than its
causes. However, given the widespread benefits of collective efficacy, it is important to
understand the factors that might generate and sustain collective efficacy over time.
Policing is thought to be one such factor, but the role of police in building collective
efficacy has not been adequately explored empirically. Three potential mechanisms have
4 European Journal of Criminology 00(0)

been proposed in the literature. These are not mutually incompatible but do differ in
more or less nuanced ways.
First, some scholars have suggested that trust in the police fosters collective efficacy,
particularly across dimensions of fairness and effectiveness. When residents view the
police as a capable and effective resource, and believe officers exercise their authority in
a fair and just manner, they may be more inclined to take collective action to address
neighbourhood problems and may feel more empowered ‘to intervene when confronted
with local acts of deviance’ (Silver and Miller, 2004: 558). On the other hand, when resi-
dents do not feel the police are a viable resource, they may feel too vulnerable to inter-
vene in neighbourhood issues, because they may see their own actions as both less
effective and more risky (Drakulich and Crutchfield, 2013; Kubrin and Weitzer, 2003).
Second, researchers have suggested that the police may facilitate collective efficacy
through their legitimacy (LaFree, 1998). Legitimacy refers to the extent to which people
believe the police behave in an appropriate manner and feel a normatively grounded
obligation to obey the police (Jackson et al., 2013). Through their role as moral guardi-
ans, the police construct and enforce shared norms and values and provide guidance
about appropriate and inappropriate behaviour (Kochel, 2012; LaFree, 1998; Triplett
et al., 2003). When residents do not see the police as legitimate, the validity and force of
these mutually shared norms and values diminishes, and residents are less willing to
cooperate with police, less willing to grant police discretion, and even less likely to obey
the law (Jackson et al., 2013; Sunshine and Tyler, 2003; Tyler and Fagan, 2008; Van
Damme et al., 2013). Similarly, when police lack legitimacy, neighbourhood social pro-
cesses may break down, such that residents ‘struggle to develop a working trust and so
cannot be confident that other neighbours will act in the best interest of the neighbour-
hood’ (Kochel, 2012: 389).
The third proposed mechanism is that certain types of policing strategies will increase
collective efficacy within communities. Community or neighbourhood policing is a law
enforcement approach that emphasizes community involvement in crime prevention and
seeks to increase contact between police and local residents (Gill et al., 2014). Community
policing scholars argue that, if crime is the result of social disorganization, policing strate-
gies should seek to build and sustain vital social processes within neighbourhoods
(Rosenbaum, 1987; Skogan, 1990). Community policing is expected to increase collec-
tive efficacy by providing more opportunities for residents to interact with one another, by
increasing access to police resources, and by stimulating ‘self-help’ within communities
(Renauer, 2007; Sargeant et al., 2013; Scott, 2002). Another aspect of community policing
– police presence or visibility – is also thought to contribute to collective efficacy through
reassuring residents of safety and reducing fear of crime, allowing them to confidently
engage in their own informal social control behaviours (Kochel and Weisburd, 2019).
Variables underlying these three mechanisms are likely to be associated with one
another in multiple ways. Most pertinently for the current purposes, a key aim of com-
munity policing is very often to enhance public trust and legitimacy by making police
more present, visible and engaged in local areas and thus strengthening relationships
with residents (Skogan, 2019). There is significant evidence to support this idea. Police
visibility has been shown to be a consistent positive predictor of trust in the police, in the
UK at least (Bradford et al., 2009; Sindall and Sturgis, 2013), and studies have shown
Yesberg et al. 5

that engagement with communities via information provision can have a direct effect on
public trust (Hohl et al., 2010). Since trust seems a more proximate predictor of the
actions implied by the notion of collective efficacy than, for example, the flow of infor-
mation between police and public, any effect of community policing on collective effi-
cacy may be partly or even fully mediated by trust (and/or legitimacy).
A recent rapid evidence assessment (REA) (Yesberg and Bradford, 2021) reviewed
the literature on policing and collective efficacy. Overall, trust in police was the aspect of
policing most consistently associated with collective efficacy. There was also some evi-
dence that community policing activities, such as visibility and community engagement,
predicted collective efficacy. By contrast, police legitimacy was largely unrelated to col-
lective efficacy. Over half (54 percent) of the 39 studies reviewed tested the impact of
collective efficacy on measures of policing, such as trust and legitimacy, rather than the
other way around, arguing that neighbourhood context and concerns about social order
play an important role in shaping attitudes toward the police (for example, Jackson et al.,
2013; Jackson and Sunshine, 2007; Nix et al., 2015). Of course, in reality, any associa-
tion between policing and collective efficacy is likely to be bi-directional. Furthermore,
the vast majority of studies reviewed used cross-sectional designs, which the limits the
conclusions that can be drawn about causal processes.
Only a handful of studies in the REA used a longitudinal design to test relationships
between policing and collective efficacy. Kochel and Gau (2019) used a panel commu-
nity survey from the US, with three waves of data collection, to test the impact of percep-
tions of police visibility, police–community engagement and satisfaction with police
tactics on perceptions of informal social control and social cohesion. They found that
satisfaction with police visibility, satisfaction with police tactics and police–community
engagement (at wave 1) were significant predictors of social cohesion (at wave 2), and,
through social cohesion, these policing measures indirectly predicted informal social
control (wave 3). There were no significant direct effects of the policing measures on
informal social control, and perceptions of police visibility (as opposed to satisfaction
with visibility) were not a significant predictor of social cohesion or informal social
control.
In another study, Kochel (2018) used three waves of resident surveys in Trinidad and
Tobago to explore how legitimacy and police competence influence collective efficacy
and violent victimization. Police legitimacy did not produce greater collective efficacy
(or vice versa), but there was a direct positive relationship between police competence (a
measure including effectiveness, procedural justice and trust) and collective efficacy.
Kochel (2018) concluded that the way police are perceived to act is more important for
neighbourhood social processes than people’s general impressions of the policing insti-
tution. In other words, how people perceive the actions of police officers and organiza-
tions may be more enabling of their engagement in informal social control than any sense
of duty toward the police generated by legitimacy. This idea fits with the overall findings
from the REA, which showed police legitimacy had only a weak relationship with col-
lective efficacy.
Taken together, prior research suggests that perceptions of the police, particularly
perceptions linked to the action of individual officers (that is, trust in fairness and effec-
tiveness), are associated with collective efficacy. When people feel the police are a
6 European Journal of Criminology 00(0)

Figure 1.  Conceptual model.

trustworthy and effective resource, they may be more inclined to take collective action.
There is also some evidence that aspects of community policing enhance collective effi-
cacy: police visibility and police–community engagement significantly predict collec-
tive efficacy. There is much to suggest, therefore, that policing has a potential role to play
in helping neighbourhoods develop the vital community social processes needed to pro-
tect against crime.

Current study
The current study uses three waves of repeated cross-sectional survey data from neigh-
bourhoods in London to test the conceptual model shown in Figure 1. First, we expect
that trust in police fairness and effectiveness will be important predictors of collective
efficacy. Second, given that aspects of community policing have been found to be impor-
tant in encouraging collective efficacy, particularly visibility and engagement, we also
test direct relationships between police visibility and police–community engagement and
collective efficacy, as well as indirect pathways through trust in police. We use multilevel
Structural Equation Models to test these relationships. We measure police visibility and
police–community engagement at wave 1, trust in the police at wave 2 and collective
efficacy at wave 3, and include lagged (wave 1) controls for trust and collective
efficacy.

Method
Data
Data are drawn from a large sample of 40,080 respondents to the Mayor’s Office for
Policing and Crime’s (MOPAC) Public Attitude Survey (PAS) who were interviewed
Yesberg et al. 7

Table 1.  Demographic characteristics of sample.

Characteristic Wave 1 Wave 2 Wave 3

Percenta N Percent N Percent N

Gender Male 45.75 6595 45.84 5884 45.55 5838
Female 54.25 7820 54.16 6952 54.45 6978
Age range 16–24 10.92 1566 11.49 1472 10.26 1312
25–34 20.04 2874 19.23 2463 20.45 2615
35–64 47.74 6848 47.68 6108 47.78 6109
65+ 21.30 3055 21.60 2767 21.51 2750
Ethnicity White 67.14 9528 67.54 8267 67.89 8268
Black 7.90 1090 8.02 982 7.32 891
Asian 17.53 2417 16.63 2036 16.96 2065
Mixed 2.75 379 2.41 295 2.83 345
Other 4.68 646 5.39 660 5.00 609
Country of UK 55.58 8014 56.71 7282 57.67 7394
birth Non-UK 44.42 6405 43.29 5558 42.33 5427
Employment Employed 55.36 7982 55.01 7063 55.22 7080
status Unemployed 2.39 345 2.39 307 1.79 229
Student 6.39 922 6.78 870 6.17 791
Other 35.86 5170 35.83 4600 36.82 4721

a. Percentages calculated with missing values excluded.

between October 2014 and September 2017 (see Table 1 for sample characteristics). The
PAS is conducted on a rolling basis and includes a representative sample of residents
from across London. Approximately 3200 Londoners are interviewed face-to-face each
quarter at pre-selected addresses aiming to achieve 100 interviews in each of the 32
London boroughs. The PAS asks about people’s experiences of crime and anti-social
behaviour and, pertinent to the current study, includes questions about perceptions of
community policing, trust in the police and collective efficacy. The dataset also contains
electoral ward identifiers, making it possible to look at area-level effects. Wards are the
basic unit of electoral geography in London, with an average size of 3 km2 and an aver-
age population of around 13,800 in 2015.
Some 629 wards are represented in the dataset, with an average of 64 respondents per
ward (ranging from 12 to 116 respondents). To investigate potential causal effects, we
split the data into three waves (wave 1: October 2014 – September 2015; wave 2: October
2015 – September 2016; wave 3: October 2016 – September 2017) and use multilevel
Structural Equation Models to connect levels of police visibility and police–community
engagement (wave 1) to trust in police fairness and effectiveness (wave 2), to collective
efficacy (wave 3). Importantly, because different respondents complete the survey at
each wave, these pathways (and hence our focus of inference) are estimated at the ward
level.1 At the individual level, correlations between the three waves must be fixed to 0,
since respondents are interviewed only once.2
8 European Journal of Criminology 00(0)

Measures
Collective efficacy. Collective efficacy was measured using eight items on a five-point
agree/disagree scale. Four items measured respondents’ perceptions of the social cohe-
siveness of their neighbourhood, including: ‘people in this neighbourhood can be trusted’
and ‘people act with courtesy to each other in public space in this area’; and four items
measured their perceptions of informal social control, including: ‘local people and
authorities have control over the public space in this area’ and ‘if I sensed trouble whilst
in this area, I could get help from people who live here’ (α = .81 for all eight items; see
Table 4 in the Appendix for the other item wordings).

Trust in police fairness.  Trust in police fairness was measured with seven items asking
residents their agreement with statements about police in their area (on a five-point
agree/disagree scale), including: ‘they can be relied on to be there when you need them’
and ‘the police in this area treat everyone fairly regardless of who they are’ (α = .89; see
Appendix Table 4 for the other item wordings).

Trust in police effectiveness.  Trust in police effectiveness was also measured using seven
items. Respondents were asked – on a seven-point scale from very well (1) to not at all
well (7) – how well they think the Metropolitan Police: prevent terrorism; respond to
emergencies promptly; provide a visible patrolling presence; tackle gun crime; support
victims and witnesses; tackle drug dealing and drug use; and tackle dangerous driving (α
= .86).

Police visibility and police–community engagement.  We include two items capturing aspects
of community policing. The first item measures police visibility on a six-point ordinal
scale, asking respondents how often they see police patrolling on foot, bicycle or horse-
back in their area (defined as within 15 minutes’ walk from their home) from: at least
daily (1) to never (6):3 29 percent of respondents said they saw police patrolling at least
daily or weekly; 24 percent said they saw police at least fortnightly or monthly; and 46
percent said they saw police less often or never. The second variable measured police–
community engagement on a three-point ordinal scale. Respondents were asked ‘how
well informed do you feel about what the police in this area have been doing over the last
12 months?’ and responded on a three-point scale from very well informed (1) to not at
all informed (3). Only 6 percent of respondents said they felt very well informed about
what the police in their area are doing; 40 percent felt fairly well informed; and 55 per-
cent felt not at all informed.

Neighbourhood controls.  Finally, we control for two aspects of neighbourhood context in

the models: crime rate and deprivation. Crime rate was measured at the ward level using
police-recorded crime data from 2014/15 and represents the rate of total crime per 1000
population (the logged rate of crime is used in all analyses). Deprivation was measured
using the 2015 Index of Multiple Deprivation (IMD) (Noble et al., 2004). IMD is a meas-
ure of unique deprivation at a small local area level (Lower-layer Super Output Areas;
LSOA) across England. IMD scores are based on seven domains of deprivation: income,
Yesberg et al. 9

Table 2.  Descriptive statistics of measures at the ward level.

Measure Range Mean SD

Endogenous constructs (latent variables)
  Collective efficacy (wave 3) −1.37 to 2.38 0 0.47
  Fairness (wave 2) −1.19 to 3.04 0 0.61
  Effectiveness (wave 2) −1.60 to 2.75 0 0.64
Exogenous variables (latent mean decomposition)
  Visibility (wave 1) 1 to 6 0 1.68
  Community engagement (wave 1) 1 to 3 0 0.60
Neighbourhood controls
  Crime rate (2014/2015) 24.50 to 1212.13 82.27 71.15
  Index of multiple deprivation (IMD) 2015 4.85 to 52.88 23.35 10.24

employment, education, health, crime, barriers to housing and services, and living envi-
ronment. We calculated ward-level IMD scores by using the average score for LSOAs
within each ward (see Table 2 for descriptive statistics).

Analytic strategy
We use multilevel Structural Equation Models (Muthén and Muthén, 2012) to specify the
ward-level causal pathways outlined in Figure 1 between police visibility and police–
community engagement (measured at wave 1), trust in police fairness and police effec-
tiveness (wave 2), and collective efficacy (wave 3). This is an extension to standard
Structural Equation Modelling – combining a latent variable modelling framework with
directed regression pathways – that correctly distinguishes the variation that occurs
within each ward from the variations that occur between wards. Two multilevel measure-
ment models (CFA) are first specified for collective efficacy (wave 3) and for trust in
police fairness and effectiveness (wave 2).4 These have a within-ward component, which
is the estimated factor structure derived from the correlations between the identified
survey items at the individual level. We use this within-ward factor model to identify the
between-ward component, allowing the estimated intercepts for each item from this fac-
tor model to vary randomly across wards, with the associations between these varying
intercepts used to define the between-ward factor structure. Following Jak (2019), the
ward-level factor loadings are constrained to equal the within-ward loadings to ensure
that the ecological measures of each construct correctly represent ward-level aggregates
of the scores for each resident. The between-ward factors can then be connected using
directed pathways such as a standard Structural Equation Model.5 Factor loadings, ICCs
and model fit are reported in Tables 5 and 6 in the Appendix.6
The two single-item exogenous predictors – police visibility and police-community
engagement – measured at wave 1 are included at the ward-level using the latent mean
decomposition specification (Muthén and Muthén, 2012). This separates respondent-
level differences in ratings of each item from the ward mean, with the ward mean score
then used as a predictor in the ward-level model. Importantly, by deriving these ward-
level scores during the model estimation process, this corrects for unreliability at the
10 European Journal of Criminology 00(0)

ward-level and results in unbiased estimates of the ward-level effects by incorporating

the uncertainty associated with the within-ward individual deviations (Lüdtke et al.,
2008).7 Table 2 presents descriptive statistics at the ward level for the latent variables
measured at waves 2 (trust in police fairness and effectiveness) and 3 (collective effi-
cacy), the ward-level mean decompositions at wave 1 (police visibility and police–com-
munity engagement) and the neighbourhood controls.
At the ward level we estimate directional pathways between the ward-level mean
decompositions (wave 1) and the ward-level latent variables measured at waves 2 and 3.
Here collective efficacy measured at the final time point is regressed on trust in police
fairness and trust in police effectiveness measured at wave 2. In addition, collective effi-
cacy, police fairness and police effectiveness are all regressed on the latent mean of
police visibility and police–community engagement from wave 1. Finally, we also
include ward-level lagged controls for collective efficacy, police fairness and police
effectiveness measured at wave 1 (see Table 7 in the Appendix for the CFA results for the
lagged controls). Accounting for prior levels of each construct in each ward reduces the
potential impact of temporal autocorrelation, reflecting the high degree of stability
exhibited by each construct over time (models without these lagged control variables
demonstrated the same substantive results – compare Models 1 and 2 in Table 3 –
although the magnitude and significance of some directional pathways were larger).8
Here we constrain the factor loadings for each latent variable to be equal at each time
point to ensure we are consistently measuring each construct (models freely estimating
the factor structure at each time point showed the same result – Table 8 in the Appendix,
Model 2A).
Although directional pathways are estimated at the ward level across the three waves
of data, to enable identification of the model, at the within-ward level the associations
between collective efficacy, police fairness and effectiveness are fixed at 0. Similarly, the
associations between the latent constructs and the observed within-ward measures of
police visibility and community engagement are also fixed at 0. Fixing these pathways
to 0 reflects the fact that the underlying individual data are cross-sectional, with all
respondents interviewed at only a single time point. Conversely, the within-ward correla-
tion between effectiveness and fairness (measured at the same time point) is freely esti-
mated. The model is estimated using a robust maximum likelihood approach (MLR)
using Mplus version 7.11, which is robust to non-normally distributed data (Muthén and
Muthén, 2012).

Results
Findings from the multilevel Structural Equation Model are presented in Figure 2 and
Table 3 (Model 2).9 First, as the figure shows, police visibility has a significant associa-
tion with both trust in police fairness and police effectiveness. Police–community
engagement, on the other hand, has a significant association only with police effective-
ness at the p < .10 level. Second, trust in police fairness has a significant association
with collective efficacy, whereas trust in police effectiveness does not. Third, neither
police visibility nor police–community engagement have direct associations with collec-
tive efficacy.
Yesberg et al. 11

Figure 2.  Multilevel Structural Equation Model of policing measures on collective efficacy
(Model 2).
Note: Standardized coefficients. Lagged effects omitted for visual ease.

Examining the sequential processes, there was a small indirect effect of police visibil-
ity on collective efficacy, through trust in police fairness, but this effect was only signifi-
cant at the p < .10 level (B = .039, SE = .020, p = .055). These findings suggest that,
when police are more visible in neighbourhoods, this leads to more positive perceptions
from residents about officer behaviour (that is, the fairness with which the police treat
residents), which in turn leads to more collectively efficacious neighbourhoods.
Looking last at our neighbourhood controls, the logged crime rate has a significant
negative association with trust in police effectiveness (Table 3). In other words, in areas
with lower rates of crime, people perceived the police to be more effective. Deprivation,
on the other hand, has a significant positive association with trust in police effectiveness
and a significant positive association with collective efficacy. Areas with more depriva-
tion are more likely to have high social cohesion and informal social control. Overall, the
model explained approximately 36 percent of the variance in collective efficacy (R2 =
.358) and 9 percent of the variance for both trust in police fairness (R2 = .088) and effec-
tiveness (R2 = .086).

Discussion
This study used a longitudinal design and multilevel Structural Equation Modelling to
test the impact of police visibility and police–community engagement, and trust in
12 European Journal of Criminology 00(0)

Table 3.  Multilevel Structural Equation Models.

Model 1: Structural variables Model 2: Temporal controls

B SE Sig. Beta B SE Sig. Beta

Collective efficacy
Fairness 0.705 0.184 <.001 0.521 0.592 0.180 .001 0.454
Effectiveness −0.241 0.145 .095 −0.218 −0.181 0.147 .218 −0.168
Visibility −0.015 0.027 .578 −0.037 −0.009 0.027 .733 −0.023
Community engagement 0.193 0.129 .134 0.112 0.056 0.131 .425 0.033
Crime rate (log) −0.011 0.029 .709 −0.024 −0.004 0.029 .891 −0.009
Deprivation 0.010 0.001 <.001 0.490 0.007 0.001 <.001 0.364
Collective efficacy (wave 1) 0.287 0.064 <.001 0.317
Fairness
Visibility 0.073 0.027 .007 0.239 0.066 0.027 .014 0.219
Community engagement 0.223 0.120 .063 0.175 0.179 0.124 .151 0.141
Crime rate (log) −0.024 0.024 .314 −0.071 −0.023 0.024 .325 −0.069
Deprivation 0.001 0.001 .544 0.040 0.000 0.001 .840 0.014
Fairness (wave 1) 0.136 0.092 .137 0.125
Effectiveness
Visibility 0.074 0.029 .012 0.198 0.073 0.029 .013 0.199
Community engagement 0.268 0.141 .057 0.172 0.263 0.139 .058 0.171
Crime rate (log) −0.055 0.023 .018 −0.133 −0.054 0.023 .017 −0.134
Deprivation 0.002 0.001 .036 0.126 0.002 0.001 .037 0.126
Effectiveness (wave 1) −0.012 0.079 .879 −0.011
Level 2 variance
Fairness 0.021 0.003 0 0.021 0.003 0  
Effectiveness 0.032 0.004 0 0.031 0.004 0  
Collective efficacy 0.026 0.003 0 0.025 0.003 0  
Visibility 0.249 0.020 0 0.249 0.020 0  
Community engagement 0.014 0.002 0 0.014 0.002 0  
Level 1 variance
Fairness 0.435 0.014 0 0.440 0.012 0  
Effectiveness 0.474 0.022 0 0.458 0.017 0  
Collective efficacy 0.279 0.011 0 0.290 0.010 0  
Visibility 2.507 0.029 0 2.507 0.029 0  
Community engagement 0.361 0.005 0 0.361 0.005 0  
Model fit
RMSEA 0.012 0.010  
TLI 0.954 0.937  
CFI 0.956 0.938  

police, on collective efficacy. Using three waves of repeated cross-sectional survey data
collected from London residents over three consecutive years, we tested for direct and
indirect effects, with trust in police fairness and effectiveness as the intermediary
Yesberg et al. 13

constructs. First, we found that police visibility was a significant predictor of both trust
in police fairness and effectiveness at the neighbourhood level. Trust in the police
tended to be higher in wards where police visibility was also higher. This finding fits
with previous research showing the significant effect police visibility can have on per-
ceptions that the police are ‘doing something’ to control crime (Hawdon et al., 2003:
481) and to their overall judgements of trust, satisfaction and confidence in police
(Innes and Innes, 2011; Sindall and Sturgis, 2013).
Police–community engagement, on the other hand, was associated with effectiveness
judgements only at the p < .10 level. Although we were limited in this study by our
measure of engagement, other research has shown that police presence has a stronger
effect on trust and confidence than do other community policing variables such as knowl-
edge about police tactics and police engagement with local businesses and community
groups (Hawdon et al., 2003; Kochel and Gau, 2019). We know that, when people view
police as trustworthy, legitimate authorities, they are more likely to comply and cooper-
ate with the structures and rules the institution represents and to engage in other pro-
social behaviours (Bolger and Walters, 2019; Jackson et al., 2013; Tyler and Huo, 2002;
Walters and Bolger, 2019). Therefore, teasing apart the types of policing activities that
generate trust has important implications for crime prevention. Our findings suggest that
visibility is one important aspect of this equation.
Our second main finding was that trust in police fairness significantly predicted col-
lective efficacy. When residents believed the police exercised their authority in a fair and
just manner, perceptions of collective efficacy were higher. Trust in police effectiveness,
on the other hand, was not significantly related to collective efficacy, and, in fact, the
coefficient between these variables was negative. This finding is in contrast with research
from the UK showing that measures of police effectiveness are significantly associated
with collective efficacy judgements (for example, Jackson et al., 2013; Jackson and
Sunshine, 2007).
An argument proposed by some scholars is that, when the police are seen to be effec-
tive, residents will be less likely to exercise informal social control themselves because
they believe the police are capable of dealing with local issues on their own (Silver and
Miller, 2004). This idea is rather similar to the bystander effect in the psychological lit-
erature, whereby the likelihood of intervening decreases as the number of other people
present increases (Fischer et al., 2011). The bystander effect is due, in part, to a diffusion
of responsibility (that is, people assume that others are responsible for taking action);
similarly, when people feel the police are effective, they may believe the police will act
for them to address issues of crime and disorder. However, when the police are seen as
ineffective, this may actually encourage informal social control because residents assume
the responsibility for instilling order in their communities (Kochel, 2018; Kubrin and
Weitzer, 2003). This idea has some empirical support; for example, Schaefer and col-
leagues (2017) found that police effectiveness was negatively associated with taking
action to solve local crime problems. The negative (although not significant) coefficient
between effectiveness and collective efficacy provides some small further support for
this hypothesis, but more research is clearly needed to understand how perceptions of
police effectiveness influence residents’ willingness to take action.
14 European Journal of Criminology 00(0)

Finally, we found a small indirect effect of police visibility on collective efficacy,

through trust in police fairness. When the police are more present in neighbourhoods,
perceptions of fairness are higher, which, in turn, enhances perceptions of collective
efficacy. It seems it is not enough for the police to be a visible presence; people also
need to feel they can trust the police to exercise their authority in a fair and just manner.
The implications of these findings seem to be that police can maximize their influence
on collective efficacy through not only making themselves more visible to residents, but
by treating them fairly and respectfully, and by giving them voice in their interactions
with them.
That it is community understandings of police fairness, not effectiveness, that appear
to be important in generating collective efficacy means that the results presented here
resonate with the wider literature on procedural justice. At the individual level, people
are more ready to cooperate with the police when they feel that officers behave in a pro-
cedurally just fashion (Bolger and Walters, 2019). We find that at the neighbourhood
level this extends to more general efforts to intervene and maintain order. There are two
mutually compatible interpretations of this finding. First, believing that the police oper-
ate in a fair and just manner may provide reassurance to residents that, should they inter-
vene in a particular situation, the police will support them in an appropriate manner.
Second, research has shown that perceptions and experiences of procedural justice, spe-
cifically, are linked to feelings of security and belonging within wider society (Bradford,
2014; Murphy et al., 2015). It may therefore be that fair policing strengthens social
bonds within communities and neighbourhoods by fostering a sense of collective inclu-
sion within wider social structures and, therefore, collective efficacy.
Considering the two control variables in our analysis, the association between crime
rates, trust and collective efficacy was much as expected: when crime was higher, for
example, trust in police effectiveness was lower in following years. However, the posi-
tive association between deprivation and collective efficacy is at first sight puzzling. A
key aspect of Sampson and colleagues’ conceptualization of collective efficacy is that
it is the mechanism linking deprivation to crime: more deprived neighbourhoods are
less able to organize themselves to combat crime problems because they lack the eco-
nomic and social capital required (for example, Sampson, 2012). Yet, in these London
data, we find that, all else equal, more deprived areas had greater collective efficacy in
following years.
The answer to this puzzle may partly lie in the residential mobility patterns seen in
London and indeed elsewhere in the UK. Around 10 percent of households in London
move each year (GLA, 2020). Although unemployment and other markers of depriva-
tion are associated with greater mobility (Gambaro and Joshi, 2017), studies have
shown that, when moving home, people living in social housing, for example, tend to
move less far, whereas people with more education move further (Langella and
Manning, 2019). Furthermore, an important feature of the London property market spe-
cifically is the number of properties in the richest parts of town bought by overseas
investors and then left empty (for example, Transparency International UK, 2017),
meaning many such areas have relatively few permanent residents. These and other fac-
tors may combine to mean that people living in some of the more deprived parts of
London are more likely to know their neighbours and feel they can get help from other
Yesberg et al. 15

residents if they need it – key items used in the collective efficacy scale used in this
article and many other studies.
Of course, this study has a number of limitations, most notably the limited measures
available in the dataset, particularly those representing different aspects of community
policing. We had no measures of important community policing activities, such as prob-
lem-solving or partnership working. Our only measure of police engagement was
whether people knew the priorities of their local policing team. Although knowledge of
these priorities could represent active engagement by the police to ensure residents are
kept informed of their work, other measures would more effectively capture police–
community engagement. For example, Kochel and Gau’s (2019) police engagement
measure asked residents whether the police met and worked with local businesses or resi-
dents to address crime and other problems. There was also no measure of legitimacy in
the dataset, so we were unable to test whether trust in individual officers is more or less
important for fostering collective efficacy than perceptions of the legitimacy of the polic-
ing institution (Kochel, 2018; Yesberg and Bradford, 2021). Future research should seek
to replicate these findings using better measures of community policing, and there should
be a more thorough investigation into potential intervening variables (for example,
legitimacy).
In addition, we were limited to data from repeated cross-sections of London residents,
rather than a fully longitudinal design where the same respondent views are tracked over
time. Our focus is at the ward level, and as such we assume that sampled residents at each
survey wave are providing unbiased assessments of the true level of collective efficacy
and police activity in their local area. But, without repeated measurements from the same
individuals, the possibility remains that our results are actually a reflection of differences
in the sample of residents measured at each time point. However, we have no reason to
believe sampling within wards was biased, and ICCs for each construct were similar in
magnitude at each wave. Our model specification also required us to fix all within-ward
correlations to 0 over time, reflecting the absence of repeated measurements from
individuals.
To assess the sensitivity of our results to this model specification, we re-estimated all
models using a two-stage process. First, predicted scores derived from single-level CFA
were saved and aggregated to the ward level to generate a set of ward-level characteris-
tics. In a second stage we estimated the path model outlined in our conceptual model
(Figure 1) on this ward-level dataset. We estimated equivalent models using predicted
scores derived at the ward level from multilevel CFA. Results using both specifications
were consistent with the results reported in Table 3. Future research using a true longitu-
dinal design is needed to further corroborate our findings.

Conclusion
This study adds to the literature on policing and collective efficacy. Using three waves of
data from London, and multilevel Structural Equation Modelling, we demonstrate a clear
link between collective efficacy and prior levels of trust in police fairness. Trust was, in
turn, predicted by prior levels of perceived police activity, most notably visible patrolling
16 European Journal of Criminology 00(0)

presence. In line with other studies, we conclude that collective efficacy, at the neigh-
bourhood level, may be enhanced when policing is visible and above all, fair.
Our results have important implications for policy. The London Metropolitan Police
‘Vision and Values’ statements argues that to ‘[t]o make London the safest city it can be
it is clear that everyone needs to play their part in preventing crime’ (MPS, 2018: 14).
More widely, the police and policy-makers around the world speak constantly of the
need to engage citizens in the task of order production and maintenance. Results from
this study underline the importance of fairness: it is this aspect of police behaviour, not
effectiveness, that appears most likely to bolster community engagement in such activity.
Resonating again with the procedural justice literature, it would seem that, in a city such
as London, fairness should be at the top of police priorities, both for their own sake and
because these aspects of police–community relations are so closely linked to outcomes
such as, here, collective efficacy.
This message has often been difficult for police leaders and policy-makers to swal-
low, both because of a widely held belief that effectiveness is more important and
because making policing fairer can seem a nebulous and hard to achieve policy aim.
The link between police presence and trust should therefore provide some reassur-
ance. Neighbourhoods where the police were more visible tended to be significantly
higher in trust (on average – an important proviso is that not all communities will
respond the same way to visible policing); and trust was linked to collective efficacy.
It seems visible patrolling can, perhaps simply by indicating that police are present in
and care about an area, bolster outcomes of significant value to both police and
community.
This is certainly an easy, familiar, policy recipe to follow. It does, though, run some-
what against current fashion in police science and management, which stresses that (a)
undirected police patrol is ineffective in reducing crime and (b) patrols should be tar-
geted toward high-crime locations (Braga et al., 2019). Both these things may very well
be true, but they may also miss an important aspect of police–community relations, in
London at least, which is that people and communities value a visible police presence in
their neighbourhood and that this may encourage them to play an active role in address-
ing threats to social order. Withdrawing police from the majority of neighbourhoods to
focus on a relatively small number of high-crime locations may therefore have negative
longer-term effects. This possibility is poorly understood at the current time, has signifi-
cant resource and other implications for the police and other security actors, and would
be a fruitful topic for future research.

Acknowledgements
The authors would like to thank the Mayor’s Office for Policing and Crime – in particular Dr Paul
Dawson – for access to the data used in this study.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship,
and/or publication of this article.
Yesberg et al. 17

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/
or publication of this article: This work was supported by the first author’s Economic and Social
Research Council (ESRC) New Investigator Grant [ES/S010629/1].

ORCID iDs
Julia Yesberg https://orcid.org/0000-0002-8511-321X
Ian Brunton-Smith https://orcid.org/0000-0002-0077-5602
Ben Bradford https://orcid.org/0000-0001-5480-5638

Notes
1. The vast majority of wards were represented in all three waves (n = 627); however, two
wards were missing wave 1 data and one ward was missing wave 2 data.
2. Addresses are sampled at the start of each year (before the first quarter, January–March) and
the prior year’s selected addresses are excluded from the current year sample. Because our
data span three years, it is possible – but highly unlikely – that a sampled address in wave 1
was resampled in wave 3. Give the large population of Greater London and the number of
possible addresses, the chances of reselecting an address and a person from that address is
extremely low. We do not have address details in our dataset to confirm this.
3. We also had a variable measuring satisfaction with police visibility (1 = satisfied; 0 = not
satisfied), similar to Kochel and Gau (2019), but this variable was strongly correlated with
police visibility (r = −.60) so we chose to retain the latter measure in analysis.
4. We ran two separate measurement models because the trust and collective efficacy measures
were measured at different time periods and thus with different samples of respondents.
5. Additional residual correlations at the individual level between specific items within the same
latent variable were estimated as suggested by modification indices.
6. Comparisons of RMSEA, TFI and CLI between models where the ward-level factor loadings
are freely estimated and constrained to equal the within-ward factor loadings reveal no sub-
stantial change to fit.
7. An additional model estimated using the naive mean score at ward level for each exogenous
predictor showed the same substantive results. See Table 8 in the Appendix, Model 2B.
8. Model fit indices showed there was a fall in model fit with the inclusion of lagged controls.
However, model fit indices may be less reliable because of the specification of the model
(that is, respondents were included at only one time point and various within-ward corelations
were fixed to zero).
9. Note this model includes (1) controls for neighbourhood crime rate and deprivation and (2)
lagged controls for collective efficacy, police fairness and police effectiveness measured at
wave 1 (see Table 3).

References
Armstrong TA, Katz CM and Schnebly SM (2015) The relationship between citizen perceptions
of collective efficacy and neighborhood violent crime. Crime & Delinquency 61(1): 121–142.
Bolger CP and Walters GD (2019) The relationship between police procedural justice, police legit-
imacy, and people’s willingness to cooperate with law enforcement: A meta-analysis. Journal
of Criminal Justice 60: 93–99.
Bradford B (2014) Policing and social identity: Procedural justice, inclusion and cooperation
between police and public. Policing and Society 24(1): 22–43.
18 European Journal of Criminology 00(0)

Bradford B, Jackson J and Stanko EA (2009) Contact and confidence: Revisiting the impact of
public encounters with the police. Policing and Society 19(1): 20–46.
Braga AA, Turchan BS, Papachristos AV et al. (2019) Hot spots policing and crime reduction:
An update of an ongoing systematic review and meta-analysis. Journal of Experimental
Criminology 15: 289–311.
Browning CR and Cagney KA (2002) Neighborhood structural disadvantage, collective efficacy,
and self-rated physical health in an urban setting. Journal of Health and Social Behavior
43(4): 383.
Browning CR, Feinberg SL, Dietz RD et al. (2004) The paradox of social organization: Networks,
collective efficacy, and violent crime in urban neighborhoods. Social Forces 83(2): 503–534.
Brunton-Smith I, Sutherland A and Jackson J (2014) Bridging structure and perception: On the
social ecology of beliefs and worries about neighbourhood violence in London. British
Journal of Criminology 54: 503–526.
Burchfield KB and Silver E (2013) Collective efficacy and crime in Los Angeles neighborhoods:
Implications for the Latino Paradox. Sociological Inquiry 83(1): 154–176.
Drakulich KM and Crutchfield RD (2013) The role of perceptions of the police in informal social
control. Social Problems 60(3): 383–407.
Fischer P, Greitemeyer T, Kastenmüller A et al. (2011) The bystander-effect: A meta-analytic
review on bystander intervention in dangerous and non-dangerous emergencies. Psychological
Bulletin 137(4): 517–537.
Gambaro L and Joshi H (2017) Moving home in the early years: What happens to children in the
UK? Longitudinal and Life Course Studies 7(3): 265–287.
Gau JM (2014) Unpacking collective efficacy: The relationship between social cohesion and infor-
mal social control. Criminal Justice Studies 27(2): 210–225.
Gerell M and Kronkvist K (2017) Violent crime, collective efficacy and city-centre effects in
Malmö. British Journal of Criminology 57(5): 1185–1207.
Gill C, Weisburd D, Telep CW et al. (2014) Community-oriented policing to reduce crime, dis-
order and fear and increase satisfaction and legitimacy among citizens: A systematic review.
Journal of Experimental Criminology 10(4): 399–428.
GLA (2020) Housing in London 2020: The Evidence Base for the London Housing Strategy.
Greater London Authority.
Hawdon JE, Ryan J and Griffin SP (2003) Policing tactics and perceptions of police legitimacy.
Police Quarterly 6(4): 469–491.
Hipp JR and Wickes R (2017) Violence in urban neighborhoods: A longitudinal study of collective
efficacy and violent crime. Journal of Quantitative Criminology 33(4): 783–808.
Hohl K, Bradford B and Stanko EA (2010) Influencing trust and confidence in the London
Metropolitan Police: Results from an experiment testing the effect of leaflet drops on public
opinion. British Journal of Criminology 50(3): 491–513.
Horne C (2004) Collective benefits, exchange interests, and norm enforcement. Social Forces
82(3): 1037–1062.
Innes H and Innes M (2011) Police Presence and Public Confidence in Local Policing: An Analysis
of the British Crime Survey. Cardiff.
Jackson J and Sunshine J (2007) Public confidence in policing: A neo-Durkheimian perspective.
British Journal of Criminology 47(2): 214–233.
Jackson J, Bradford B, Stanko B et al. (2013) Just Authority? Trust in the Police in England and
Wales. New York: Routledge.
Jak S (2019) Cross-level invariance in multilevel factor models. Structural Equation Modeling: A
Multidisciplinary Journal 26(4): 607–622.
Yesberg et al. 19

Kochel TR (2012) Can police legitimacy promote collective efficacy? Justice Quarterly 29(3):
384–419.
Kochel TR (2018) Applying police legitimacy, cooperation, and collective security hypotheses
to explain collective efficacy and violence across neighbourhoods. International Journal of
Comparative and Applied Criminal Justice 42(4): 253–272.
Kochel TR and Gau JM (2019) Examining police presence, tactics, and engagement as facilitators
of informal social control in high-crime areas. Justice Quarterly.
Kochel TR and Weisburd D (2019) The impact of hot spots policing on collective efficacy:
Findings from a randomized field trial. Justice Quarterly 36(5): 900–928.
Kubrin CE and Weitzer R (2003) New directions in social disorganization theory. Journal of
Research in Crime and Delinquency 40(4): 374–402.
LaFree G (1998) Losing Legitimacy: Street Crime and the Decline of Social Institutions in
America. New York: Routledge.
Langella M and Manning A (2019) Residential Mobility and Unemployment in the UK. CEP
Discussion Paper, Centre for Economic Performance.
Lüdtke O, Marsh HW, Robitzsch A et al. (2008) The multilevel latent covariate model: A new,
more reliable approach to group-level effects in contextual studies. Psychological Methods
13(3): 203–229.
Mazerolle L, Wickes R and McBroom J (2010) Community variations in violence: The role of
social ties and collective efficacy in comparative context. Journal of Research in Crime and
Delinquency 47(1): 3–30.
MPS (2018) Keeping London Safe: The Met’s Direction: Our Strategy 2018–2025. Executive
Summary. Metropolitan Police.
Murphy K, Sargeant E and Cherney A (2015) The importance of procedural justice and police
performance in shaping intentions to cooperate with the police: Does social identity matter?
European Journal of Criminology 12(6): 719–738.
Muthén LK and Muthén BO (2012) Mplus User’s Guide, 7th Edn. Los Angeles: Muthén & Muthén.
Nix J, Wolfe SE, Rojek J et al. (2015) Trust in the police: The influence of procedural justice and
perceived collective efficacy. Crime and Delinquency 61(4): 610–640.
Noble N, Wright G, Dibben C, Smith GAN, McLennan D, Anttila C, Barnes H, Mokhtar C, Noble
S, Avenell D, Gardner J, Covizzi I and Lloyd M (2004) The English Indices of Deprivation,
2004. Office of the Deputy Prime Minister, Neighbourhood Renewal Unit, London.
Renauer BC (2007) Is neighborhood policing related to informal social control? Policing 30(1):
61–81.
Rhineberger-Dunn GM and Carlson SM (2011) An analysis of the mediating effects of social
relations and controls on neighborhood crime victimization. Western Criminology Review
12(1): 15–34.
Rosenbaum DP (1987) The theory and research behind neighbourhood watch: Is it a sound fear
and crime reduction strategy? Crime and Delinquency 33(1): 103–134.
Sampson RJ (2010) Collective efficacy theory. In: Cullen FT and Wilcox P (eds) Encyclopedia of
Criminological Theory, Vol. 2. Thousand Oaks, CA: Sage.
Sampson RJ (2012) Great American City: Chicago and the Enduring Neighborhood Effect.
Chicago, IL: University of Chicago Press.
Sampson RJ and Wikström PO (2008) The social order of violence in Chicago and Stockholm
neighborhoods: A comparative inquiry. In: Kalyvas S, Shapiro I and Masoud T (eds) Order,
Conflict, and Violence. Cambridge: Cambridge University Press, 97–119.
Sampson RJ, Raudenbush SW and Earls F (1997) Neighbourhoods and violent crime: A multilevel
study of collective efficacy. Science 277(5328): 918–924.
20 European Journal of Criminology 00(0)

Sargeant E (2017) Policing and collective efficacy: The relative importance of police effectiveness,
procedural justice and the obligation to obey police. Policing and Society 27(8): 927–940.
Sargeant E, Wickes R and Mazerolle L (2013) Policing community problems: Exploring the role
of formal social control in shaping collective efficacy. Australian and New Zealand Journal
of Criminology 46(1): 70–87.
Schaefer L, Mazerolle L and Kapnoulla M (2017) Different actions for different crimes: Explaining
individual action in local crime problems. Journal of Community Psychology 45(7): 922–939.
Scott JD (2002) Assessing the relationship between police–community coproduction and neigh-
borhood-level social capital. Journal of Contemporary Criminal Justice 18(2): 147–166.
Shaw C and McKay H (1942) Juvenile Delinquency and Urban Areas. Chicago: University of
Chicago Press.
Silver E and Miller L (2004) Sources of informal social control in Chicago neighborhoods.
Criminology 42(3): 551–584.
Sindall K and Sturgis P (2013) Austerity policing: Is visibility more important than absolute num-
bers in determining public confidence in the police? European Journal of Criminology 10(2):
137–153.
Skogan WG (1990) Disorder and Decline: Crime and the Spiral of Decay in American Cities. New
York: The Free Press.
Skogan WG (2019) Advocate: Community policing. In: Weisburd D and Braga AA (eds) Police
Innovation: Contrasting Perspectives. Cambridge: Cambridge University Press.
Sunshine J and Tyler TR (2003) The role of procedural justice and legitimacy in shaping public
support for policing. Law and Society Review 37(3): 513–548.
Transparency International UK (2017) Faulty Towers: Understanding the Impact of Overseas
Corruption on the London Property Market. URL (accessed 16 July 2021): https://www.trans-
parency.org.uk/sites/default/files/pdf/publications/TIUK_Faulty_Towers_August_24.pdf.
Triplett RA, Gainey RR and Sun IY (2003) Institutional strength, social control and neighborhood
crime rates. Theoretical Criminology 7(4): 439–467.
Tyler TR and Fagan J (2008) Legitimacy and cooperation: Why do people help the police fight
crime in their communities? Ohio State Journal of Criminal Law 6(1): 231–276.
Tyler TR and Huo YJ (2002) Trust in the Law: Encouraging Public Cooperation with the Police
and Courts. New York: Russell Sage Foundation.
Van Damme A, Pauwels L and Svensson R (2013) Why do Swedes cooperate with the police? A
SEM analysis of Tyler’s Procedural Justice Model. European Journal on Criminal Policy and
Research 21(1): 15–33.
Walters GD and Bolger PC (2019) Procedural justice perceptions, legitimacy beliefs, and compli-
ance with the law: A meta-analysis. Journal of Experimental Criminology 15(3): 341–372.
Warner BD (2007) Directly intervene or call the authorities? A study of forms of neighborhood
social control within a social disorganization framework. Criminology 45(1): 99–129.
Weisburd D, White C and Wooditch A (2020) Does collective efficacy matter at the micro geo-
graphic level? Findings from a study of street segments. British Journal of Criminology
60(4): 873–891.
Wickes R, Hipp JR, Sargeant E et al. (2013) Collective efficacy as a task specific process:
Examining the relationship between social ties, neighborhood cohesion and the capacity
to respond to violence, delinquency and civic problems. American Journal of Community
Psychology 52(1–2): 115–127.
Wikström P, Oberwittler D, Treiber K et al. (2012) Breaking Rules: The Social and Situational
Dynamics of Young People’s Urban Crime. Oxford: Oxford University Press.
Williams KR and Guerra NG (2011) Perceptions of collective efficacy and bullying perpetration
in schools. Social Problems 58: 1533–8533.
Yesberg et al. 21

Wright EM and Benson ML (2011) Clarifying the effects of neighborhood context on violence
‘behind closed doors’. Justice Quarterly 28(5): 775–798.
Yesberg JA and Bradford B (2021). Policing and collective efficacy: A rapid evidence assessment.
International Journal of Police Science and Management. Epub ahead of print 19 July 2021.
DOI: 10.1177/14613557211026938.

Appendix

Table 4.  Item wordings endogenous constructs.

Items
Collective People in this neighbourhood can be trusted [SC1]
efficacy People act with courtesy to each other in public space in this area [SC2]
You can see from the public space here in the area that people take pride in
their environment [SC3]
This local area is a place where people from different backgrounds get on
well together [SC4]
Local people and authorities have control over the public space in this area
[IC1]
If I sensed trouble whilst in this area, I could get help from people who live
here [IC2]
The people who live here can be relied upon to call the police if someone is
acting suspiciously [IC3]
If any of the children or young people around here are causing trouble, local
people will tell them off [IC4]
Trust in police To what extent do you agree with these statements about the police in your area?
fairness They can be relied on to be there when you need them [PF1]
They would treat you with respect if you had contact with them for any
reason [PF2]
The police in this area treat everyone fairly regardless of who they are [PF3]
They are dealing with the things that matter to people in this community
[PF4]
The police in this area listen to the concerns of local people [PF5]
The police in this area are helpful [PF6]
The police in this area are friendly and approachable [PF7]
Trust in police How well do you think the Metropolitan Police. . .
effectiveness Prevent terrorism [EFF1]
Respond to emergencies promptly [EFF2]
Provide a visible patrolling presence [EFF3]
Tackle gun crime [EFF4]
Support victims and witnesses [EFF5]
Tackle drug dealing and drug use [EFF6]
Tackle dangerous driving [EFF7]
22 European Journal of Criminology 00(0)

Table 5.  Measurement model for police fairness and effectiveness (wave 2).

Constrained

B SE Sig.
Within-level
PF1 1.000 0.000  
PF2 0.759 0.016 .000
PF3 0.942 0.018 .000
PF4 0.988 0.014 .000
PF5 0.938 0.015 .000
PF6 0.924 0.014 .000
PF7 0.791 0.016 .000
EFF1 1.000 0.000  
EFF2 1.189 0.022 .000
EFF3 1.508 0.036 .000
EFF4 1.399 0.027 .000
EFF5 1.470 0.031 .000
EFF6 1.396 0.032 .000
EFF7 1.293 0.032 .000
Between-level
PF1 1.000 0.000  
PF2 0.759 0.016 .000
PF3 0.942 0.018 .000
PF4 0.988 0.014 .000
PF5 0.938 0.015 .000
PF6 0.924 0.014 .000
PF7 0.791 0.016 .000
EFF1 1.000 0.000  
EFF2 1.189 0.022 .000
EFF3 1.508 0.036 .000
EFF4 1.399 0.027 .000
EFF5 1.470 0.031 .000
EFF6 1.396 0.032 .000
EFF7 1.293 0.032 .000
Level-2 variance
Fairness 0.024 0.003 .000
Effectiveness 0.036 0.004 .000
Level-1 variance
Fairness 0.435 0.014 .000
Effectiveness 0.474 0.022 .000
ICC
Fairness .05  
Effectiveness .07  
RMSEA 0.035  
TLI 0.958  
CFI 0.960  
Yesberg et al. 23

Table 6.  Measurement model for collective efficacy (wave 3).

Constrained

B SE Sig.
Within-level
SC1 1.000 0.000  
SC2 0.986 0.018 .000
SC3 1.069 0.024 .000
SC4 0.561 0.015 .000
IC1 0.898 0.024 .000
IC2 0.929 0.020 .000
IC3 0.768 0.020 .000
IC4 0.969 0.029 .000
Between-level
SC1 1.000 0.000  
SC2 0.986 0.018 .000
SC3 1.069 0.024 .000
SC4 0.561 0.015 .000
IC1 0.898 0.024 .000
IC2 0.929 0.020 .000
IC3 0.768 0.020 .000
IC4 0.969 0.029 .000
Level-2 variance 0.043 0.004 .000
Level-1 variance 0.278 0.011 .000
ICC 0.13  
RMSEA 0.034  
TLI 0.961  
CFI 0.952  

Table 7.  Measurement model for lagged controls (wave 1).

Constrained

B SE Sig.
Within-level
PJ1 1.000 0.000  
PJ2 0.750 0.015 .000
PJ3 0.896 0.017 .000
PJ4 0.978 0.015 .000
PJ5 0.947 0.015 .000
PJ6 0.934 0.015 .000
PJ7 0.795 0.015 .000
EFF1 1.000 0.000  
EFF2 1.202 0.023 .000
(Continued)
24 European Journal of Criminology 00(0)

Table 7. (Continued)

Constrained

B SE Sig.
EFF3 1.532 0.036 .000
EFF4 1.466 0.030 .000
EFF5 1.557 0.030 .000
EFF6 1.444 0.033 .000
EFF7 1.308 0.034 .000
SC1 1.000 0.000  
SC2 0.958 0.016 .000
SC3 1.063 0.021 .000
SC4 0.580 0.016 .000
IC1 0.767 0.020 .000
IC2 0.949 0.019 .000
IC3 0.828 0.022 .000
IC4 0.927 0.024 .000
Between-level
PJ1 1.000 0.000  
PJ2 0.750 0.015 .000
PJ3 0.896 0.017 .000
PJ4 0.978 0.015 .000
PJ5 0.947 0.015 .000
PJ6 0.934 0.015 .000
PJ7 0.795 0.015 .000
EFF1 1.000 0.000  
EFF2 1.202 0.023 .000
EFF3 1.532 0.036 .000
EFF4 1.466 0.030 .000
EFF5 1.557 0.030 .000
EFF6 1.444 0.033 .000
EFF7 1.308 0.034 .000
SC1 1.000 0.000  
SC2 0.958 0.016 .000
SC3 1.063 0.021 .000
SC4 0.580 0.016 .000
IC1 0.767 0.020 .000
IC2 0.949 0.019 .000
IC3 0.828 0.022 .000
IC4 0.927 0.021 .000
Level-2 variance
Fairness 0.020 0.002 .000
Effectiveness 0.027 0.004 .000
Collective efficacy 0.046 0.004 .000
(Continued)
Yesberg et al. 25

Table 7. (Continued)

Constrained

B SE Sig.
Level-1 variance
Fairness 0.396 0.013 .000
Effectiveness 0.473 0.020 .000
Collective efficacy 0.253 0.010 .000
ICC
Fairness 0.05  
Effectiveness 0.05  
Collective efficacy 0.15  
RMSEA 0.027  
TLI 0.954  
CFI 0.951  

Table 8.  Additional results.

Model 2A: Temporal controls Model 2B: Naive mean for

(unconstrained) exogenous

B SE. Sig. Beta B SE Sig. Beta

Collective efficacy
Fairness 0.582 0.178 .001 0.451 0.589 0.179 .001 0.451
Effectiveness −0.175 0.142 .217 −0.168 −0.175 0.147 .234 −0.161
Visibility −0.009 0.026 .739 −0.022 −0.010 0.017 .545 −0.033
Community engagement 0.053 0.128 .677 0.033 0.038 0.055 .496 0.035
Crime rate (log) −0.003 0.028 .906 −0.008 −0.006 0.029 .835 −0.014
Deprivation 0.007 0.001 <.001 0.363 0.007 0.001 <.001 0.360
Collective efficacy (wave 0.277 0.062 <.001 0.316 0.287 0.064 <.001 0.315
1)
Fairness
Visibility 0.066 0.027 .014 0.218 0.044 0.017 .008 0.181
Community engagement 0.178 0.124 .151 0.141 0.070 0.052 .178 0.085
Crime rate (log) −0.023 0.024 .331 −0.069 −0.024 0.023 .311 −0.071
Deprivation 0.000 0.001 .845 0.013 0.00 0.001 .807 0.017
Fairness (wave 1) 0.136 0.091 .137 0.125 0.145 0.091 .113 0.132
Effectiveness
Visibility 0.075 0.030 .013 0.200 0.043 0.018 .019 0.146
Community engagement 0.266 0.141 .060 0.170 0.127 0.059 .030 0.128
Crime rate (log) −0.055 0.023 .017 −0.134 −0.058 0.023 .010 −0.144
Deprivation 0.002 0.001 .037 0.126 0.002 0.001 .035 0.127
Effectiveness (wave 1) −0.013 0.082 .876 −0.012 −0.005 0.078 .944 −0.005
(Continued)
26 European Journal of Criminology 00(0)

Table 8. (Continued)

Model 2A: Temporal controls Model 2B: Naive mean for

(unconstrained) exogenous

B SE. Sig. Beta B SE Sig. Beta

Level 2 variance
Fairness 0.021 0.003 0 0.019 0.002 0  
Effectiveness 0.032 0.005 0 0.029 0.004 0  
Collective efficacy 0.025 0.003 0 0.048 0.004 0  
Visibility 0.249 0.020 0  
Community engagement 0.014 0.002 0  
Level 1 variance
Fairness 0.435 0.014 0 0.440 0.012 0  
Effectiveness 0.474 0.022 0 0.459 0.017 0  
Collective efficacy 0.279 0.011 0 0.290 0.010 0  
Visibility 2.507 0.029 0  
Community engagement 0.361 0.005 0  
Model fit
RMSEA 0.010 0.010  
TLI 0.937 0.954  
CFI 0.938 0.954