Triggers of change to public confidence in the police and criminal justice system: Findings from the Crime Survey for England and Wales panel experiment

Abstract

Accounts of public ‘trust and confidence’ in criminal justice agencies often fall into one of two camps. Instrumental accounts suggest that people trust police and the criminal justice system (CJS) when they believe them to be effective in fighting crime and reducing offending. Expressive or affective accounts, by contrast, suggest people place as much or more emphasis on the social meaning of justice institutions as on their instrumental activities. In this article we add to recent studies that have sought to weigh up the balance between instrumental and expressive factors. Using data from the Crime Survey for England and Wales panel experiment, we present evidence that trust in police and the wider CJS is implicated in public concerns about the nature of local order and cohesion. The expressive account appears to offer a better understanding of why people may grant trust to, or withdraw trust from, the police and the CJS.

Keywords

criminal justice system police public opinion trust and confidence

Trust and confidence in the police and criminal justice system (CJS) is a topic that continues to receive significant attention in the criminological and policy literatures. Linked both to legitimacy and, consequently, public co-operation and compliance (Jackson et al., 2012a; Tyler and Huo, 2002), public trust in the police has been an object of particular attention. Events such as the 2011 riots in England and the Leveson inquiry into the culture of the press, which implicated a significant number of officers in corruption, serve to almost continuously ramp up levels of debate and contest around policing, a process that many would argue has been going on since the 1970s at least (Reiner, 2010). While there is less debate around public confidence in the CJS as a whole, the latter continues to be a subject of almost equal policy concern, as witnessed by repeated attempts to devise ways of enhancing people’s confidence in the courts, prisons and other organizations (Chapman et al., 2002; Singer and Cooper, 2008).

There are, broadly speaking, two accounts of the sources of public trust in criminal justice agencies (Freiburg, 2001). The first can be characterized as instrumental or cognitive: people believe the police, courts and prisons are there to ‘fight’ crime, and place trust in these institutions when and to the extent that they judge them to be effective in this task. The second account can be characterized as expressive or affective. Here the argument is that people are (a) less concerned with instrumental factors than we might expect, and (b) much more concerned with both how they are treated by the police and other agencies, and by a wider range of factors that indicate whether police, in particular, are successful in maintaining, and indeed in representing, order, stability and cohesion (Jackson and Bradford, 2009; Jackson and Sunshine, 2007; Jackson et al., 2009).

Understanding which of these accounts provides greater insight into what generates and modulates public trust in the police and the CJS is important not only because of what criminal justice agencies stand to gain from higher levels of public trust – legitimacy, co-operation and compliance – but also because citizens have a right to be served by trusted justice institutions and to expect that those institutions will both represent their core interests and beliefs and concentrate on issues they believe to be important. The policy implications are obvious. If trust and confidence are motivated primarily by instrumental concerns, police and other agencies should concentrate on being seen to be firm and effective in dealing with crime. If, on the other hand, expressive factors are more important, too great an emphasis on ‘crime-fighting’ may not only be ineffective in terms of enhancing trust but actually counter-productive if, for example, it generates many confrontational encounters between police and citizens.¹ Moreover, if people are more concerned with issues of low-level disorder and community cohesion, policies that aim to increase public confidence by emphasizing the police’s ‘crime-fighting’ role may turn out to be misconstrued.

From one perspective the evidence on this question appears clear. When it comes to personal contact with officers, people strongly value expressive ‘means’, such as quality of treatment, fairness, dignity and respect, over instrumental ‘ends’ in terms of ‘getting results’. Panel and experimental data have been used in a number of recent studies to demonstrate that contact experiences do appear to change people’s opinions of the police (Myhill and Bradford, 2012), that fairness judgements play a key role in this process (Tyler and Fagan, 2008), and that the experience of procedural fairness – or unfairness – during face-to-face interactions with officers is linked to legitimacy judgements (Mazerolle et al., 2012).

Yet only a minority of people come into contact with the police and other criminal justice agencies from year to year – what influences trust and confidence on a more general level? Here, the evidence is mixed. Studies conducted in England and Wales have generally found that it is expressive concerns about low-level disorder, social cohesion, collective efficacy and change in society that are associated most strongly with trust and confidence (Jackson and Bradford, 2009; Jackson and Sunshine, 2007; Jackson et al., 2012b; Merry et al., 2012), with concerns about crime and victimization less important. Such studies have tended to be cross-sectional in design, however, making it difficult to identify factors that might effect change in levels of trust. In an important recent study, Sindall et al. (2012) used time-series analysis to show that, at the aggregate level, only instrumental factors have any long-run relationship with levels of confidence in the police. It may be that while expressive concerns about (dis)order and cohesion are associated with trust and confidence on a cross-sectional basis, they are relatively stable over time and are therefore unlikely to be the cause of change in levels of trust.

In this article we build on recent studies to examine whether change in (a) instrumental factors (the perceived risk of victimization and actual victimization), and (b) expressive concerns about disorder and collective efficacy, are associated with changes in trust in the police and the CJS at the individual level. Using data from the Crime Survey for England and Wales (CSEW)² panel experiment (Fong and Williams, 2011) we find that changes in perceptions of disorder and collective efficacy were consistently associated with changes in confidence in the local police and in the CJS as a whole. Change in the perceived risk of victimization had, by contrast, no consistent association with change in confidence; recent victimization had no such association at all, although earlier victimization did have some link with later confidence. We therefore present evidence in support of both instrumental and affective accounts of trust and confidence, although the affective account retains a significant explanatory advantage. The article concludes with some reflections on the implications of our findings.

Trust in the Police and the CJS

There is a voluminous literature on attitudes to the police, concentrated in the USA but drawing increasingly on research from across the world (e.g. Brandl et al., 1994; Brown and Benedict, 2002; Hinds and Murphy, 2007; Skogan, 2012; Sun et al., 2013). While until recently only a relatively small number of studies have considered public trust in the police in a UK context (Jackson and Sunshine, 2007), here, too, the literature is growing (e.g. Bradford et al., 2009; Kautt, 2011; Merry et al., 2012; Sindall and Sturgis, 2013; Sindall et al., 2012). Measuring public attitudes towards the ‘criminal justice system’ as an entity, by contrast, seems a largely UK phenomenon (e.g. Bradford, 2011; Van Der Walle, 2009).

Brown and Benedict (2002) reviewed over 100 studies, highlighting several methodological and conceptual problems associated with measuring attitudes towards the police that should be considered when attempting to synthesize research findings. Of particular note is the nature of the dependent variable. Studies have for example distinguished between ‘diffuse’ and ‘specific’ support for the police, with the former relating to an overall evaluation of an organization’s performance and the latter to specific policies or individuals (Brandl et al., 1994; Brown and Benedict, 2002: 564). In England and Wales, a distinction is often made between satisfaction with specific encounters with officers, and more general ‘trust and confidence’ in the police as an organization. This latter phrase, subject to significant and often anguished conceptual debate about the distinction between trust and confidence (see Luhmann, 1979), is perhaps best seen as wrapping up people’s perceptions of the current effectiveness and good intentions of the police (Hardin, 2006: 17) with their assessments of the extent to which police will be effective and officers well intentioned were they to encounter them in the future (Bottoms and Tankebe, 2011). Satisfaction, by contrast, is a retrospective assessment of specific police actions and performance. Yet, satisfaction and trust/confidence are clearly related: satisfaction with prior contact will shape trust judgements, and the extent to which people trust the police will colour how they ‘read’ officer behaviour during encounters (Myhill and Bradford, 2012; Rosenbaum et al., 2005).

A second important methodological issue relating to attitudes research is the question of whether to assess public opinions using several related attitude statements or a single ‘catch-all’ question. The former approach is preferable both methodologically, since it allows more accurate measurement of the underlying construct, and theoretically, as there are different varieties or components of trust (Jackson et al., 2012b; Stoutland, 2001). Indeed, people can have high trust in some elements of policing and low trust in others (Stone et al., 2005), and satisfaction with encounters with the police can be associated with some elements of wider trust and not others (Bradford et al., 2009). Single indicator measures of confidence and trust are, however, appealing to policy-makers for their simplicity (and to survey managers seeking to reduce the content of large surveys), and have, consequently, been used in several UK studies of attitudes to the police (e.g. Kautt, 2011; Merry et al., 2012). Although they will always be somewhat crude and reductive, such indicators have been shown to ‘summarize’ wider trust in the police relatively well (Jackson and Bradford, 2010).

Notwithstanding these issues, three individual-level variables – age, ethnicity and contact with the police – seem to be consistently associated with attitudes towards the police (Brown and Benedict, 2002). Individual-level variables may, however, have complex interactions with contextual variables. The effect of ethnicity, in particular, on attitudes may be influenced by factors operating at the neighbourhood level – the style of policing adopted in particular communities, for example. The influence of victimization and the fear of victimization on attitudes and/or trust seems, significantly, to vary between studies.

Brown and Benedict (2002) further identified a ‘contextual-level’ variable that was associated consistently with public attitudes: type of neighbourhood. They suggested however that there is ‘no consensus about why attitudes toward the police vary by neighbourhood’ (Brown and Benedict, 2002: 556). One reason for this lack of consensus may be found in the suggestion that some US studies, while statistically advanced, have been largely atheoretical (Kautt, 2011), with little attempt to explain the correlates that emerge from complex multivariate analyses. Yet, there are significant examples of theory-driven research in the US context. Sampson et al. (1997) published a landmark study linking collective efficacy – social cohesion, trust and informal social control – to levels of violent crime in Chicago neighbourhoods. Concentrated socio-economic disadvantage appears to be highly predictive of low collective efficacy (Sampson, 2004), and of lower satisfaction with the police (Sampson and Bartusch, 1998). Sampson (2004) and others (e.g. Gau and Brunson, 2010) suggest that negative attitudes to the police in deprived neighbourhoods may be caused in part by ‘enforcement’ styles of policing which are frequently applied in such neighbourhoods. In essence, poor areas receive heavy handed policing that alienates local residents and undermines their trust in the police.

Another approach has been to think not so much about whether disadvantage and disorder promote policing styles that undermine trust, but about what low-level disorder and its correlates (social cohesion and collective efficacy) mean for local residents (Jackson and Bradford, 2009; Jackson and Sunshine, 2007; Jackson et al., 2012b). Trust and distrust are active stances (Luhmann, 1979), and the relational and expressive sense of cohesion and efficacy that exists (or not) among residents of a particular locality may be one factor influencing the stances they take in relation to criminal justice agents. Key here is the idea that police, in particular, are embedded in a particular kind of ‘value-bearing narrative’ (Earle and Cvetkovich, 1995); representing, for many, a sense of order and cohesion, and serving as an explanatory mechanism for understanding how order and cohesion are maintained over time (Girling et al., 2000). Experiences that support this narrative (by promoting a sense of order or by suggesting that local people are ‘pulling together’) increase trust. Experiences or impressions of community breakdown, by contrast, diminish trust because they undermine the narrative of policing – they suggest that there is a failure to maintain order and cohesion, and the police are implicated in this perceived failure (Jackson et al., 2012b).

The expressive nature of the concepts of social cohesion, collective efficacy and disorder, as perceptions people have about their social environment, sits comfortably with the notion that policing in British contexts is inscribed with symbolic meaning relating to cultural and social order (see, for example, Loader and Mulcahy, 2003). Indeed, several previous studies of public attitudes in England and Wales have shown perceptions of collective efficacy and neighbourhood disorder to be associated more strongly with public trust in the police than instrumental factors such as experiencing victimization or generalized worry about crime (Jackson and Bradford, 2009; Jackson and Sunshine, 2007; Jackson et al., 2012b). A possible explanation is that while serious victimization is a rare event for most people, the (perceived) nature and extent of local social bonds are both apprehensible and immediate. Perceptions of disorder, for example, may serve as a guide or heuristic for people when they are forming their trust judgements in relation to the police because disorder indicates threats (Sampson, 2009) the police are failing to address. The effect on trust of concerns about order and cohesion may go yet deeper, however, outweighing the effects of victimization (threat made real) even when this does occur. When people perceive their local environment as orderly, for example, this may indicate not only the absence of crime but also that the local community is efficacious in upholding shared norms and values. On this account police, representative of community and order, garner trust when community appears unified and social order seems strong (Jackson et al., 2012b).

Limitations of recent UK research

Despite the regular suggestion of strong associations between concerns about disorder and neighbourhood, on the one hand, and (for example) trust in the police on the other, much existing UK research has used cross-sectional survey data to investigate these issues and so might be considered somewhat limited, especially when it comes to investigating processes of change. In particular, it seems possible that perceptions of disorder and collective efficacy are relatively stable over time, because they are associated with ideological postures or because the nature of local areas are themselves relatively stable (Sampson, 2012). Such perceptions may provide an underlying structure to the narrative people have concerning the nature of social order and the trustworthiness of the police, but be unlikely to trigger change to that narrative.

Furthermore, two recent UK studies have produced findings inconsistent to some degree with other extant research. Kautt (2011) combined several sweeps of CSEW data to examine the correlates of trust and confidence on a cross-sectional basis. Crime victims, it was suggested, were less likely to rate the police as doing a good job as they felt ‘the police failed them’ (Kautt, 2011: 369), a primarily instrumental judgement. Kautt did not, however, have measures of victims’ assessments of the quality of their contacts with the police, or of perceptions of social cohesion and collective efficacy, expressive concerns that might contrast with instrumental concerns about victimization.

Sindall et al. (2012) also pooled data from several sweeps of the CSEW, but took a different analytic approach, using time-series analysis to identify factors associated with public trust in the police at the aggregate level and over time. They found that ‘only aggregate perceptions of crime and the property crime rate have any long-run predictive relationship with public confidence in the police’ (Sindall et al., 2012: 756). It may be that positive change in trust and confidence will be generated by more instrumental judgements about, for example, crime rates.

This latter study, in particular, represents an attempt to derive as much value as possible from cross-sectional data. As Skogan (2012) contends, however, what is really required to move towards causal explanations of attitudes to the police is panel (and experimental) data. Myhill and Bradford (2012) used panel data to show that contact with the police – judged primarily in expressive terms – was associated independently with wider trust in the police. Measures of change in perceptions of disorder and collective efficacy were not utilized in this study, however, and the sample was drawn in part from neighbourhoods subject to a pilot programme of intensive community engagement and targeted problem-solving activity, placing the generalizability of its findings in some doubt.

Another limitation of existing research is that, despite policy interest in public opinion of the wider CJS, it has focused primarily on the police. Part of the reason for the relative absence of research on confidence in the CJS may be scepticism about the extent to which, as an aggregate body, it can be an ‘object’ of trust in any meaningful sense. Survey questions about confidence in the effectiveness and fairness of a range of CJS organizations have, though, been shown to load onto single underlying factors, suggesting the CJS may indeed be a ‘thing’ that people may, or may not, trust (Bradford, 2011; Hough et al., 2013). Moreover, Bradford (2011) demonstrated that contact with Victim Support, a very specific part of the CJS, was associated with variation in levels of confidence in the system as a whole. It remains an open question, therefore, as to whether people have a similar orientation – instrumental or affective – towards the CJS as they do the police, or whether the sources of confidence in these two institutions differ. Previous work has found that service provision premised on principles of procedural justice (Bradford, 2011), and perceptions of disorder (Hough et al., 2013) are associated with such confidence in the CJS. Expressive factors may therefore be important, but this idea has yet to be tested in any systematic manner.

Research hypotheses

Two key issues have, then, hampered extant UK research on attitudes to the police and the CJS. The first is an overreliance on cross-sectional data. The second is the availability of robust measures of key constructs. The present study develops the field in two respects. First, we utilize a dataset hitherto unique in this UK research context: a nationally representative panel sample. Panel data allow us to examine the association between changes in individuals’ victimization experiences, perceptions of crime and their local area and change in their attitudes to the police. Second, we include measures of instrumental and expressive factors, in order to examine which have a greater bearing on ‘trust and confidence’. Our specific hypotheses are that:

H1: Perceiving an increase in disorder (which we treat here as an expressive factor) over time will be associated with a decline in confidence in the police and the CJS.

H2: Perceiving an increase in local collective efficacy (an expressive factor) over time will be associated with an increase in confidence in the police and the CJS.

H3: An increase in the perceived likelihood of victimization (an instrumental factor) over time will be associated with a decline in confidence in the police and the CJS.

H4: Controlling for earlier victimization experiences (an instrumental factor), a report of recent victimization will be associated with a decline in confidence in the police and the CJS.

Data and Methods

A panel experiment was conducted in the spring and summer of 2011 to investigate the possibility of boosting the sample size of the CSEW at the Police Force Area level by re-interviewing respondents from the previous year and adding them to the current year’s sample (Fong and Williams, 2011). Four re-contact methods were tested on some of the 85 per cent of respondents to the 2009/2010 and 2010/2011 surveys who had agreed to be re-contacted when originally interviewed: telephone (using CATI – Computer Aided Telephone Interviewing), postal and two further methods involving on-line surveys. To minimize the risk of ‘mode-effects’ (forms of bias introduced by the different methods of survey administration), in this article we use only the CATI and postal samples. Wave 1 of the current study therefore comprises data collected from January to March 2010 as part of the standard CSEW, using face-to-face interviews. Wave 2 comprises data collected approximately one year later, January to March 2011, by either CATI or postal survey (the intention was that all follow-ups would be 12 months after the initial contact, but this was not uniformly achieved). The response rate to the first wave of the panel was by definition the response rate to the CSEW (around 75 per cent). The response rate for both telephone and postal follow-ups was the same (60 per cent).

In the original dataset, there were some 1503 CATI- and 2002 postal-mode respondents. All respondents received the same questionnaire at Wave 2. Given the modular design of the standard CSEW, however, many questions of interest were asked only of one-quarter of respondents at Wave 1. Since we required our measures to be fielded to all respondents in both waves, the final sample size was therefore 879 (362 CATI and 517 postal). The structure of the sample is shown in Table 1.

Table 1.

Structure of the sample (N = 879).

		Percentages
Gender		Car Access (Wave 2)
Male	45	Owns or has regular use	84
Female	55	No	16
Age		Employment status (Wave 2)
16–24	5	Worked in last 7 days	51
25–44	25	No	49
45–64	41
65–74	19	Area type
75+	10	Rural	25
Ethnicity		(of which) inner city	10
White	95
Non-White	5	Recent victim of crime (Wave 2)
		No	84
		Yes	16
Housing tenure
Owner-occupier	79	Recent contact with police (Wave 2)
Social renter	13	No	77
Private renter	8	Yes	23

Response variables

Two response variables were needed. The first, confidence in the local police, is represented by answers to the question ‘Taking everything into account, how good a job do you think the police in this area are doing?’, with answers on a five-point scale ranging from ‘very poor’ to ‘excellent’. The second response variable, confidence in the CJS, is represented by a latent variable measured by two observed indicators: ‘How confident are you that the criminal justice system as a whole is effective?’; and ‘How confident are you that the criminal justice system as a whole is fair?’. Latent variable modelling in the package Mplus was used to estimate and derive this second measure (see later and Appendix Table 1). We take a pragmatic approach and treat both response variables as measures of ‘trust and confidence’, and use these terms interchangeably throughout the article. While we agree there may be some level of conceptual difference between trust and confidence, the present article is perhaps not the place to explore it.

Explanatory variables

Four explanatory variables were needed. One, victimization, was represented by a dummy variable (the panel survey did not, unfortunately, ask victims of crime how satisfied they were with any police response). Confirmatory factor analysis (CFA) in Mplus was utilized to estimate and validate the other three explanatory variables; factor scores for both waves were then extracted for further use (see Appendix Table 1 for full question wordings and factor loadings).

The first latent construct represented respondents’ views about low-level disorder and anti-social behaviour (disorder), measured by five items, including ‘How much of a problem in this area is vandalism, graffiti and other deliberate damage?’. The second latent construct represented prospective perceived risk of victimization. Respondents were asked how likely they thought they were to be victims of burglary and physical attack in the next year. Finally, respondents’ perceptions of local collective efficacy were measured by two items that asked how likely they thought it was that local people would intervene if a fight was occurring or local children were spraying graffiti.

Appendix Table 1 shows that a four-factor model that included the three explanatory variables described above, and confidence in the CJS, was an adequate fit to the data when estimated using Wave 1 and Wave 2 data simultaneously (Appendix Table 2 displays a correlation matrix for all the explanatory and response variables – note that within waves no pairwise correlation exceeded .54, suggesting good discriminant validity). Full information maximum likelihood estimation was used, meaning that cases with some missing values on the observed indicators were retained in the analysis. Note further that the factor loadings of individual items were relatively stable between waves, although the loading of the item assessing the likelihood of suffering a violent attack did differ somewhat between the two sampling points. On balance, however, we do not believe this difference represents too great a threat to the validity of the underlying latent variable (perceived risk of victimization).

Finally, we included the measure of confidence in the local police as an explanatory variable in the models predicting confidence the CJS. It does not seem unreasonable to suggest that confidence in the police, as the most visible and accessible CJS organization, is a key factor shaping confidence in the system as a whole (Hough et al., 2013).

Issues arising from the mixed mode administration

A significant threat to the validity of the latent variables derived from the CSEW panel – and indeed all analysis of these data – arises from the fact that the second wave was administered in two different modes (CATI and postal). Research on mixed mode surveys has shown that responses to questions can differ substantially depending on the way the questionnaire is administered. For example, respondents to aural modes (e.g. CATI) are more likely to give extreme positive answers than respondents to visual modes (e.g. postal) – implying that the mean scores on any observed variable may be more positive among aural mode respondents (Dillman et al., 2009).

To address the issue of measurement equivalence we used the multiple group function in Mplus. We first compared the factor structure of the latent variables measured at Wave 2. Models that allowed factor loadings to vary between the CATI and postal respondents did not improve model fit over a baseline model that constrained all Wave 2 factor loadings to be equal. We then estimated a second series of models that allowed factor loadings and the thresholds of the observed indicators to vary across groups; again, none significantly improved model fit over the baseline. These results suggest that the different survey modes did not influence the ‘meaning’ of the latent variables.

It is important to note that, as suggested by Dillman et al.’s (2009) study, there were significant differences between groups in the means of the latent variables measured at Wave 2. For example the mean of the disorder variable in the CATI group was −0.077, while the mean of the postal group was 0.077 (t = −3.72, p = < .005). To correct for this effect we included a dummy variable for survey administration in all analyses, thus partialing out structural differences caused by the use of two survey modes (and see later for more discussion of this issue).

Control variables

A number of control variables were included in all models: gender; age; ethnicity (dichotomized as White/non-White due to the small sample size); employment status (dichotomized as employed/other); household car access; area type (inner city/rural area/other); a measure of interviewer-coded disorder derived from CSEW interviewers’ assessments of the condition of respondents’ homes and their immediate neighbourhood (logged to correct for skewness); and contact with the police in the past year (measured at Wave 2 only).

Results

Table 2 shows results from a series of ordinal logistic regression models predicting confidence in the local police at Wave 2. Ordinal logistic regression is the preferred method for measures of this type as it is recognizant of the structure of the response variable (Britt and Weisburd, 2011).³ Models 1 and 2 in Table 2 used only Wave 1 predictors. In Model 1, victimization (negative), disorder (negative), collective efficacy (positive) and perceived risk of victimization (negative) were all significant predictors of confidence in the local police at Wave 2. However once confidence in the police at Wave 1 was entered in Model 2 (for the sake of simplicity as a continuous variable) the coefficients for all but disorder lost significance, suggesting that the statistical effect of the other Wave 1 variables on Wave 2 confidence were mediated by confidence at Wave 1.

Table 2.

Ordinal logistic regression models predicting confidence in the local police at Wave 2.

	Model 1ß	Model 2ß	Model 3ß
Follow-up mode (ref: telephone)
Postal	0.02	−0.09	0.06
Gender (ref: male)
Female	0.14	0.13	0.09
Age	−0.01**	−0.01+	−0.01+
Ethnic group (ref: all others)
White	−0.31	−0.19	−0.25
Employment status at Wave 2 (ref: all others)
Employed	−0.09	−0.03	−0.04
Car access (ref: no)
Yes	−0.18	−0.10	−0.09
Area type (ref: urban)
Inner city	−0.16	−0.21	−0.10
Rural	−0.48**	−0.32+	−0.40*
Interviewer-coded disorder (logged) (high=more)	−0.03	−0.03	−0.01
Wave 1 explanatory variables
Victim	−0.29+	−0.21	−0.17
Disorder (high=more)	−0.68**	−0.57**	0.31
Collective efficacy (high=more)	0.33**	0.08	−0.29+
Risk of victimization (high=more)	−0.35*	−0.13	−0.18
Confidence in local police (high=better)		1.33**	1.28**
Wave 2 explanatory variables
Victim			−0.19
Contact with police (ref: none)			0.16
Disorder (high=more)			−1.08**
Collective efficacy (high=more)			0.45**
Risk of victimization (high=more)			0.09
N	813	813	813

Notes: ⁺p < .10; *p < .05; **p < .01.

Brant test for Model 3: Chi sq. = 71.33, p = .096, d.f. = 57.

Model 3 adds the Wave 2 explanatory variables. It is this model that allows us to assess the potential effects of change in experiences of crime, disorder and collective efficacy on confidence in the police. Since perceptions of disorder (for example) at Wave 1 are included (held constant) in the model, a positive coefficient for disorder at Wave 2 would represent a predicted increase in confidence between Waves 1 and 2 (specifically, the increase in the log odds of being in a higher category of the response variable associated with a one unit increase in the explanatory variable), while a negative coefficient would represent a predicted decline in confidence. Strikingly, we find that while an increase in perceptions of disorder was associated with a decline in confidence in the police, and that an improved perception of local collective efficacy was associated with an increase in confidence, change in the perceived risk of victimization was not associated with such a change. Furthermore, conditioning on victim status at Wave 1, respondents who fell victim to crime between waves did not, once the other variables were taken into account, report significantly lower levels of confidence in the police at Wave 2.

Note that few other variables included in the models were significant. Confidence fell slightly with age, and people living in rural areas tended to express less trust in the police than those living in towns and cities. Most notably, interviewer-coded disorder, which might be considered as a more objective measure of the orderliness of respondents’ immediate physical environments, was not significant in any model. This finding provides some support for the idea that what is at stake here are respondents’ subjective assessments of their local areas, not some unobserved feature of ‘objectively’ disorderly neighbourhoods (e.g. aggressive policing).

Table 3 repeats the above process, this time with confidence in the CJS as the response variable, using linear regression models. Here, given inclusion in the model of disorder (for example) at Wave 1, the coefficient for disorder at Wave 2 represents the expected change in the mean of the confidence measure from Wave 1 to Wave 2 associated with a unit change in the disorder measure over the same period. There are some marked similarities with the police models. Perceptions of disorder, risk of victimization and confidence in the local police, measured at Wave 1, were all significant predictors of confidence in the CJS at Wave 2 (Model 4); and most, although not all, of the statistical effect of these variables appeared to be mediated by confidence in the CJS at Wave 1 (Model 5). Finally, Model 6 shows that change in perceptions of disorder, in collective efficacy, and in confidence in the local police between Waves 1 and 2 were all factors associated with change in confidence in the CJS over the same period, with statistical effects all in the hypothesized directions.

Table 3.

Linear regression models predicting confidence in the criminal justice system at Wave 2.

	Model 4ß	Model 5ß	Model 6ß
Follow-up mode (ref: telephone)
Postal	−0.21**	−0.21**	−0.17**
Gender (ref: male)
Female	0.01	0.03	0.02
Age	−0.01**	0.00	0.00
Ethnic group (ref: all others)
White	−0.06	0.02	0.01
Employment status at Wave 2 (ref: all others)
Employed	0.11+	0.08*	0.08+
Car access (ref: no)
Yes	−0.10	−0.05	−0.03
Area type (ref: urban)
Inner city	−0.26**	−0.20**	−0.17**
Rural	0.04	0.02	0.02
Interviewer-coded disorder (logged) (high=more)	0.01	0.01	0.03
Wave 1 explanatory variables
Victim	−0.04	−0.02	0.01
Disorder (high=more)	−0.16**	−0.08*	0.23**
Collective efficacy (high=more)	0.03	−0.01	−0.19**
Risk of victimization (high=more)	−0.18**	0.03	−0.12
Trust in local police (high=better)	0.27**	0.06*	−0.01
Confidence in the CJS (high=more)		0.79**	0.75**
Wave 2 explanatory variables
Victim			−0.06
Contact with police (ref: none)			−0.03
Disorder (high=more)			−0.29**
Collective efficacy (high=more)			0.22**
Risk of victimization (high=more)			0.14*
Trust in local police (high=better)			0.10**
Constant	−0.47*	−0.10	−0.24
R²	0.22	0.61	0.66
N	813	813	813

Notes: +p < .10; *p < .05; **p < .01.

Rather perplexingly, conditioning on these other factors an increase in perceived risk of victimization was associated with an increase in confidence in the CJS. Further analysis (results not shown here) demonstrated that this effect arose only once perceptions of disorder were included in the model, and, absent disorder the coefficients for perceptions of risk at both Waves 1 and 2 were not significant predictors of confidence. It may be that, holding perceptions of disorder constant, the ‘idea’ of the CJS becomes more salient as people perceive greater risk, and this greater salience has the effect of increasing trust in the system; or, more accurately perhaps, their sense that it is important, something on which they in a sense rely and therefore should support (see Van Der Toorn et al., 2011).

This finding aside, however, the results in relation to confidence in the CJS echo those in relation to the police. Change in the more expressive variables – disorder, collective efficacy – was associated with change in confidence exactly as hypothesized. Change in the more instrumental factors of risk and victimization did not have the hypothesized effect on people’s views, although perceived risk of victimization at Wave 1 did have an indirect effect, mediated by Wave 1 confidence, on Wave 2 confidence.

Additional analysis

Fong and Williams (2011: 20) reported that in the CSEW panel experiment the telephone model provided responses most congruent with the original face-to-face administration. Given the potential problems created by the mixed mode administration at Wave 2, we re-estimated the models shown above using only the CATI respondents (full results available from the lead author). Results were similar to those shown above. Changes in both disorder and collective efficacy between Waves 1 and 2 were associated with significant change in confidence in the police, in the expected directions. Change in perceived risk, and recent victimization, had no significant association with change in confidence. Changes in disorder, collective efficacy and confidence in the police were, again, associated with significant change in confidence in the CJS, in the expected directions; change in perceived risk, and recent victimization, were not.

Discussion

We found strong evidence to support our first two research hypotheses. Perceiving an increase in disorder over time was in the CSEW panel associated with a decline in confidence in both the police and the CJS (H1); perceiving an increase in local collective efficacy was associated with an increase in confidence, again both in the police and the CJS (H2). Evidence to support our remaining hypotheses, which related to instrumental factors, was weaker. While there was some suggestion perceived likelihood of victimization at Wave 1 was associated with lower levels of confidence in the police and the CJS, there was little evidence that an increase in perceived risk between waves was associated with a decline in confidence (H3); in the case of the CJS, increased risk may even have been associated with increased confidence. Finally, we found that between waves victimization (H4) was not associated with a decline in confidence in either the police or the CJS. Again, however, there is some evidence that people who reported victimization at Wave 1 expressed less confidence in the police at Wave 2 (although not the CJS), suggesting that victimization has some association with trust in the police, at least.

In sum, while we find evidence in support of both instrumental and expressive accounts of ‘trust and confidence’ the latter appears significantly stronger. Change in risk perceptions, and recent victimization, did not have hypothesized associations with our response variables. By contrast, collective efficacy, the most obviously ‘expressive’ explanatory variable, was a consistent predictor of confidence. On average, people whose assessments of the likelihood that others in their neighbourhood would intervene if a fight was occurring or local children were spraying graffiti grew more favourable between survey waves also tended to place more trust in the police and the CJS by Wave 2. Assessments of community efficacy on the face of it quite unconnected with the direct activity of the police and the CJS were therefore associated with trust in these institutions, which would appear to strengthen the claim that there is a significant expressive or affective strand to public trust in both. Representing and embodying efforts to maintain normative social order, the police and the CJS garner trust when collective efforts to maintain such order, as indicated by a perception that local people are willing to step in and help maintain it, seem to be strong (Jackson and Bradford, 2009).

Similar associations were found in relation to perceived disorder. On average, confidence in both the police and the CJS declined among people who perceived an increase in disorder in their local environment. In this article we have treated perceptions of disorder as an affective variable, representing perceptions of the quality of local social order. Yet perceived disorder is, as measured, something of a mixed bag, wrapping up assessments of clearly illegal behaviour (vandalism and graffiti) with other activity which might not be illegal but may certainly be threatening to some (drunk and disorderly behaviour). Concerns about disorder may therefore actually reflect a more instrumental concern with the extent of crime in local areas (Mackenzie et al., 2010). While our models controlled for actual victimization and perceived risk of sanction, and thus to an extent partialed out the overlap between ‘crime’ and ‘disorder’, this is unlikely to have been total.

On balance, however, we believe perceptions of disorder are more expressive than instrumental. Most importantly, two key measures of concern, ‘noisy neighbours’ and ‘teenagers hanging around’, deal with essentially non-criminal behaviour that many people nevertheless experience as threatening to their sense of order and cohesion. The most obvious interpretation here is that of metaphor (Mackenzie et al., 2010). Local disorder represents a set of social processes and concerns associated with

a splintering of society (that leads) … social groups to retreat from engagement with others in the public realm … to segregate dynamically into privatised enclaves … and to become increasingly intolerant of those perceived to be different or holding different values. (Mackenzie et al., 2010: 10)

If this is indeed what perceptions of disorder represent, the extent of their association with trust in the police and the CJS is striking. Trust in the police and the CJS is withdrawn when people turn away from their neighbours and, concomitantly, begin to perceive low-level disorder and criminality as more threatening (see Jackson and Sunshine, 2007).

Weaknesses of this study

We should note a number of limitations to the present study. Most importantly, although we use panel data these were collected at two time points only; we cannot therefore truly identify potential causal pathways. Multi-wave longitudinal or experimental data will be required in order to assess whether trust and confidence really is more expressive than instrumental. Furthermore, we were constrained by the variables available in the CSEW panel, and a greater variety of observed and latent variables would have enhanced the analysis. In particular our measure of collective efficacy was very limited, and a fuller set of measures might have altered the results. Equally, better-specified and more accurate measures of ‘trust and confidence’ would also have been desirable. For example, if trust in the police is composed of distinct components – for example fairness, effectiveness and engagement with the community (Bradford et al., 2009) – do these different components have different ‘triggers of change’?

Implications for Policy and Practice

What are the implications of these findings for policy and practice? First, this study emphasizes the extent to which people’s expressive or affective concerns about policing and criminal justice are important, and very possibly more important than their instrumental concerns about crime. Research has shown that there is a tendency for community-focused police activity, in particular, to be viewed as ‘nice to do but not essential’ (Foster and Jones, 2010). This view is often accompanied by the argument that if criminal justice agencies get the ‘essentials’ of crime-fighting right, public confidence will follow. The evidence presented here suggests this may not always, or even often, be the case. People’s experiences of crime per se do not seem to be strongly linked to their confidence in justice institutions, at least at the level of the ‘general’ population.

‘Crackdowns’ on crime are thus unlikely to enhance public confidence. Of course there is no reason – aside perhaps from prevailing occupational cultures that associate ‘real’ police work with action and ‘thief-taking’ (Reiner, 2010) – to believe that long-lasting reductions in crime will in any case be achieved by an enforcement model. While enforcement-focused policing can achieve demonstrable improvements in crime in the short term (Ratcliffe et al., 2011), problem-solving and related approaches may achieve more sustainable gains in the longer term (Braga et al., 1999). Indeed, policing strategies that incorporate a community engagement element may be more likely to be regarded positively by local residents. Policies and practices which resonate with the association of police, in particular, with efforts to maintain community order and stability may be more acceptable to residents, and seen as more legitimate.

Second, if perceptions of community cohesion and social order, and trust and confidence, are as inextricably linked as the current article and much other research suggests, then fostering collective efficacy seems a promising approach for improving both confidence and community safety. Such a process might create a virtuous circle – as local residents begin to trust police more, they become more ready to work both with police, other agencies and each other to solve local problems (Sampson, 2004). The potential benefits of community capacity building remain enticing – ‘greater activation of informal social control’ (Sampson, 2004: 111) should mean communities require fewer resources from formal social control agencies. If the police and other criminal justice agents are to be required to undertake community capacity building, however, they must be given the skills and tools this work demands. Direct involvement by the police in local communities that is not well managed and, for example, makes promises that are not delivered, risks being counter-productive in terms of generating long-term public trust.

Yet, the extent to which criminal justice organizations (particularly the police) should engage in community capacity building remains an open question. While police will always be required to detect crime, undertake enforcement activity and, we would argue, prioritize local, collaborative crime prevention and problem-solving activity, how far should they go in terms of wider community engagement – or building – activity? Is this role appropriate, or does it constitute ‘mission creep’ and an over-involvement of criminal justice organizations in wider public affairs?

Answers to these questions are yet to be forthcoming. It may be more realistic, given a policy agenda that continues to privilege ‘crime fighting’, to suggest the key message of this article is simply that the public take a much broader view of policing, in particular, than such an agenda would support. People are, in this sense, arguably more attuned to some of the key themes in classic police sociology than many policy-makers either know or suspect. Empirical studies of police activity (e.g. Ericson, 1983; see also Brodeur, 2010), and more theoretical accounts (Bittner, 1990), have long stressed that order maintenance is a, if not the, defining feature of policing in liberal democracies. As well as being crime-fighters, police are peace-officers (Banton, 1964), the secret social service (Punch, 1979) and street corner politicians (Muir, 1977), engaged in dealing with a wide range of usually sub-criminal activity that, nevertheless, influences the character of social order in local communities. While this order is, of course, also constituted by social, economic and cultural factors that go far beyond the activity of the uniformed police (or indeed of the CJS as a whole), they remain inextricably linked in people’s minds. From the perspective of the public, as much as students of policing, the police help reproduce order – concomitantly, when order maintenance appears successful, so do the police (Jackson et al., 2012b).

Calls for a focus in police activity on crime fighting ‘pure and simple’ thus appear sociologically and psychologically naïve. Not only do they cut against what is known about actual police activity, which is predominantly focused on non-crime-related activities, but, equally importantly, they appear ignorant of the factors that actually shape public trust, which are often highly expressive in nature and only tangentially related to crime. Naturally, it remains something of an irony that the police, and the CJS, may gain or lose public confidence as a result of social processes largely beyond the control of justice agents or organizations. Efforts to ‘enhance’ public trust must be cognizant of the extent to which external factors help shape people’s perceptions and understandings.

Footnotes

Appendix

Appendix Table 2.

Correlation matrix of response and explanatory variables.

	1	2	3	4	5	6	7	8	9	10
Disorder (Wave 1) – 1	1
Disorder (Wave 2) – 2	0.82	1
Collective efficacy (Wave 1) – 3	−0.37	−0.36	1
Collective efficacy (Wave 2) – 4	−0.40	−0.47	0.77	1
Risk of victimization (Wave 1) – 5	0.54	0.49	−0.28	−0.24	1
Risk of victimization (Wave 2) – 6	0.36	0.53	−0.30	−0.32	0.84	1
Confidence in the CJS (Wave 1) – 7	−0.27	−0.27	0.23	0.29	−0.33	−0.29	1
Confidence in the CJS (Wave 2) – 8	−0.26	−0.34	0.19	0.35	−0.27	−0.28	0.76	1
Rating of local police (Wave 1) – 9	−0.26	−0.26	0.28	0.31	−0.25	−0.23	0.41	0.36	1
Rating of local police (Wave 2) – 10	−0.28	−0.34	0.21	0.29	−0.22	−0.23	0.34	0.40	0.51	1

Notes

Author biographies

Ben Bradford is Departmental Lecturer at the Centre for Criminology, University of Oxford. His research interests include public trust and confidence in the police, legitimacy and the idea of procedural justice, and the social and cultural place of policing.

Andy Myhill is a Senior Research Officer at the College for Policing. Previously employed by the National Policing Improvement Agency, he has published both government reports and academic papers on topics including community policing and police–public contacts.

References

Banton

(1964) The Policeman in the Community. London: Tavistock.

Bittner

(1990) Aspects of Police Work. Boston, MA: Northeastern University Press.

Bottoms

Tankebe

(2011) Beyond procedural justice: A dialogic approach to legitimacy in criminal justice. Journal of Criminal Law and Criminology 102(1): 101–150.

Bradford

(2011) Voice, neutrality and respect: Use of victim support services, procedural fairness and confidence in the criminal justice system. Criminology and Criminal Justice 11(4): 345–366.

Bradford

Jackson

Stanko

(2009) Contact and confidence: Revisiting the impact of public encounters with the police. Policing and Society 19: 20–46.

Braga

Wesiburd

Waring

Green Mazerolle

Spelman

Gajewski

(1999) Problem-oriented policing in violent crime places: A randomized controlled experiment. Criminology 37(3): 541–580.

Brandl

Frank

Worden

Bynum

(1994) Global and specific attitudes toward the police: Disentangling the relationship. Justice Quarterly 11(1): 119–134.

Britt

Weisburd

(2011) Logistic regression models for categorical outcome variables. In: Piquero

Weisburd

(eds) Handbook of Quantitative Criminology. New York: Springer, pp. 649–682.

Brodeur

J-P

(2010) The Policing Web. Oxford: Oxford University Press.

10.

Brown

Benedict

(2002) Perceptions of the police: Past findings, methodological issues, conceptual issues and policy implications. Policing: An International Journal of Police Strategies & Management 25(3): 543–580.

11.

Chapman

Mirrlees-Black

Brawn

(2002) Improving Public Attitudes to the Criminal Justice System: The Impact of Information. Home Officer Research Study 245. London: Home Office.

12.

Dillman

Phelps

Tortora

. (2009) Response rate and measurement differences in mixed-mode surveys using mail, telephone, interactive voice response (IVR) and the Internet. Social Science Research 38: 1–18.

13.

Earle

Cvetkovich

(1995) Social Trust: Toward a Cosmopolitan Society. Westport, CT: Praeger.

14.

Ericson

(1983) Reproducing Order: A Study of Police Patrol Work. Toronto: University of Toronto Press.

15.

Fong

Williams

(2011) British Crime Survey: Feasibility of boosting Police Force Area (PFA) Sample Sizes Using Supplementary Recontact Surveys. Report for the Home Office. London: BMRB.

16.

Foster

Jones

(2010) ‘Nice to do’ and essential: Improving neighbourhood policing in an English police force. Policing: A Journal of Policy and Practice 4(4): 395–402.

17.

Freiberg

(2001) Affective vs effective punishment: Instrumentalism and emotionalism in criminal justice. Punishment and Society 3(2): 265–278.

18.

Gau

Brunson

(2010) Procedural justice and order maintenance policing: A study of inner-city young men’s perceptions of police legitimacy. Justice Quarterly 27(2): 255–279.

19.

Girling

Loader

Sparks

(2000) Crime and Social Change in Middle England: Questions of Order in an English Town. London: Routledge.

20.

Hardin

(2006) Trust. Cambridge: Polity Press.

21.

Hinds

Murphy

(2007) Public satisfaction with police: Using procedural justice to improve police legitimacy. Australian and New Zealand Journal of Criminology 40(1): 27–42.

22.

Hough

Bradford

Jackson

Roberts

(2013) Attitudes to Sentencing and Trust in Justice: Exploring Trends from the Crime Survey for England and Wales. London: Ministry of Justice Analytical Services.

23.

Jackson

Bradford

(2009) Crime, policing and social order: On the expressive nature of public confidence in policing. British Journal of Sociology 60(3): 493–521.

24.

Jackson

Bradford

(2010) What is trust and confidence in the police? Policing: A Journal of Policy and Practice 4(3): 241–248.

25.

Jackson

Sunshine

(2007) Public confidence in policing: A neo-Durkheimian perspective. British Journal of Criminology 47(2): 214–233.

26.

Jackson

Bradford

Hohl

Farrall

(2009) Does the fear of crime erode public confidence in policing? Policing: A Journal of Policy and Practice 3(1): 100–110.

27.

Jackson

Bradford

Hough

Myhill

Quinton

Tyler

(2012a) Why do people comply with the law? Legitimacy and the influence of legal institutions. British Journal of Criminology 52(6): 1051–1071.

28.

Jackson

Bradford

Stanko

Hohl

(2012b) Just Authority? Trust in the Police in England and Wales. Oxon: Routledge.

29.

Kautt

(2011) Public confidence in the British police: Negotiating the signals from Anglo-American research. International Criminal Justice Review 21(4): 353–382.

30.

Loader

Mulcahy

(2003) Policing and the Condition of England: Memory, Politics and Culture. Oxford: Oxford University Press.

31.

Long

Freese

(2006) Regression Models for Categorical Dependent Variables Using Stata. College Station, TX: Stata Press.

32.

Luhmann

(1979) Trust and Power. Chichester: John Wiley and Sons.

33.

Mackenzie

Bannister

Flint

Parr

Millie

Fleetwood

(2010) The Drivers of Perceptions of Anti-Social Behaviour. Home Office Research Report 43. London: Home Office.

34.

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(4): 343–367.

35.

Merry

Power

McManus

Alison

(2012) Drivers of public trust and confidence in police in the UK. International Journal of Police Science & Management 14(2): 118–135.

36.

Muir

(1977) Police: Streetcorner Politicians. Chicago, IL: Chicago University Press.

37.

Myhill

Bradford

(2012) Can police enhance public confidence by improving quality of service? Results from two surveys in England and Wales. Policing and Society 22(4): 397–425.

38.

Punch

(1979) The secret social service. In: Holdaway

(ed.) The British Police. London: Edward Arnold.

39.

Ratcliffe

Taniguchi

Groff

Wood

(2011) The Philadelphia Foot Patrol Experiment: A randomized controlled trial of police patrol effectiveness in violent crime hotspots. Criminology 49(3): 795–831.

40.

Reiner

(2010) The Politics of the Police. Oxford: Oxford University Press.

41.

Rosenbaum

Schuck

Costelloe

Hawkins

Ring

(2005) Attitudes toward the police: The effects of direct and vicarious experience. Police Quarterly 83: 343–365.

42.

Sampson

(2004) Neighbourhood and community. New Economy 11(2): 106–113.

43.

Sampson

(2009) Disparity and diversity in the contemporary city: Social (dis)order revisited. British Journal of Sociology 60(1): 1–31.

44.

Sampson

(2012) Great American City: Chicago and the Enduring Neighborhood Effect. Chicago, IL: University of Chicago Press.

45.

Sampson

Bartusch

(1998) Legal cynicism and (subcultural?) tolerance of deviance: The neighbourhood context of racial differences. Law and Society Review 32(4): 777–804.

46.

Sampson

Raudenbush

Earls

(1997) Neighbourhoods and violent crime. Science 277: 918–924.

47.

Sindall

Sturgis

(2013) Austerity policing: Is visibility more important than absolute numbers in determining public confidence in the police? European Journal of Criminology 10(2): 137–153.

48.

Sindall

Sturgis

Jennings

(2012) Public confidence in the police: A time-series analysis. British Journal of Criminology 52(4): 744–764.

49.

Singer

Cooper

(2008) Inform, Persuade and Remind: An Evaluation of a Project to Improve Confidence in the Criminal Justice System. Ministry of Justice Research Series 15/08. London: Office for Criminal Justice Reform.

50.

Skogan

(2012) Assessing asymmetry: The life course of a research project. Policing and Society 22(3): 270–279.

51.

Stone

Agulnik

Grant

Fitzpatrick

(2005) Measuring Public Confidence in the Police. Report for the Home Office, January. London: BMRB.

52.

Stoutland

(2001) The multiple dimensions of trust in resident/police relations in Boston. Journal of Research in Crime and Delinquency 38: 226–256.

53.

Sun

(2013) Public assessments of the police in rural and urban China: A theoretical extension and empirical investigation. British Journal of Criminology 53(4): 643–664.

54.

Tyler

Fagan

(2008) Legitimacy and cooperation: Why do people help the police fight crime in their communities? Ohio State Journal of Criminal Law (6): 231–275.

55.

Tyler

Huo

(2002) Trust in the Law. New York: Russell-Sage.

56.

Van Der Toorn

Tyler

Jost

(2011) More than fair: Outcome dependence, system justification, and the perceived legitimacy of authority figures. Journal of Experimental Social Psychology 47(1): 127–138.

57.

Van Der Walle

(2009) Confidence in the criminal justice system: Does experience count? British Journal of Criminology 49(3): 284–398.