Different Things to Different People? The Meaning and Measurement of Trust and Confidence in Policing Across Diverse Social Groups in London

Electronic copy available at: http://ssrn.com/abstract=1628546

1

Different Things to Different People?

The Meaning and Measurement of Trust and Confidence in Policing

Across Diverse Social Groups in London

Ben Bradford1, Scottish Centre for Crime and Justice Research, University of Edinburgh

Jonathan Jackson, Methodology Institute and Mannheim Centre for Criminology, LSE

This is an extended version of: Jackson, J. and Bradford, B. (2010). ‘What is Trust and

Confidence in the Police?’, Policing: A Journal of Policy and Practice, 4, 3, 241-248.

Abstract

One of the first actions of the new Home Secretary was to scrap public confidence as the

single performance indicator of policing in England and Wales. But public trust and

confidence will remain important to policing policy and practice. Trust and confidence can (a)

encourage active citizen participation in priority setting and the running of local services, (b)

make public bodies more locally accountable and responsive, and (c) secure public

cooperation with the police and compliance with the law. Analysing survey data from London

we find that overall 'public confidence' condenses a range of complex and inter-related

judgements concerning the trustworthiness of the police. This is the case across different

population groups and those with different experiences of crime and policing. Even recent

victims and those worried about crime seem to place less priority on police effectiveness

compared to police fairness and community alignment when responding to summary

confidence questions. We argue that confidence summarises a motive-based trust that is

rooted in procedural fairness and a social alignment between the police and the community.

This social alignment is founded upon public assessments of the ability of the police to be a

'civic guardian' who secures public respect and embodies community values (Loader &

Mulcahy, 2003). By demonstrating their trustworthiness to the public, the police can

strengthen their social connection with citizens, and thus encourage more active civic

engagement in domains of security and policing.

Key words: public confidence in policing; trust, legitimacy

1 Dr Ben Bradford, Scottish Centre for Crime and Justice Research, University of Edinburgh, 15

Buccleuch Place, Edinburgh, United Kingdom, Tel: 0044 131 651 3780, Email: [email protected]


2

Changes in performance target regimes have raised the profile of public trust and confidence

in policing in England and Wales to a level not seen since the early 1980s. Until very recently

the police worked against the statutory performance indicator derived from the British Crime

Survey (BCS) – “Percentage of people who agree that the police and local councils are

dealing with the anti-social behaviour and crime issues that matter in their area.” Police

forces were tasked by the British Government to increase the proportion of local residents

who felt confident that their local police (and council) were dealing with the things that matter

in the local community.

The positioning of public confidence as the paramount performance indicator for

police has now been abandoned. At the time of writing the new government’s policies

regarding police performance – and how to measure it – appear to be in a state of flux. But the

underlying shift away from micro-managing police and towards emphasising the relationship

between the police and public seems set to continue (Home Office 2008; 2010). The Home

Secretary may have stated that the ‘fight against crime’ is the only target the police should

work to. Yet, there is a strong emphasis within the new government on (a) making public

bodies more locally accountable and responsive, and (b) encouraging active citizen

participation in priority setting and indeed the running of local services.

While the situation remains fluid, one possible outcome might be the emergence of

more localised performance targets and measures, perhaps set by local forces or even the

putative elected police commissioners. Such local targets will attempt to capture what is

important to people living in individual police force areas. But what criteria do the public use

when judging the performance of the police? Are aspects of effectiveness (catching criminals,

responding in emergencies) paramount, or does the public place greater importance on

procedural fairness (treating citizens with fairness and dignity, giving ‘voice’ and providing

information) or community engagement (understanding and responding to the needs of the

community)? And on what basis might cooperation and participation be encouraged?

In this paper we discuss what confidence in policing means and we assess how it is

empirically measured. Comparing the PSA23 performance indicator (as the latest example of

a single indicator of confidence) with prior measures of public trust in the police, we argue

that such summary questions can be useful single measures of overall public confidence in

policing. But more nuanced viewpoints are vital if we are to properly understand public

opinion and work toward greater citizen involvement in the policies and practise of policing.

‘Global’ confidence in policing, however measured, is rooted in quite specific public

assessments of police fairness and shared values/priorities; assessments of police

effectiveness appear to be less important. This is so across diverse population groups and

those with different experiences of crime and policing. To improve global confidence, the

police must focus first on how they treat members of the public, and second on how they

engage and align themselves with the specific problems that face the community (cf. Myhill

& Quinton 2010).

The paper comprises two sections. In Section One we discuss how and how ‘global’

or ‘overall’ confidence in the police has been measured and the comparability of different

question formats. This section both reprises and elaborates on earlier work (Jackson and

Bradford 2010). Its main purpose in the current context is to lay the ground for the second

part of the paper, which addresses the need for a more nuanced understanding of public

opinions of the police and investigates whether trust and confidence mean different things to

different people. Drawing on the key distinction between fairness and effectiveness Section

Two investigates the empirical relationships between ‘trust’ and ‘confidence’ as these play

out on a population-averaged basis and within specific groups living in London. It also

introduces propensity to cooperate with police as an ‘outcome’ of trust, a relationship that

should inspire a continued interest in public opinion among police managers and policy

makers, particularly in the light of the new Government’s plans for a ‘big society’ in which

members of the public are encourage to work alongside public services, including the police

(Home Office 2010)


3

SECTION ONE: MEASURING PUBLIC CONFIDENCE IN POLICING The Government’s 2008-2011 Public Service Agreements (PSA) set out the framework in

which the recently scrapped measure of confidence operated. The PSA23 agreement aimed to

‘make communities safer,’ with Priority Activity 3 stipulating to ‘Tackle the crime, disorder

and antisocial behaviour issues of greatest importance in each locality, increasing public

confidence in the local agencies involved in dealing with these issues.’ PSA23 encouraged

local agencies and partnerships to address the priorities determined by local debate and to

promote a police force that is sensitive and responsive to a broad range of citizens’ needs.

Previous performance management regimes included measures of public opinion that

focused primarily on the views of users of police services and people from communities with

problematic relationships with the police. But the new arrangement heralded a shift to the

general population. Whatever new regime emerges, it seems probable that forces will

continue be judged against the views of all those they serve, including individuals who have

little experience, knowledge or awareness of actual police activities. Such an emphasis may

not be as unreasonable as it first might sound. The police exist to serve all citizens equally:

one section of the public should not have a higher stake in or claim to judgments about a

service which is in many ways the paradigmatic ‘public good’. Active citizen participation in

the running of a public service as important as the police requires, by definition, input from as

wide a range of people as possible. Just as importantly, research evidence increasingly finds

robust associations between trust and confidence and the readiness of public to involve,

cooperate and defer to officers (Tyler & Fagan 2008; Bradford & Jackson 2010). Since any

individual could find themselves in a situation where they might need or be able to help the

police for some reason, the opinions of all citizens are relevant.

Yet, if police performance is to be measured by public opinion surveys, the key

questions needs to be ‘fit for purpose.’ Does, for example, the PSA23 confidence question

allow survey respondents to accurately summarize their views? Or is it ‘contaminated’ by

opinions about the local council? Is the likely complexity of people’s viewpoints straight-

jacketed by single item formats? And are responses to such questions linked to more concrete

views about the police, for example ideas about how effective or fair it is? If the answers to

these questions are ‘no’, this would seriously undermine the usefulness – and indeed fairness

– of single confidence measures. Such a finding might have major implications for the new

system which is yet to emerge, and which, it might be hoped, may well rely on research

findings to shape policing policy and practice.

PSA23 and single indicators of public confidence in policing The two performance indicators for Priority Activity 3 in PSA23 were measures of public

perceptions of anti-social behaviour and a single measure of public confidence in ‘local

service providers.’ We address here only the ‘PSA23 confidence’ measure, which seems to

invite respondents to provide some form of ‘job-rating’ of the police and to some extent the

local council. The full wording is:

• It is the responsibility of the police and local council working in partnership to deal

with anti-social behaviour and crime in your local area. How much would you agree

or disagree that the police and local council are dealing with the anti-social

behaviour and crimes issue that matter in this area? [Strongly agree, tend to agree,

neither agree nor disagree, tend to disagree, and strongly disagree]

A central component of the old ‘PSA23’ performance management regime for police

in England and Wales left forces with the objective to increase the proportion of local

residents confident that their local police (and council) are dealing with the crime and anti-

social behaviour (ASB) issues that matter in the local community. The regime shifted the

emphasis firmly toward public opinion and away from the more ‘objective’ measures of

previous targets – concerning arrest rates, operational efficiency and so forth. But it retains, in

common with those earlier regimes, a core concern with the effectiveness of the police in

4

addressing crime and ASB. This concern sits naturally with both media and organisational

images, which have always positioned the police primarily as crime-stoppers and thief-takers

(Reiner 2000). It appears, on this account, that the way to improve confidence in the police is

to demonstrate to the public an adequate level of efficacy with regard to some core police

functions. Under the PSA23 regime, it was the opinions about police effectiveness that

seemed in some ways to comprise confidence, since an increased proportion answering the

question above positively would be interpreted as an increase in confidence. According to the

previous government, increasing this type of confidence will lead to increased cooperation

with the police, among the ‘law-abiding majority’ at least (Casey 2008). This emphasis is

again shared by the new regime, as evidenced by statements such as the following from the

2010 Policing White Paper: “The police are charged with keeping people safe; cutting crime

and anti-social behaviour” (Home Office 2010: 4).

However policing is and always has been about more than narrow effectiveness.

Police work possesses a heavy symbolic content, whether in relation to the link between

police, nation and state (Loader and Mulcahy 2003), the drawing and maintaining of social

boundaries (Choongh 1997; Waddington 1999), or the transmission of messages concerning

shared group membership or exclusion (Tyler 1990; Tyler and Huo 2002). While it is unlikely

that the public are uninterested in the effectiveness of the police, the more symbolic and

relational aspects of policing may be just as important in public sensibilities and trust in the

police. A good deal of research in the US and UK suggests that most important of these is the

extent to which people feel fairly treated by police, both as individuals and as members of

communities (Bradford, Jackson and Stanko 2009; Sunshine and Tyler 2003a; 2003b; Tyler

and Fagan 2008). More broadly, however, it may be that what overall confidence measures

such as the PSA23 indicator are really measuring is a broader trust in the police that is based

in large part on assessments of the motives and motivations of the police. When answering

such questions it may be that individuals express their sense of whether the police seem to

share their values and interests, and are visible and accessible representatives of the

community.

Naturally, fairness and a sense of engagement between police and public are unlikely

to be the only factors in play. The maintenance of social order defined far more broadly than

anti-social behaviour has always been a core police function, and while the new target nodded

toward disorder with its emphasis on ASB this appeared a rather narrow definition which

elided many of the order-maintenance activities police officers are engaged in on a daily

basis. An emphasis on the demonstration of police effectiveness to the public may also serve

to exclude the active participation of the public in the maintenance of social order that is

needed if police are to be truly effective. Dealing with ASB (and for that matter crime) is not

something the police can do on its own. Consideration of public opinion in this area should

not stop at confidence but extend into public behaviours that flow from confidence – namely,

in the context of this paper, cooperation with the police (and this, indeed, appears to be a key

aim of the current government). Finally, the emphasis in recent government policies on ‘the

law-abiding majority’ may occlude the views of minority or marginalised groups who might

have histories of problematic relationships with the police and/or be stigmatised as a criminal

other. What might ‘confidence’ mean for these groups, and is the relationship between

confidence and cooperation the same?

In order to investigate how people might be answering summary survey questions

such as the PSA23 measure we draw upon data from the Metropolitan Police Public Attitudes

Survey (PAS). The PAS is a large-scale, face-to-face, representative sample survey of

Londoners (running annually), which asks respondents the PSA23 question. Importantly, the

survey also probes for public views about how effective they think the police are across a

range of activities, as well as public trust in police fairness and community engagement, and it

also asks respondents the standard British Crime Survey question (‘taking everything into

account, how good a job do you think the police in this area are doing?’). The availability of

this suite of questions means answers to the global questions can be ‘tied to’ each other and to

a range of more specific opinions (see Section Two), allowing us to get under the skin of

these very broad-brush indicators.

5

The first task is to investigate whether the PSA23 measure, as a single indicator of

police performance, was ‘contaminated’ by opinions of the local council. Table 1 suggests

that the PSA23 ‘global’ confidence measure generates public evaluations of the police more

than the local council. Shown are the overall frequencies for (a) the PSA23 confidence

measure, (b) a follow-up question that asks ‘in terms of dealing with anti-social behaviour and

crime issues in this area, how would you rate the performance of the police?’, and (c) a

second follow-up question that ask ‘in terms of dealing with anti-social behaviour and crime

issues in this area, how would you rate the performance of the local council?’. The most

important findings are at the top and bottom ends of the scales. Combining ‘strongly agree’

and ‘tend to agree,’ 52 per cent of respondents agreed to some degree that police and local

councils were dealing with the issues that mattered, while 51 per cent rated the performance

of their local police as excellent or good, compared with 37 per cent who gave their local

council this rating. Similarly, 13 per cent disagreed with the first statement, while 12 per cent

gave the police a ‘poor’ or ‘very poor’ rating, compared with 24 per cent who did the same

for the council.

Table 1. Overall confidence in the local police (and council) Percentages

Strongly

agree

Tend to

agree

Neither

agree nor

disagree

Tend to

disagree

Strongly

disagree

The police and local council are dealing with the

anti-social behaviour and crime issues that matter in this area?1

6 46 35 11 2

Excellent Good Fair Poor Very poor

In terms of dealing with anti-social behaviour and crime issues in this area, how would you rate the

performance of the police?2

3 48 37 10 2

In terms of dealing with anti-social behaviour and

crime issues in this area, how would you rate the performance of the local council?3

2 35 40 20 4

Source: Weighted data from the first quarter of the 2009/2010 London Metropolitan Police Safer Neighbourhoods Survey. Total n = 5,120.

1. 1,089 (21.3%) missing

2. 1,151 (22.5%) missing

3. 1,789 (34.9%) missing

Further light can be shed by comparing the PSA23 question with the standard BCS

‘local police are doing a good job’ measure. Having greater face validity, this is a ‘tried and

tested’ standard. Inspecting the data, there appears to be a considerable amount of agreement

between the old and new measures (see Table 2). Collapsing each item into three levels,

around two-fifths of the sample believed that both (a) the local police were doing an

excellent/good job and (b) that the police and their partners were dealing with the things that

mattered in their community. Just under a fifth thought that their local police were doing a fair

job and ‘neither agreed nor disagreed’ that the police and their partners were dealing with the

things that matter in their community. Points of strong disagreement were rare. Only 1 per

cent of the sample thought that their local police were doing a poor or very poor job and that

the police and their partners were dealing with the things that mattered in their community.

Less than one in twenty thought that their local police were doing an excellent or good job

and that the police and their partners were not dealing with the things that mattered in their

community.2

2 Another way to investigate the extent of similarity between the two questions is to compare and

contrast their socio-demographic and attitudinal correlates. We found some variation. While some

socio-demographic and experiential factors had almost identical associations with the two measures,

other important characteristics – such as ethnicity, social class and victim status – had varying

relationships (for more details see Bradford & Jackson, 2010b). However, the substantive differences

were rather small.

6

Table 2. PSA23 confidence and local police ‘good-job’ measures

Percentages The police and local council are dealing with the anti-social

behaviour and crime issues that matter in this area?

Agree Neither agree nor

disagree Disagree

Total

Excellent or good 39 15 4 58

Fair 12 17 5 34

Taking everything into account,

how good a job do you think the

police in this area are doing? 1 Poor or very poor 1 2 5 8

Source: Weighted data from the first quarter of the 2009/2010 London Metropolitan Police Safer Neighbourhoods Survey.

Valid n = 3,898.

1. Gamma .597; Kappa .308

This suggests that the PSA23 measure generates public evaluations of the police that

are similar to those produced by the classic ‘are the local police doing a good job?’ survey

question. The PSA23 confidence measure and the ‘good-job’ measure are rather similar,

albeit not functionally equivalent.

Another way to investigate the extent of similarity and difference between the two

questions is to assess the socio-demographic and attitudinal correlates of each. If these were

shown to be markedly different – if, that is, individual characteristics such as age, ethnicity or

personal contact with the police predict different answers to each question – this would again

suggest that the two are not comparable and that the PSA23 question is poorly suited for

assessing police performance. To address this issue we specified two binary logistic

regression models, the first predicting the odds of agreeing that the police and local council

are dealing with the anti-social behaviour and crimes issue that matter (the PSA23 measure)

and the second predicting the odds of giving the local police an excellent or good rating

according to the old BCS question. The same set of explanatory variables was used in each.

Results from the models are summarised in Table 3.

Table 3. Correlates of the PSA23 confidence and ‘good-job’ measures

Similar correlates Different correlates

• Gender No variation by gender according to either

measure.

• Employment/economic activity status Those in full-time employment more likely to give

favourable ratings than others on both measures.

• Car access Those with access to a car more likely to give less

favourable ratings on both measures.

• Marital status Married people slightly more likely to give higher

ratings on both measures.

• Experience of contact with the police The effect of recent contact with the police on

opinions is very similar across both measures.

• Age Different patterns of variation by age across the

two indicators.

• Ethnicity White British group less likely than BME groups

to give positive rating on PSA23 measure – no

difference between White British/BME on ‘good

job’ measure.

• Housing tenure Social renters less likely to give positive rating on

‘good job’ – no variation by tenure in ratings on

PSA23 measure.

• Social class Being from social class C associated with less

favourable scores on both measures – Ds and Es

score lower on PSA23 measure only.

• Newspaper readership Reading a newspaper generally associated with

greater odds of positive opinions according to

‘good job’ – less so for the PSA23 measure.

• Victim status Recent victims of crime more likely to give lower

ratings on good job but not on PSA23.

A considerable level of variation is suggested by the findings reported in Table 3.

While some socio-demographic and experiential factors had almost identical associations

with the two measures, other important characteristics – such as ethnicity, social class and

victim status – had varying relationships. However it is important not to overestimate the size

of these effects. Appendix Table 1 shows predicted probabilities of giving favourable ratings

7

on the two indicators, generated from the models summarised in Table 3.3 It demonstrates that

even where there was significant variation in the socio-demographic correlates of the two

measures the substantive size of these was rather small. For example, the White/BME ‘gap’

on the ‘good job’ measure was a statistically insignificant 1 percentage point; according to the

PSA23 measure it was statistically significant, but still only amounted to 3 percentage points

(discrepancy due to rounding).

Overall the evidence from the socio-demographic correlates of the two questions is

somewhat inconclusive. While there was significant variation across the two indicators the

size of any difference tended to be small. On balance it seems that while, again, the two

indicators are clearly tapping into different underlying attitudes or experiences, these are not

so different as to raise serious concerns about the suitability of the PSA23 measure.

Summary of Section One The ‘take home’ message of this first section appears clear. While there are inevitably some

differences between the PSA23 and good job measures, no doubt caused in part by confusion

arising from the poor formulation of the PSA23 measure, points of similarity between the two

seem much stronger. When answering the PSA23 question PAS respondents appeared to be

thinking firstly and most importantly about the police. This is in itself an intriguing finding –

despite the poor phrasing of the indicator it behaved in a very similar manner to a question

alluding only to the police. One interpretation of this might be that the link between crime,

disorder and policing is so strong respondents largely ignored mention of the council and

focussed in the police as the institution charged with ‘doing something about it’ (despite the

significant amount of inter-agency collaboration that now exists, in some areas at least,

around issues of crime and disorder). It may be that the image of the police is so powerful, or

at least so deeply entrenched, that it is very easy for people to form an overall attitude

orientation toward it, and this dominates any other considerations raised by specific question

formats. But on what bases do the public make their overall assessments of the police? It is to

this issue that we now turn.

PART TWO: TRUST, CONFIDENCE AND COOPERATION

The relationship between ‘confidence in policing’ and ‘trust in the police’

‘To say we trust you means we believe you have the right intentions toward us and

that you are competent to do what we trust you to do.’ (Hardin, 2006: 17).

Thus far we have addressed public confidence in policing, as measured using single survey

questions. We have referred to this as overall or ‘global’ confidence. But, we contend, the

public engage in quite sophisticated inferences about the trustworthiness of the police. Public

sensibilities towards the police are fraught with issues of authority, social order and security

(Loader & Mulcahy, 2003). A trustworthy police force is seen by the public to be effective, to

be fair, and to have shared values, interests and a strong commitment to the local community

(Tyler & Huo, 2002; Sunshine & Tyler, 2003a; Jackson & Sunshine, 2007; Jackson &

Bradford, 2009). Trust extends beyond narrow public assessments that police perform their

duties effectively and efficiently to include: a sense that the police understand the needs of the

community; that they treat people fairly and with dignity; that they give them information;

and that they allow members of the community a voice to highlight local problems.

Trust is a complex and somewhat nebulous concept. But most accounts agree that it

is deeply embedded in social relationships (Tilly 2005); and the expectations people have of

each other or of institutions (Barber 1983) are central. These expectations range from the very

general (that the behaviour of the trusted party will serve to maintain and replicate the

accepted moral and social order) to the more specific (that the trustee will be technically

competent in their role and that they will place the interests of the trustor above their own in

3 The values shown in Appendix Table 1 are indicative only. They are not population proportions for

the different groups shown.

8

certain circumstances). When we trust a person or an institution we make a set of

assumptions about the way they (or it) will behave in the future. These assumptions are based

on assessments of competence, predictability, and motives (Luhman 1977; Hardin 2006).

Trust thus implies a social connection between the parties involved – the trustor must be able

to imagine that those in whom they place their trust can understand what their interests are,

and that both parties share an understanding of what it is right and wrong to do in a given

circumstance.

Within the context of the trust relationship between public and police it may be that it

is public assessments of the motives of the police are most significant. Motive-based trust

(Lind and Tyler 1988; Tyler 1990; Tyler and Huo 2002) stems from the notion that the

trustworthiness of organisations or institutions is founded in a very important sense on

estimates of character and affect, perceptions that the trustee has the best interests of the

truster at heart. Motive-based trust is primarily social rather than instrumental in character,

premised on an acknowledgement of shared social bonds. Implicated in its formation and

negotiation is communication of group membership and status – fostering the idea that police

and public are on the same side. Motive-based trust might be considered particularly

important for the police. Due to the nature of the job they may find it difficult to be

predictable or keep promises. Officers may be unable to arrive at a household at a pre-

arranged time to take a statement because another call intervened, for example. In such cases

motive-based trust provides a buffer against disappointed expectations since it provides

reassurance that the reasons for tardiness are valid.

According to the procedural justice approach motive-based trust is produced and

reproduced by and through the experience of procedural fairness. If people find authorities

treat them fairly, with dignity and respect, and that they are given a say in the processes by

which the relationship is managed, this will enhance the extent to which they trust the motives

of those authorities. Taking a slightly broader approach, we might argue that motive-based

trust will always be complemented by and formed in the light of assessments of other aspects

of police behaviour, and the other types of trust outlined above. A trustworthy police force is

seen by the public to be effective, fair, as having shared values and a strong commitment to

the local community (Tyler & Huo, 2002; Sunshine & Tyler, 2003a, 2003b; Jackson &

Sunshine, 2007; Jackson & Bradford, 2009). Trust thus extends beyond narrow

instrumentality (that police perform their duties effectively) to include a sense that they

understand the needs of the community, that they treat people fairly and with dignity, that

they give them information, and that they allow them a voice to highlight the problems facing

a given neighbourhood.

In the British context, within which strong links between police and community

continue to be ideologically important for all that they have also been attenuated by the

pressures of late modernity (from the decline in deference to the advent of New Public

Management – see for example Hough 2007), the extent to which people perceive police to be

engaged with local problems and issues may in many ways represent the extent to which they

trust the motives of the police. If people believe that police understand what is important to

them and are prepared to act on that understanding they will trust the motives of officers

within a given situation. Indeed, since community seen from an individual’s subjective

viewpoint may essentially be the individual writ large, sensing that the police understand the

interests of one’s community may in many cases be coterminous with sensing they understand

one’s personal interests.

Taken together, the three aspects or components of trust may in some sense sum

together to produce an overall orientation toward the police. Or, to put it another way, the way

survey respondents answer summary questions such as the PSA23 and old BCS ‘good job’

measures will be built up, in part at least, from their ideas about police effectiveness, fairness

and shared day-to-day priorities. If only on a pragmatic basis, then, the claim can be made

that in as much as they are separable concepts, overall confidence in the police is a product of

judgements made about its trustworthiness across the different components identified above.

However, following the work of Tyler, we should also expect people’s sense of motive-based

trust in the police to be formed in large part on the basis of assessments of the fairness with

9

which the police operate. How police treat people during face-to-face encounters, or how

individuals think officers would treat them during hypothetical encounters, will be an

important source of information concerning the extent to which they can trust the motives of

the police. There is then an implicit ordering of concepts which may be important to take into

account – trust in fairness (and perhaps effectiveness) leads to motive-based trust which leads

to overall confidence as measured by PSA23 and similar questions.

Moreover, we should examine potential variation in what different groups or

individuals want from the police, and in how they come to form their opinions of the police.

Perhaps those who are recent victims of crime place more emphasis on police effectiveness,

for example, while for those who are regularly the objects of police attention (for example for

stop and search activity) the fairness of officers may be the most important concern. While

addressing such different priorities would certainly not be impossible, it might be difficult –

how much ‘tailored’ provision can the police service provide in a era of fiscal shortage, and

how many competing priorities and needs can be reconciled? Use of the current single

measure, even in sub-group analysis, does not allow such variation to be teased out. Put

simply, we do not yet know whether opinions of the police are formed in different ways

across different population groups. If this was found to be the case the use of a single measure

averaged across the whole population would be problematic to say the least, quite aside from

the policy implications of public requirements for variegated policing practice.

Finally, ordering the concepts in the way described above also allows insight into

why overall confidence may be constructed in different ways by different groups. Returning

to the two examples briefly alluded to above may suffice to illustrate the point. Perhaps

people from ethnic minority groups with problematic relationships with the police are more

sensitive to police fairness as an indicator of the extent to which they can trust the motives of

the police, precisely because they have experienced much unfairness (and indeed have had

cause to question to motives of police officers on many occasions). On the other hand,

perhaps people who have recently been victims of crime place less emphasis on these

relational aspects of police and more on the instrumental aspects of effectiveness. In this case,

we would expect a strong direct relationship between trust in police effectiveness and overall

confidence.

Why do people cooperate with the police? A high level of ‘overall’ public confidence in policing might be a desirable end in and of

itself. But if low public confidence is to be a stick with which to beat police managers,

perhaps enhancing public cooperation is a more meaningful carrot than force’s positions in

the confidence rankings produced until recently by the Home Office. Indeed, encouraging

public cooperation with the police appears to be a key component of the policies emerging

from the new coalition government.

Tyler’s model of procedural justice and process-based policing offers a powerful

explanatory tool for understanding how and why people come to cooperate with the police

(Tyler, 1990; 2004; 2006; Tyler and Blader, 2000; Tyler and Huo, 2002). The basic premise

of the model is simple. In their dealings with legal authorities, citizens value fairness, decency

and transparency over instrumental concerns; in other words, people care more about how

they are treated than about the outcomes they receive. As noted above, the experience of

procedural justice fosters trust in the police, which is a feeling and judgement that the police

share values and goals, and this trust is powerfully linked to the legitimacy of the police.

Legitimate authorities command voluntary cooperation and compliance on behalf of those

they govern, so, by fostering police legitimacy through the use of fair process, propensities to

cooperate are enhanced. While most of the extant work on procedural justice originates from

the US, there is increasing evidence that key elements of Tyler’s model play out in Great

Britain (Bradford et al. 2009; Bradford 2009; Jackson et al. 2009). Indeed, an emphasis on

fairness and the relationship between police and public seems to spread far beyond those with

recent personal experience of the police to encompass, arguably, the majority of the

population (Jackson et al. 2009; Stanko and Bradford 2009). In terms of the three components

10

of trust outline above, the procedural justice model would suggest that fairness and

community engagement are more important than effectiveness.

At the heart of the idea of procedural justice is a social-psychological understanding

of the role of group membership in people’s lives. It is important for individuals that they feel

valued members of the social order around them. The way they are dealt with by “important

group representatives” (Tyler and Huo 2002: 167) provides individuals with information

about their place, role and status within that order. The police may constitute just such a

group representative (Sunshine and Tyler 2003a) (an idea which resonates strongly with

accounts which position the British police as representatives of nation, state and community,

see: Jackson and Bradford 2009; Loader and Mulcahy 2003; Reiner 2000; Waddington 1999).

But what of those individuals (perhaps marginalised by force or separate by choice from the

dominant social order) who do not feel shared group-membership with the police? Does the

role of procedural justice hold across diverse groups within society, or do those who feel

different group allegiances to the majority place more emphasis on instrumental outcomes

(Huo and Tyler 2000; Tyler and Huo 2002)? In short, are the routes to public cooperation

with the police different among groups who might be considered to be outside the

‘mainstream’?

Conversely, in as much as the people from excluded or marginalised groups are more

likely to be the objects of police attention and come into personal contact with officers, they

may place more emphasis on fair treatment than others. Such judgements might be rather

pragmatic acknowledgements of long histories of negative relations with the police. But

nevertheless, if people from marginalised groups were to care more about fairness this would,

on the account of the procedural justice model at least, suggest they feel at least as strong a

sense of membership of the social groups police represent than others fortunate enough not to

be seen as ‘police property’.

Tyler’s model is of course only one way of looking at the general relationship

between police and public. Despite the apparent robustness of the evidence for a dominant

procedural justice effect, it is at least plausible that some people (no matter what their

relationship to the social groups the police represent) place more emphasis on instrumental

concerns. That is, some may simply care more about how effective the police are than how

fairly they treat people or how engaged they are with local communities. Likely candidates

for such understandings may be those who have been recent victims of crime, or who are very

worried about crime and disorder. Such people may have very real reasons for caring more

about efficacy than fair treatment; and also, perhaps, for having a rather low level of overall

confidence. Furthermore, if they feel let down by perceived police failures, they may be less

ready to offer cooperation in the future.

Research Questions The discussion above suggests that individual’s trust judgements about the police sum

together in some way to produce their overall level of confidence. Furthermore, based on the

predictions of the procedural justice model, assessments of police fairness should be more

important influences on decisions to cooperate than views about effectiveness. If so, does this

hold for all social groups? We have three goals:

1. What is the relationship between ‘overall’ confidence as measured by the PSA23

indicator and other summary questions and more subtle and distinct trust assessments?

More specifically, are overall confidence, trust in police fairness, effectiveness and

community engagement separable concepts, and if they are how do the trust components

‘sum’ to produce confidence?

2. Do these relationships vary over different social groups? More specifically, do different

types of people think about the police in substantively different ways?

3. And for these same groups, what is the relationship between trust and cooperation? More

specifically, do some people take a more instrumental view, basing their decisions to

cooperate with police on assessments of effectiveness, while others, as predicted by the

procedural justice model, place more emphasis on fairness?

11

Results Our first question, then, concerns the extent to which overall confidence judgements (as

measured by for example the new PSA23 confidence indicator) summarise prior aspects of

trust. To address this issue we investigated the relationship between overall confidence

judgements (as measured by the PSA23 confidence target and other summary indicators such

as the old good job indicator) and three dimensions of trust: effectiveness, fairness and group

engagement. If the summary questions are indeed tapping into more subtle and precise

assessments of the police, then we should find a strong association between overall

confidence and some or all of the different dimensions of trust.

On the other hand, some distinction in opinions across the different dimensions

should also be expected. If there was little or no variation this would suggest that Londoner’s

opinions of the police draw on some single underlying trait or orientation, perhaps an ‘idea’

of the police that bears relatively little relationship to things officers and organisation actually

do. Such a finding would be sociologically and psychologically interesting, but it would offer

little in the way of insight to those seeking ways to enhance public confidence.

Latent variable modelling (in the statistical package Mplus 5.0) was the primary

method of analysis. We treat the different dimensions of trust are treated as ‘latent’ variables

measured by a set of observed indicators – that is, survey questions. We assess the level of

trust that people hold in police effectiveness (for example) by using their answers to a series

of questions about how effective police are across a number of tasks to ‘build up’ a picture of

their underlying assessment of its efficacy. Such a process is more accurate than simply

asking them ‘how effective do you think the police are?’.4

Using confirmatory factor analysis (CFA) we tested a series of statistical models to

investigate the dimensionality of confidence and trust. These models specified ‘trust and

confidence’ in one of five different ways:

Model 1: One single underlying ‘thing,’ which spans overall confidence and trust in

police effectiveness, fairness and community engagement;

Model 2: Two underlying ‘things’: overall confidence; and trust in police

effectiveness, fairness and community engagement;

Model 3: Two underlying ‘things’: trust in police effectiveness; and overall confidence

combined with trust in fairness and community engagement;

Model 4: Three underlying ‘things’: overall confidence; trust in police effectiveness;

and trust in police fairness and community engagement; and,

Model 5: Four underlying ‘things’: overall confidence; trust in police effectiveness;

trust in police fairness; and trust in police community engagement.

So Model 4, for example, specified that three underlying factors, or latent variables, explained

much the variation in the questions used in the analysis: (a) overall confidence, (b) trust in

police effectiveness, and (c) trust in police fairness and community engagement. Using CFA

allows us to test statistically whether this specification fits the data well – that is, whether the

data support the idea that the opinions represented by the observed measures used in the

model can be thought of as being formed on the basis of the three underlying latent

orientations.

Results from the five models are shown in Table 1. Both exact and approximate fit

statistics suggest that Model 1 fitted the data poorly. Opinions of the police in London seem

not to be one homogenous ‘syndrome.’ People do distinguish between different aspects of

police performance and behaviour when making judgements about the trustworthiness of the

police. Model 2, which specified overall confidence as being separate from the more specific

trust judgements, also fitted the data poorly. The primary reason for this poor fit is

demonstrated by Models 3 to 5. More specifically, when trust in police effectiveness was

specified to be separate from trust in police fairness and community engagement the models

4 See the Appendix for a complete list of the survey questions used.

12

all fitted well, according to the approximate fit statistics,5 no matter how the rest of the model

was specified. At the bare minimum, then, we need to distinguish between on the one hand

trust in police effectiveness, and on the other hand trust in police fairness and engagement.

Table 1Confirmatory Factor Analysis testing the underlying dimensions of 'confidence'

Model 1 Model 2 Model 3 Model 4 Model 5

1 single factor 2 factors 2 factors 3 factors 4 factors

Trust in police effectiveness

Tackle gun crime .45 .45 .73 .73 .73

Tackle drug crime .45 .45 .78 .78 .78

Tackle dangerous driving .46 .46 .73 .74 .74

Trust in police fairness

Treat you with respect .72 .73 .73 .74 .77

Treat everyone fairly .78 .78 .79 .79 .83

Are friendly and approachable .76 .76 .77 .77 .82

Trust in police community engagement

Understand community issues .84 .84 .84 .84 .85

Dealin with the things that matter .87 .87 .87 .87 .88

Listen to local people's concerns .83 .83 .84 .84 .85

Overall confidence

Good job local area .65 .67 .66 .67 .67

Confidence in police in this area .86 .89 .86 .89 .89

PSA23 measure .55 .57 .60 .57 .57

Chisquare 37246.8 37614.2 5750.0 4913.7 3436.0

p <0.000 <0.000 <0.000 <0.000 <0.000

CFI .717 .714 .957 .963 .974

TLI .893 .892 .986 .987 .991

RMSEA .252 .253 .093 .087 .075

Source: Metropolitan Police Public Attitudes Survey 2008/09

But as noted, Models 3, 4 and 5 fitted the data well. Furthermore, inspection of the

correlations between overall confidence and trust in police fairness and engagement (in

models 4 and 5) suggests a very strong association. It appears on this basis that people in

London draw on very similar underlying views or orientations when thinking about how fair

the police are and how engaged it is with local communities. Equally, Londoner’s overall

level of confidence in the police is very closely related to how fair and engaged they think it

is.

However, it is also true that Model 5 fitted the data better than Models 4 and 5,

according to the approximate and especially the exact fit statistics. Furthermore, there is

evidence that aggregate levels of trust in fairness, community engagement and overall

confidence react differently to police communication and other activities (Hohl, Bradford and

Stanko 2010). While these may be very closely related constructs they are not inseparable.

This might be taken to suggest that, where possible, the indicators should be kept separate.

But on the whole it seems to be a pragmatic choice whether one thinks of trust in fairness,

trust in engagement and overall confidence as one ‘thing’ or as three extremely highly

correlated phenomena. At the very least we can conclude – in response to the first research

question laid out above – that overall confidence measures are very closely related to active

assessments of police behaviour that relates to personal treatment (particularly fairness) and

engagement with the community, and that these assessments are empirically separable from

judgments about police effectiveness.

Finally, the data shown in Table 1 confirm earlier findings that, despite

understandable concerns, the new PSA23 measure is very closely related to more specific

5 Note that as is common practise more credence is given to the approximate rather than the exact fit

statistics. The extremely large sample size of the PAS means that the exact fit statistics are of little

practical use in determining how well a specific model fits the data.

13

opinions of the police, and as such probably is suitable for use as an overarching public

confidence target (Jackson and Bradford 2010). The question not only scales well with other

summary question such as the BCS ‘good-job’ measure, but it is also closely related to trust in

both police fairness and community engagement.

Does confidence in the police mean different things to different people? Confidence measures seem to tap most closely into public judgements that the police ‘are on

their side.’ Such a judgement is based more around notions of fairness and engagement rather

than effectiveness. Yet this raises an important issue. By lumping all individuals together in

one group – ‘the public’ – we may miss out on important variation between key groups in the

general population. This variation may emerge, for example, through differences between

those who are the object of police attention, those who draw upon the police as a service and

resource, and those who look to the police to keep them safe and tackle crime.

To address our second research question we estimated a simple structural equation

model in which, firstly, trust in police fairness and effectiveness predicted trust in police

community engagement. We see this latter opinion set police as a British version of motive-

based trust, turning on whether the police are seen to represent the interests of the community

and share the values, goals and priorities of community members. As specified in the model

people formulate their sense of motive-based trust in the police on the basis of judgements or

inferences about the fairness and effectiveness with which it operates (and, crucially, the

model will allow assessments of which is more important). Secondly, all three components or

elements of trust predict overall confidence in policing. All the latent variables in the equation

were predicted by observed indicators identical to those shown in Table 2, with one

exception: ‘overall’ confidence was measured by the PSA23 indicator and the local/London

good-job ratings, thus allowing for the fact that people from, for example, some ethnic

minority groups might have a different relationship with ‘the Met’ (responsible for the

Stephen Lawrence enquiry and the de Menezes shooting, perhaps) and the police in their

particular borough or local area. The model tested was therefore as specified in Figure 1.

Figure 1. Trust and confidence in the police

The model shown in Figure 1 suggests a number of things. First, as demonstrated by

the CFA modelling above, trust in police fairness and trust in police community engagement

are very strongly related. However in contrast to the previous model trust in police fairness is

here given a formative role, one interpretation of which is to say that trust in community

engagement is formed in large part on the basis of trust in the fairness of police officers in

their interactions with members of the public – fairness leads to a sense of motive-based trust.

People infer the level of engagement of local police to a large extent on the basis that they

14

perceive officers to be fair and respectful (although, as other work has shown, trust in police

community engagement can also be fostered in other ways, for example by the provision of

tailored information about local police activities – Hohl et al. 2010). There is also another,

complementary, interpretation – that the effect of trust in police fairness on overall confidence

is mediated by trust in police community engagement which, again, we might see as a

peculiarly British version of Tyler’s motive-based trust.

Trust in both police fairness and community engagement has strong and substantively

large statistical effects on overall confidence. By contrast trust in police effectiveness has

only a relatively weak association with overall confidence (and with the community

engagement measure). On this basis it appears that overall confidence is built up most

importantly from the public’s trust that the police are fair and engaged with local

communities. Despite the wording of the PSA23 measure perceptions of the effectiveness of

the police in ‘fighting crime’ are considerably less important, although they are certainly still

relevant.

The model shown in Figure 1 was tested on a range of different population sub-

groups. Results from these sub-group analyses are shown in Tables 2 and 3. Looking first at

the patterns by ethnic group shown in Table 2, the extent of the similarity across the different

groups is striking. The top part of the table shows the measurement component of the model,

and is exactly analogous to model 5 in Table 1 (note that the coefficients for the indicators of

trust in effectiveness, fairness and community engagement are the same in model five in

Table one and in the ‘all people’ column in Table 2). It shows that for each group shown the

measurement properties of the latent constructs representing the various forms of trust and

overall confidence are very similar. In terms of the survey questions used in the analysis, at

least, people from all the ethnic groups shown think about trust in police fairness, for

example, in highly comparable ways, placing very similar weights on respect, fairness and

friendliness. There is only one really significant exception to this pattern – trust in police

effectiveness among those from the Mixed group, who appear to place less stress on

effectiveness in dealing with drug crime and dangerous driving than those from other groups.

This may be explained by the very young age profile of this group (Dobson, Zealey and

Brown 2006) – perhaps particularly in London traffic policing may be less relevant to many

young people, while some may have rather ambivalent opinions of the way police drug crime.

Overall, however, it appears that the three components of trust, and overall confidence, are

‘built up’ in very similar ways across the diverse groups shown.

15

Table 2Structural equation modeles using components of trust to predict 'overall' confidence: sub-group analysis by ethnic group

Standardised coefficients

White British/ Pakistani/ Black Black

All people Irish Mixed Indian Bangladeshi Caribbean African

MEASUREMENT COMPONENT


Tackle gun crime .73 .72 .73 .76 .78 .75 .76

Tackle drug crime .78 .79 .60 .84 .80 .80 .79

Tackle dangerous driving .74 .74 .57 .84 .79 .75 .72


Treat you with respect .77 .77 .74 .73 .82 .79 .80

Treat everyone fairly .83 .83 .78 .81 .87 .82 .83

Are friendly and approachable .82 .83 .74 .82 .84 .80 .87


Understand community issues .85 .86 .81 .87 .82 .88 .84

Dealin with the things that matter .88 .89 .84 .89 .89 .90 .89

Listen to local people's concerns .85 .86 . .84 .84 .88 .87

Overall confidence

Good job local area .67 .65 .70 .62 .72 .70 .68

Confidence in police in this area .88 .90 .83 .88 .89 .87 .87

PSA23 measure .57 .58 .66 .56 .41 .68 .

STRUCTURAL COMPONENT

Effectiveness --> engagement .07 .08 .00 .06 .10 .04 .03

Fairness --> engagment .88 .86 .94 .86 .89 .90 .91

Effectiveness <--> fairness .29 .28 .23 .37 .28 .51 .36

Effectiveness --> confidence .15 .17 .09 .13 .16 .07 .12

Fairness --> confidence .10 .10 .27 .02 .03 .17 .10

Engagement --> confidence .82 .82 .70 .90 .89 .79 .84

Proportion of variance explained

Trust in police community engagement .81 .74 .87 .78 .85 .85 .85

Overall confidence .95 .94 .97 .94 .98 .97 .98

FIT STATISTICS

Chisquare 3436.1 2352.8 137.5 171.9 399.7 302.4 243.7

p <0.000 <0.000 <0.000 <0.000 <0.000 <0.000 <0.000

CFI .97 .97 .99 .98 .97 .96 .98

TLI .99 .99 .99 .99 .98 .99 .99

RMSEA .08 .09 .05 .08 .10 .09 .08

Unweighted n 20214 11198 1735 836 1322 1080 1511

Note: Due to model fit problems only two indicators were used to measure trust in police community engagement for the Mixed groups and overall confidence

for the Black Carribean group. The models for these groups are therefore not strictly comparable with those for the other groups.


A very similar pattern can be seen in the structural component of the models

summarised in Table 2. The relationships between the latent variables are similar across all

the groups. Regardless of whether we are talking about ethnic minority groups with long

histories of adversarial relations with the police, or the ethnic majority whose members much

less frequently finds themselves to be the object of police attention, people place most

importance on the fairness of police behaviour. Trust judgements premised on the idea of

fairness overwhelming drive the sense that the police are (or of course are not) engaged with

the community and are a valid repository of motive-based trust and, through this, overall

confidence. The Mixed group is again a slight outlier – note that the unmediated direct path

from fairness to overall confidence is somewhat stronger for this group, something which

again may reflect its younger age profile and, perhaps, less stress on the link between police

and community. By contrast, there is a hint in the data that those from the Indian and

Pakistani groups place even more stress on such a link than others, although in substantive

terms the differences are likely to be very small.

Finally, the approximate fit statistics suggest that the model fits the data from

moderately to very well for all the groups shown (comparison of the exact fit statistics is

unsuitable because of the different sample sizes shown). Since the fit statistics apply primarily

to the measurement component of the models, this is further evidence that the latent

constructs specified do indeed mean largely the same thing to the different groups. On the

basis of the structural component of the models, however, it also seems justifiable to claim

that the relationships between them are also largely similar.

To investigate the extent of variation in trust and opinions among those with different

experiences of crime and policing, Table 3 repeats the above process for, respectively, those

who had and had not been recent victims of crime; those who were ‘unworried’ about crime

16

compared with those who were ‘functionally’ and ‘dysfunctionally’ worried (see Jackson and

Gray 2009); and those with no recent contact experience with the police compared with those

who had had positive and negative recent contact experiences.

Table 3Structural equation modeles using components of trust to predict 'overall' confidence: sub-group analysis by experienece of crime and the police

Recent victim of crime? Worry about crime Recent contact with police

Yes No Dysfunctional Functional Unworried None Negative Positive



Tackle gun crime .75 .73 .67 .70 .76 .73 .77 .72

Tackle drug crime .85 .77 .77 .74 .78 .77 .85 .77

Tackle dangerous driving .69 .74 .69 .70 .78 .75 .65 .66


Treat you with respect .80 .76 .76 .77 .77 .76 .82 .75

Treat everyone fairly .84 .83 .83 .84 .83 .83 .82 .77

Are friendly and approachable .83 .81 .83 .81 .82 .81 .86 .76


Understand community issues .85 .85 .82 .85 .88 .85 .85 .82

Dealin with the things that matter .89 .88 .89 .90 .88 .88 .87 .86

Listen to local people's concerns .88 .85 .86 .83 .85 .84 .87 .82

Overall confidence

Good job local area .73 .65 .67 .72 .65 .65 .75 .55

Confidence in police in this area .90 .88 .91 .88 .89 .88 .91 .87

PSA23 measure .62 .56 .55 .55 .57 .54 .68 .52


Effectiveness --> engagement .14 .05 .10 .11 .04 .06 .16 .05

Fairness --> engagment .79 .90 .86 .88 .89 .89 .80 .85

Effectiveness <--> fairness .43 .27 .32 .18 .29 .27 .33 .24

Effectiveness --> confidence .16 .14 .17 .21 .11 .15 .16 .11

Fairness --> confidence .11 .12 .01 .23 .17 .11 .06 .16

Engagement --> confidence .80 .81 .89 .69 .77 .82 .85 .77


Trust in police community engagement .74 .83 .79 .84 .81 .82 .75 .74

Overall confidence .96 .95 .96 .97 .93 .95 .97 .90

FIT STATISTICS

Chisquare 367.5 3076.5 1152.9 343.0 2012.4 2947.9 134.9 407.6

p <0.000 <0.000 <0.000 <0.000 <0.000 <0.000 <0.000 <0.000

CFI .98 .97 .97 .98 .98 .97 .99 .96

TLI .99 .99 .99 .99 .99 .99 .99 .98

RMSEA .07 .08 .09 .07 .08 .08 .06 .09

Unweighted n 2368 17846 5096 2208 12552 17332 1089 1793


The extent of the similarity across the models shown in Table 3 is again striking. In

every case the measurement components of the models were highly comparable. More

importantly, the relationships between trust in effectiveness, fairness, engagement and overall

confidence were also very similar. So, for example, for both recent victims of crime and non-

victims the perceived fairness of the police was by far the most powerful predictor of trust in

community engagement, which was then the most powerful influence of overall confidence.

This pattern repeated across the different types of fear and contact experiences. So the

experiences of recent victims of crime (or those fearful of crime) do not appear to cause them

to place more importance on the instrumental aims of policing; conversely, lack of recent

contact does not seem to attenuate the importance of police fairness. Trust and confidence

appear to mean broadly the same thing across all the groups shown.

That said, there are some hints in the data that variations in recent experiences of

crime and the police are associated with some small variations in the ways opinions are built

up. Most notably, the association between effectiveness and engagement is slightly stronger

for the ‘victim’ and perhaps the ‘worried about crime’ groups, pointing to a greater salience

of instrumental concerns among these people. These variations are slight, however, and in no

way detract from the general conclusion drawn above.

In sum, then, it appears that research question two can be answered largely in the

negative. Among the population sub-groups analysed in Tables 2 and 3 there appears to be

very little variation in the way trust in the police is formulated, or in the ways the different

components of trust are related and in turn influence ‘overall’ confidence. Furthermore, the

response to question one can be further specified – overall confidence is built up most

importantly from assessments of police fairness as these are mediated by motive-based trust.

Does the relationship between trust and cooperation vary across different groups?

17

What then are the routes to cooperation with the police? How can public assistance in the

maintenance of social order be enhanced? And do the potential mechanisms vary across the

key groups of interest here? To address our third research question we repeated the process

used in answering the second, estimating a series of simple structural equations models which

used trust in police fairness and effectiveness. But this tine, instead of having overall

confidence as the final outcome variable, we included self-assessed propensity to offer

support and cooperation to the police. Such cooperation can be considered as either the

outcome of the legitimacy of the police (Sunshine and Tyler 2003b) or as constitutive of that

legitimacy (Bradford and Jackson 2010; c.f. Beetham 1990). But this distinction is not

important for current purposes. The aim is to investigate the relative roles of fairness and

effectiveness in predicting cooperation, not to produce a detailed analysis of the complex

relationship between trust, legitimacy and acts of cooperation (see Bradford and Jackson 2010

for a discussion). Results from the models are shown in Tables 4 and 5.

There are a number of points to be made in reference to the models used to address

our third research question. Firstly, data from only one quarter (Q1 2009/10, n=5120) of the

PAS were used, as this was the only quarter of the survey which fielded the cooperation

questions. The dataset obtained did no contain the victim of crime variable, so this had to be

dropped from the analysis. Secondly, engagement was also dropped from the analysis. This

was done on both pragmatic grounds (the models for some groups would not fit if

engagement was included as before, probably due to problems caused by the high correlation

between fairness and engagement being exacerbated by the small sample sizes in some

groups), and also to sharpen the focus on the key distinction developed here, that between

relational concerns about police fairness and instrumental concerns about police effectiveness.

Finally, due to a lack of question comparable with the 2008/09 survey worry about crime is

represented by only two groups, the worried and the not-worried.

Table 4Structural equation modeles using trust in police effectiveness and fairness to predict stated propensity to cooperate with

the police: sub-group analysis by ethnic group


White British/ Pakistani/ Black Black

All people Irish Mixed Indian Bangladeshi Caribbean African



Tackle gun crime .70 .68 .76 .91 .84 .64 .69

Tackle drug crime .85 .86 .93 .83 .76 .90 .80

Tackle dangerous driving .77 .79 .77 .72 .63 .79 .83


Treat you with respect .83 .80 .87 .90 .81 .86 .90

Treat everyone fairly .85 .82 .83 .86 .92 .86 .87

Are friendly and approachable .77 .77 .87 .72 .74 .81 .75

Propensity to cooperate

Call police to report a crime .96 .94 .98 .99 .98 .99 .97

Report suspicious activity .90 .92 .86 .90 .96 .85 .92

Provide information to help find a criminal .94 .96 .93 .84 .92 .90 .89


Effectiveness --> cooperation -.21 -.24 .02* -.43 -.31 -.18 -.10*

Fairness --> cooperation .36 .34 .26 .62 .37 .40 .43

Effectiveness <--> fairness .35 .32 .46 .37 .37 .43 .39


Propensity to cooperate .12 .12 .07 .37 .12 .13 .16

FIT STATISTICS

Chisquare 280.2 163.2 44.6 15.7 54 41.6 69.9

p <0.005 <0.005 <0.005 .21 <0.005 <0.005 <0.005

CFI .99 .99 .99 .99 .99 .99 .99

TLI .99 .99 .99 .99 .99 .99 .99

RMSEA .06 .07 .09 .03 .11 .09 .11

Unweighted n 5114 2715 340 276 333 306 360

* Not significant at the 5% level.


Bearing these provisos in mind, the evidence from Table 4 appears straightforward,

with a strongly significant and substantively large observed association between trust in

police fairness and stated propensity to cooperate. Holding across all the ethnic groups shown,

it does not appear that being from an ethnic minority (which might be seen as membership of

18

a social group with weaker links to that the uniformed police represent, indeed arguably

embody, namely the British nation state as a community of affect – Loader and Mulcahy

2003) was associated with any less interest in the relational aspects of police-community

relations. Indeed, if anything the link between fairness and cooperation was stronger for most

ethnic minority groups than for the White British/Irish majority.

In most cases, the association between trust in police effectiveness and cooperation

was negative. That is, people who thought the police effective in dealing with crime were on

average less likely to say they would cooperate. This finding stands in contrast to that

reported by Bradford and Jackson (2010) who found a weak but statistically significant

positive association between trust in police effectiveness and intention to cooperate with the

police. One explanation might be that the wording of the questions used had a perverse effect:

the phrasing of the preamble – ‘how likely would you be to do the following things’ – is such

that some respondents might interpret it to mean ‘how likely is it that the following things

might happen about which you might have to do something’ (see the Appendix for the precise

question wording). If the police are considered more effective, this might reduce the

likelihood that they would need to be contact/assisted. However, further analysis

demonstrated that the negative association between trust in effectiveness and cooperation

(conditional on trust in fairness) was retained even when control variables representing

respondents’ views about the prevalence of crime and disorder in their local areas were

introduced, making it appear unlikely that ‘cooperation’ was being interpreted by PAS

respondents as ‘probability that I might have to’.

There is another possibility. As noted, the negative path from trust in effectiveness to

cooperation was conditional on trust in police fairness (and these two constructs were

themselves positively correlated). If fairness was removed from the model there was no

significant association between effectiveness and cooperation. One explanation might then be

that we are detecting a residual effect from trust in police effectiveness, such that if the police

are trusted to be effective, then active cooperation is less likely to be offered because it is

assumed the police (or other official body) will intervene. This could be interpreted as

suggesting that, at the margin, people are less likely to engage in informal social control, or

perhaps ‘new parochial’ social control, i.e. active cooperation with agents of formal social

control (see Carr 2003), if they have high levels of trust in the efficacy of those agents.

Considerably more work would be needed to flesh out this idea, and it does not detract from

the fact that if people have low levels of trust in police fairness they are on average much less

likely to say they will offer cooperation.

Turning finally to the picture across different experiences of the police and worry

about crime a very similar pattern emerges. Trust in police fairness retains a dominant role.

There are though some interesting hints that those with recent negative contact with the police

may place greater emphasis on police fairness than those with positive or no contact;

certainly, compared with the positive contact group the link between fairness and cooperation

is stronger for those who have had recent negative contact with police. Equally, looking at

differences by worry about crime, fairness retains a key role among the worries and the

unworried. An intriguing feature of the models covering worry about crime is the variation in

the effect size of the effectiveness to cooperation path (-.26 for the worried, -.08 for the

unworried). As above this relationship needs to be further investigated before any firm

conclusions can be drawn.

To sum up, the third research question posed above it seems that the role of fairness

in promoting (or undermining) propensities to cooperate with the police is very similar across

the diverse groups analysed above. The data strongly support a dominant role for a procedural

justice like effect. By contrast, the role of trust in police effectiveness in influencing

cooperation is contradictory and somewhat perplexing. Further exploration of the role the

demonstration of effectiveness might have in fostering cooperation would be a useful avenue

of research.

19

DISCUSSION So what do people want from the police: fairness or effectiveness? On the basis of the

evidence presented here the answer appears clear. Echoing other UK studies (for example

Bradford, Jackson and Stanko 2009; Myhill and Beak 2008), we found that averaged across

the whole population (here of London) fairness was valued over effectiveness by some

margin, whether in terms of the way ‘overall confidence’ was built up or in terms of what

might helps explain variation in intentions to cooperate with the police. Moreover, this

relationship held even when the analysis focussed on specific population groups – people

from ethnic minority groups, those worried about crime, and those with recent, and different,

contact experiences. People who have been recently victimised or who are worried about

crime are not more instrumentally oriented, while conversely those with recent contact

experiences do not place substantially greater emphasis on fairness than those with none. An

emphasis on fairness, engagement and perhaps above all the relationship between police and

public consistently trumped assessment of police effectiveness.

It appears then that efforts to improve public confidence in policing should focus

(above all) on promoting the fairness with which officers treat – and are seen to treat – those

members of the public with whom they have contact. In terms of the models presented here

fairness and respect during interpersonal interactions may: promote the idea that the police are

engaged with and supportive of the community; be the most important ‘driver’ of overall

confidence; and, crucially, be strongly linked to active cooperation with the police.

That these associations also held across the different ethnic minority groups analysed

opens up some interesting possibilities in terms of the relationship between those groups and

the police and, perhaps, that which the British police are held to represent. Recall that

Table 5Structural equation modeles using trust in police effectiveness and fairness to predict stated propensityto cooperate with the police: sub-group analysis by experience of crime and policing


Worry about crime Recent contact with police

Worried Not Worried None Negative Positive



Tackle gun crime .77 .62 .68 .75 .75

Tackle drug crime .84 .83 .85 .90 .84

Tackle dangerous driving .73 .75 .77 .75 .77


Treat you with respect .86 .82 .82 .87 .83

Treat everyone fairly .84 .85 .86 .88 .71

Are friendly and approachable .73 .80 .77 .78 .73

Propensity to cooperate

Call police to report a crime .94 .96 .97 .90 .94

Report suspicious activity .93 .88 .90 .84 .93

Provide information to help find a criminal .93 .95 .93 .93 .99


Effectiveness --> cooperation -.26 -.08 -.24 -.15 -.04*

Fairness --> cooperation .40 .34 .34 .44 .35

Effectiveness <--> fairness .35 .33 .33 .48 .22


Propensity to cooperate .16 .10 .120 .15 .12

FIT STATISTICS

Chisquare 127.1 174.4 257.0 40.9 36.2

p <0.005 <0.005 <0.005 <0.005 <0.005

CFI .99 .99 .99 .99 .99

TLI .99 .99 .99 .99 .99

RMSEA .07 .06 .07 .08 .05

Unweighted n 2007 3035 4138 410 635

* Not significant at the 5% level.


20

according to the procedural justice model individuals value fairness and respect because it

communicates shared group membership and status. Decent treatment at the hands of the

group authorities such as the police promotes a sense that one is ‘included’ in the social

structures they embody. If individuals do not feel they share the same social group as the

police they may care much less about how they are treated and take a more instrumental

stance towards it (Tankebe 2008), one which could tip over into outright opposition (if, for

example, instrumental aims promoted a need or desire to remove police influence from an

area or social situation). Conversely, those who feel themselves to be on the margin of the

group may be even more sensitive to fairness, since they feel their position is uncertain and

are especially attuned to status messages. Yet, on this basis at least people from ethnic

minorities in London appear to see themselves firmly in the ‘mainstream’. There is little to

suggest they place any more or less emphasis on procedural fairness – and therefore on the

status messages such fairness, or its absence, transmits – than their counterparts from the

White British group. Conversely, of course, the ethnic majority place no less importance on

fairness than minority groups. Being ‘in’ the ethnic group most closely associated with the

British police (in terms of image and reality, given the problems associated with ethnic

minority recruitment to the MPS) does not appear to promote a more instrumental approach to

policing.

CONCLUSIONS We have assessed in this paper the survey question (included in both the BCS and PAS) that

asks its respondents whether they agree or disagree that the ‘police and local councils are

dealing with the anti-social behaviour and crime issues that matter in their area.’ Answers to

the PSA23 measure seem to draw on the same underlying ideas and orientations as do

answers to the question which concerns only the police. On the basis of the evidence

presented here, at least, use of the PSA23 measure did not appear to place the police at any

major disadvantage compared with the older question. The widespread fears that

accompanied its introduction may well have been mis-founded.

Naturally, the abandonment of the PSA23 target means that such concerns have in

effect been answered. But the analysis presented above underlines a more fundamental point.

Single item questions such as the PSA23 measure and the old ‘good-job’ question can act as

useful methodological tools that condense a wide range of opinions into an easily digestible

nugget. If the objective is to know at a high level what public perceptions of the police are in

a particular area (or indeed nationally), then asking people a question similar to those

discussed here is a useful place to start.

But deeper concerns remain. How can the public be expected to have extensive

knowledge of the crime problem, the activities of the police, and the impact of police

activities on crime and disorder? Can we expect people to judge the performance of the police

in these terms? Surely public perceptions of the police will be stereotypical and rather one-

dimensional.

On the contrary, people may have rather subtle views that are rooted in complex

notions of trust and closely linked to the position of the police in British social, cultural and

political life. Individuals can, and do, make nuanced and considered judgements about the

police. Existing work suggests that ‘overall confidence’ question tap most strongly into trust

in police fairness and shared values/interests. The findings presented here are consistent not

only with other work based on the PAS (see for example Bradford and Jackson 2010a), but

also recent analysis of the BCS (Myhill & Beak 2008) and the US-based work of Tyler and

colleagues (Tyler 2006; Tyler & Huo 2002). It may be that procedural fairness generates

motive-based trust and a sense of shared group membership, encouraging the idea that

citizens and the police are ‘on the same side’ and that the police are ‘civic guardians’ that

secure public respect and embody or protect community values (Loader & Mulcahy, 2003:

79-89).

The evidence presented here has some important policy implications. Recall that the

stated aim of PSA23 was to improve public confidence in policing by dealing with the crimes,

disorder and ASB issues that were important to local people. It thus appeared to prioritise

21

effectiveness over other aspects of police behaviour. This is certainly an emphasis that the

new government seems to share. On this account, the demonstration of effectiveness in the

fight against crime should be sufficient to generate public approval and support for the police.

Yet, it is the procedural fairness of the police and a sense of motive-based trust that are

consistently found to be most important to people. This may create a tension between the

intention of government policy (narrowly interpreted), what forces might think they need to

do to in response to policy-setting, and what might be the best ways to actually generate

public cooperation and ‘buy-in’ to the way policing services are provided. For example,

improving the quality of police-public interactions and re-engaging as a visible and accessible

presence in the community are, on the face of it, some way removed from narrowly

instrumental ‘thief-taking’.

We might take heart, however, from some of the stronger claims of Tyler’s

procedural justice model. If people perceive the police to be procedurally fair, and if they trust

their motives in behaving the way that they do, all current evidence suggests they are not only

more likely to actively cooperate by reporting crime, cooperating in investigations, providing

witness evidence, even intervening in situations of low-level deviance and incivility. They are

also more likely to defer to officer’s instructions and obey the laws that the police in many

ways still embody. In the long run, the fight against crime might be more efficiently, more

cost-effectively, and certainly more ethically served by treating the public with fairness,

dignity and respect than by instigating another ‘crack-down’ on crime (Hough et al.,2010).

To conclude, the public seem to see the job of the police as not just dealing with

crime, but as rooted also in the defence of civility and community, in treating people fairly

and with dignity, and in being aligned and responsive to local needs and issues. Properly

understood, the effectiveness of the police – in the public’s mind at least – may be revealed

not only in the way the police deal with crime, but also by officers being there for victims,

treating people fairly, and providing a visible and accessible source of moral authority

(Jackson & Sunshine, 2007; Myhill & Quinton 2010). If the police demonstrate to citizens of

diverse communities that they are effective, fair and aligned with local interests, then this not

only makes the service more directly accountable. It also strengthens the moral connection

between people and their police, thus encouraging greater civic participation and more active

public engagement in domains of security, policing and the regulation of social and

community life.

22

References

Barber, B. (1983). The Logic and Limits of Trust. New Brunswick, Rutgers University Press.

Beetham, D. (1991). The Legitimation of Power. London, Macmillan.

Bradford, B. (2010). ‘The Quality of Police Contact: Procedural Justice Concerns among

Victims of Crime in London’. Available at SSRN: http://ssrn.com/abstract=1596754

Bradford, B. and Jackson, J. (2010). ‘Cooperating with the Police: Social Control and the

Reproduction of Police Legitimacy’. Available at SSRN:

http://ssrn.com/abstract=1640958. Bradford, B., Jackson, J. and Stanko, E.A. (2009. ‘Contact and Confidence: Revisiting the

Impact of Public Encounters with the Police’. Policing and Society 19(1): 20-46.

Carr, P.J. (2003). ‘The New parochialism: The Implications of the Beltway Case for

Arguments Concerning Informal Social Control’. American Journal of Sociology

108(6): 1249-1291.

Casey, L. (2008). Engaging Communities in Fighting Crime. London, Cabinet Office.

Choongh, S. (1997). Policing as Social Discipline. Oxford, Clarendon Press.

Hardin, R. (2006). Trust. Cambridge: Polity Press.

Hohl, K., Bradford, B. and Stanko, E. A. (2010). ‘Influencing Trust and Confidence in the

Metropolitan Police: Results from an Experiment Testing the Effect of Leaflet-drops on

Public Opinion’, British Journal of Criminology, 50, 3, 491-513.

Home Office (2008). Policing Green Paper “From the Neighbourhood to the National:

Policing Our Communities Together”. London: Home Office.

Home Office (2010). Policing in the 21st Century: Reconnecting Police and the Public. White

Paper. London: Home Office.

Hough, M. (2007). ‘Policing, New Public Management and Legitimacy in Britain’, in J. Fagan

and T. Tyler (eds) Legitimacy, Criminal Justice, and the Law. New York: Russell Sage

Foundation Press.

Hough, M., Jackson, J., Bradford, B., Myhill, A. and Quinton, P. (2010). ‘Procedural Justice,

Trust, and Institutional Legitimacy’, Policing: A Journal of Policy and Practice, doi:

10.1093/police/paq027.

Huo, Y. J. and Tyler, T. R. (2000). How different ethnic groups react to legal authority. San

Francisco: Public Policy Institute of California.

Jackson, J. and Bradford B. (2009). ‘Crime, Policing and Social Order: On the Expressive

Nature of Public Confidence in Policing’, British Journal of Sociology, 60, 3, 493-521.

Jackson, J. and Bradford B. (2010). ‘What is Trust and Confidence in the Police?’ Policing: A

Journal of Policy and Practice.

Jackson, J. and Sunshine, J. (2007). ‘Public Confidence in Policing: A Neo-Durkheimian

Perspective’, British Journal of Criminology, 47, 2, 214-233.

Lind, E. and Tyler, T. R. (1988). The Social Psychology of Procedural Justice. New York,

Plenum Press.

Loader, I. and Mulcahy, A. (2003). Policing and the Condition of England: Memory, Politics

and Culture. Oxford: Oxford University Press.

Myhill, A. and Beak, K. (2008). Public Confidence in the Police. London: National Policing

Improvement Agency.

Myhill, A. and Quinton, P. (2010). ‘Confidence, Neighbourhood Policing, and Contact:

Drawing Together the Evidence’, Policing: A Journal of Policy and Practice, doi:

10.1093/police/paq026.

Reiner, R. (2000). The Politics of the Police (3rd

edition). Oxford: Oxford University Press.

Stanko, E. A. and Bradford, B. (2009). ‘Beyond Measuring ‘How Good a Job’ Police are

Doing: The MPS Model of Confidence in Policing’, Policing: A Journal of Policy and

Practice, 4, 3, 322-330.

Sunshine, J. and Tyler, T. (2003a). ‘Moral Solidarity, Identification with the Community, and

the Importance of Procedural Justice: The Police as Prototypical Representatives of a

Groups Moral Values’. Social Psychology Quarterly 66(2): 153-165.

23

Sunshine, J. and Tyler, T. (2003b). ‘The Role of Procedural Justice and Legitimacy in Public

Support for Policing’. Law and Society Review 37(3): 513-548.

Tankebe, J. (2008). ‘Police effectiveness and police trustworthiness in Ghana: an empirical

appraisal’. Criminology and Criminal Justice 8(2): 185-202.

Tilly, C. (2005). Trust and Rule. New York, Cambridge University Press

Tyler, T.R. (1990). Why People Obey the Law. New Haven: Yale University Press.

Tyler, T. R. (2004). ‘Enhancing Police Legitimacy’. The Annals of the American Academy

593: 84-99.

Tyler, T. R. (2006). Why People Obey the Law. New Haven: Yale University Press.

Tyler, T. R. and Blader, S.R. (2000). Cooperation in Groups. Procedural Justice, Social

Identity and Behavioral Engagement. Philadelphia: Psychology Press.

Tyler, T. R. 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, 231-275.

Tyler, T. R. and Huo, Y. J . (2002). Trust in the Law: Encouraging Public Cooperation with

the Police and Courts. New York: Russell Sage Foundation.

Waddington, P. (1999). Policing Citizens: Authority and Rights. London, University College

Press.

24

Acknowledgements We are grateful to Betsy Stanley (aka Betsy Stanko) for her continuing generosity with

sharing data and discussing ideas and findings.

Appendix

The three measures of trust in police effectiveness used were: ‘Here is a list of services that the

police provide. For each one, I would like you tell me firstly how well you think the

Metropolitan Police actually carry out each of them: Tackle gun crime; Tackle drug dealing

and drug use; and Tackle dangerous driving.’

The three measures of trust in police fairness used were: ‘To what extent do you agree

with these statements about the police in this area: They would treat you with respect if you

had contact with them for any reason; The police in this area treat everyone fairly regardless of

who they are; and The police in this area are friendly and approachable.’

The three measures of trust in police engagement used were: ‘To what extent do you

agree with these statements about the police in this area: They understand the issues that affect

this community; They are dealing with the things that matter to people in this community; The

police in this area listen to the concerns of local people.’

The three measures of overall confidence in local policing used were: ‘I would now like

to talk about how well the police perform their job. Taking everything into account, how good

a job do you think the police in this area are doing?’; ‘Taking everything into account I have a

lot of confidence in the police in this area [agree/disagree]’; and ‘The police and local council

are dealing with the anti-social behaviour and crime issues that matter in this area

[agree/disagree]’.

The three measures of intention to cooperate were: ‘If the situation arose, how likely is

it that you would: Call the police to report a crime you witnessed; Report suspicious activity

near your house to the police; provide information to the police to help find a suspected

criminal’.

25

Appendix Table: Socio-demographic variation in confidence across the PSA23 and ‘Good job’

indicators Percentage predicted to give favourable rating on each indicator

1

'Good

job'2 PSA233

'Good

job'1 PSA232

Gender Housing tenure

Male 66 60 Home owner 66 60

Female 65 60 Private renter 69 61

Age Social renter 62 60

15-17 60 73 Social class

18-21 65 62 AB 70 68

22-24 66 59 C 62 59

25-34 67 59 DE 62 57

35-44 67 60 Marital status

45-54 65 59 Married 66 61

55-64 59 58 Other 64 59

65-74 67 64 Newspaper read

75-84 67 64 None 57 57

85 and over 73 60 Quality right-wing 62 61

Economic activity status Quality left-wing 60 59

Employed full time 69 64 Tabloid right-wing 63 61

Employed part-time 64 61 Tabloid left-wing 57 60

Economically inactive 60 55 Other 65 58

Retired 62 55 Contact with the police

Unemployed 60 56 None 67 61

Student 59 55 Yes and satisfied 71 67

Car access Yes and dissatisfied 31 32

Yes 63 59 Recent victim of crime

No 70 63 No 67 61

Ethnicity Yes 57 60

White British/Irish 66 58

BME 65 62

1. Values generated from binary logistic regression models containing all variables shown. When calculating each fitted probability all other

variables were held at their mean.

2. Percentage predicted to give excellent or good rating.

3. Percentage predicted to respond strongly agree or tend to agree.

Source: London Metropolitan Police Public Attitudes Survey, 2008/09

Documents

Different Things to Different People? The Meaning and Measurement of Trust and Confidence in Policing Across Diverse Social Groups in London