Intersectional inequalities in trust in the police in England
by Ferhat Tura, Steven Pickering, Martin Ejnar Hansen, and James Hunter
Published onJul 15, 2024
Intersectional inequalities in trust in the police in England
·
Abstract
This study investigates intersectional inequalities in trust in the police in England using multilevel analysis of individual heterogeneity and discriminatory accuracy (MAIHDA) for the first time. We find that those who are non-White, from lower social classes, and reside in London show lower predicted trust levels than other people. While older people show higher predicted trust levels, younger people, especially those from marginalised backgrounds, have the lowest predicted levels of trust in the police. We also find intersectional effects. While middle-aged White males from lower social classes and living outside of London have lower than-expected trust in the police, older White females from lower social classes and living outside of London have higher then-expected trust in the police. We argue that ground-level, community engagement, coupled with extensive officer training on engaging with individuals from diverse backgrounds, are key to developing higher levels of trust in the police. Key words: Trust in the police, intersectionality, inequalities, MAIHDA
Introduction
Societies depend upon trust, and one of the most important institutions within society is the police. Yet we know that trust in the police is not uniform, both across societies, and within them (Kule et al., 2019). Trust levels in the police vary depending on many factors, such as sex, ethnicity, age, social class, location and prior experience of interaction with the police as a victim, offender or ordinary citizen (Pickering et al., 2024). Reputational damage to specific police forces arising from individual cases or the behaviour/actions of individual police officers can also lead to a reduction in trust in the police. For example, following a number of high-profile incidents, the MOPAC Trust and Confidence Dashboard reveals a decline in trust in the Metropolitan police across London since the end of 2016 – with a more marked decline in multicultural and deprived parts of the Capital in recent years (MOPAC, 2024). To foster higher levels of trust in the police, and to thereby contribute to better-functioning societies, it is necessary to understand the underlying dynamics which contribute to (dis)trust in the police. If we can identify the intersecting factors (or structural systems) which determine levels of trust in the police, instead of studying individual characteristics and locational factors that decrease trust in the police separately, then we can develop better policy tools to target initiatives to increase trust in the police.
This article aims to make a substantial analytical contribution to the academic field of policing studies. It does so by employing a relatively new “gold-standard” statistical method (Merlo, 2018) developed by Evans et al. (2018; 2024): multilevel analysis of individual heterogeneity and discriminatory accuracy (MAIHDA) for the first time in this field. We use MAIHDA instead of conventional fixed effects regression models of intersectionality that include interaction parameters (e.g., Kule et al., 2019) due to their limitations explained in Evans et al. (2024). In its simplest form, MAIHDA helps us to understand how intersecting social identities contribute to various social phenomena, in our case, trust in the police. By adopting this approach, the current study (i) maps average predicted scores of trust in the police for each stratum to identify inequalities across intersectional social strata, which are constructed by combining individuals’ socio-economic and demographic characteristics (visit the Methods section), (ii) quantifies heterogeneity of trust in the police within and between the intersectional social strata, and 3-estimates interaction effects in a way that has not been investigated before. By doing so, we are better able to understand the exact dynamics behind trust in the police, and thereby develop appropriate strategies to tackle intersecting systems of oppression such as sexism, racism, and classism.
Literature review
Trust is a key concept in the social sciences and influences a wide range of factors, from the stability of democracies (Uslaner, 2002) to delivery of policing (Mason et al, 2014; Barton and Beynon, 2015). Trust in the police is specialised trust. That is, the trust that a person has based on their knowledge of and personal experience with a particular institution (see also Levi and Stoker, 2000; Citrin and Stoker, 2018). What is of key interest for this article is that specific trust can be seen as dynamic. For example, it changes with perceptions of performance and responsiveness, and while positive interactions can help build trust, failures can make it go away (see Hetherington, 1998). When the public has a belief that the policing they experience is reliable, competent and acts with integrity, they become more likely to cooperate with the police (Jackson and Bradford, 2010). However, if the public experience is negative, for instance through perceived discriminatory practices or ineffectiveness, then trust can be eroded (Samuels-Wortley, 2021; Pickering et al., 2024). If anything, a negative experience with the police is likely to exacerbate negative views of the police (Skogan, 2006; Myhill and Bradford, 2012).
The challenge for understanding the dynamics of trust in the police is that there is strong variation across intersectional groups in their levels of trust. As noted by Panditharatne et al. (2021) the connection between the public and the police is complex when considering minority communities (Tyler, 2005; Han et al., 2020) and by explicitly exploring the intersections between sex, ethnicity, social group, age and geographic location it is possible to achieve a better understanding of intersectional individuals and their trust in the police in the United Kingdom. While the concept of intersectionality originally derives from a feminist perspective focusing on minority women (Crenshaw, 1991) it has since been applied more broadly to other fields. Indeed, Hancock’s (2007) perspective is that both the theoretical and the empirical features of the topic examined are necessary. In this regard, intersectionality is now a concept which covers sex, ethnicity and socio-economic status. To achieve a better understanding of marginalised groups and their views, we need to recognise that these categories are interdependent (see also Shields, 2008). Intersectional approaches have been used to further scholarship not just in relation to policing (Kule et al., 2019; Panditharatne et al., 2021), but also with regards to trust in public officials (Reinhardt, 2019).
Racial and ethnic minorities consistently report lower levels of trust in the police. This relationship appears to be universal across cultures with ethnic minorities having lower satisfaction and lower trust in the police in the US (e.g. Skogan, 2005; Cochran and Warren, 2011; Lee and Gibbs, 2015), New Zealand (Panditharatne et al., 2021), Denmark (Kammersgaard et al., 2023) and the United Kingdom (Pickering et al., 2024). What is important to note in this regard is that the effects estimated are for the most part based on binary variables (i.e., white versus non-White), which does not cover the potential for differences within and across ethnic minority groups. Although there are a few studies from the US which have attempted to extend the empirical approach to analyse different ethnic minorities (e.g. Cochran and Warren, 2011) and found similar relationships.
However, as noted by Cao et al. (1996), the socio-economic status of a respondent might have a stronger impact than ethnicity in shaping trust in the police. This is in line with the general findings that individuals who belong to lower socio-economic groups have lower levels of trust towards government institutions (Foster and Frieden, 2017). Individuals living in wealthier neighbourhoods generally see lower crime rates and thus have less invasive contact with the police (Weitzer and Tuch, 1999). It is also known that the policing styles used in poorer neighbourhoods are more aggressive (Cobbina et al., 2017). In countries such as the United States it is noted that there are race-class subjugated communities (Soss and Weaver, 2017) where policing is not seen as a solution and could even be seen to exacerbate problems of law and order. Where people reside is also connected with age: wealthier neighbourhoods generally contain older populations than poorer neighbourhoods (Office for National Statistics, 2018). This introduces another element of intersectionality: age. Research has shown that there are different views towards policing dependent on age, but that this also interacts with other group belonging, for instance ethnicity as recently noted by Saarikkomäki et al. (2021) and Kammersgaard et al. (2023) or gender (Hurst et al., 2005).
In a recent study, Pickering et al. (2024) found that women in London had lower trust than women elsewhere in the United Kingdom. This is different to most prior studies of policing, where it is generally the case that women have higher trust in the police than males (e.g. Cao et al.,, 1996; Mbuba, 2010; Nofziger and Williams, 2005). Several reasons for this have been noted, for instance, that media reports male and female attacks differently (Mbuba, 2010) and that women, in general, are less likely to interact with the police (Cochran and Warren, 2011). However, as argued by Hurst et al. (2005) there are differing levels of trust between men and women depending on the ethnic status of respondents, highlighting the need for an intersectional approach. In the United Kingdom, there have been significant problems within the police force in London, the Metropolitan Police, with regards to women and their experience as discussed by Pickering et al. (2024). In addition, characteristics of place in London or elsewhere (e.g., concentrated disadvantage) along with other social identity characteristics of individuals might influence police officers’ action negatively (Farrell, 2024), which might then affect the levels of trust in the police. This suggests the need for an intersectional examination of trust in police in the United Kingdom to explicitly consider the geographical location of respondents.
Methods
Data sources
Data used in the analysis are based on a series of surveys conducted by YouGov. A total of 20 monthly survey waves were conducted in England between July 2022 and April 2024. Respondents were asked a battery of questions, including a question on trust in the police. Socio-economic and demographic data on respondents were also captured, such as sex, ethnicity, age, social grade and area of residency.
Outcome variable
The outcome variable of the current study is trust in the police. Respondents were asked “how much do you trust the police” using a scale of 1 to 7 where 1 means “not at all” and 7 means “completely".
Intersectional social strata dimensions
We specified social strata as a combination of five dimensions of social identities: sex, ethnicity, age, social grade, and whether living in London. Sex is a dichotomous variable with categories of male (code as 1) and female (coded as 2). Ethnicity is a categorical variable with categories of White (code as 1) and Non-White (coded as 2). Age is a categorical variable with categories of 18-24 (coded as 1), 25-49 (coded as 2), and 50 plus. For this categorisation, we use the ONS Census 2021 3c age classification (Age classifications: Census 2021 - Office for National Statistics). Social grade is a categorical variable where categories have been combined: A=Higher managerial roles, administrative or professional and B=Intermediate managerial roles, administrative or professional (coded as 1), C1=Supervisory or clerical and junior managerial roles, administrative or professional (coded as 2), C2=Skilled manual workers (coded as 3), and combination of D=Semi-skilled and unskilled manual workers and E=State pensioners, casual and lowest grade workers, unemployed with state benefits only (coded as 4). For this categorisation, we follow the census data that uses a combined four class system (AB, C1, C2, DE; Approximated Social Grade, England and Wales - Office for National Statistics (ons.gov.uk)). Finally, we have London binary variable, and those who live in London are coded as 1 and those living outside of London are coded as 2. See Table 1 for descriptive statistics.
Table 1: Descriptive statistics of the outcome and social strata variables
Outcome variable
Frequency
Mean (SD), Min-Max
Trust in police
11,609
4.0 (1.6), 1-7
Social strata variables (reference: ref)
ID position
Frequency
Percentage
Sex
Male (ref)
1
5,243
45.2
Female
2
6,366
54.8
Ethnicity
White (ref)
1
10,053
86.6
Non-White
2
1,556
13.4
Age
18-24 (ref)
1
1,049
9.0
25-49
2
4,661
40.1
50+
3
5,899
50.8
Social grade
A + B (ref)
1
3,639
31.3
C1
2
3,397
29.3
C2
3
2,070
17.8
D + E
4
2,503
21.6
London
Living in London (ref)
1
1,411
12.2
Living outside of London
2
10,198
87.8
Constructing the strata ID
The combination of sex, ethnicity, age, social grade, and whether living in London resulted in 96 unique strata, which are results of every possible permutation of the categories of the social strata variables presented in Table 2 (2 categories of Sex * 2 categories of Ethnicity * 4 categories of social grade * 2 categories of London variable = 96 strata). We also label the strata with a five-digit ID code. For example, the digit code of an individual who is female, non-White, aged 18-24, social grade of DE, and living in London stratum would be 22141 (see Table 1).
Table 2: Sample size of simulated Intersectional social strata, defined by respondent sex, race/ethnicity, age, income and living in London (n = 96)
Sample Size Per Stratum
Number of Strata
% of Strata
100 or More
20
20.8
50 or More
40
41.7
30 or More
57
59.4
20 or More
69
71.9
10 or More
89
92.7
Less than 10
7
7.3
To evaluate whether the sample size is sufficient across various intersections, we calculated the percentage of strata that have at least X respondents, where X is 10, 20, 30, 50, or 100 respondents (Evans et al., 2024). In our case, 92.7% of the strata have at least ten respondents, 71.9% have at least twenty, and 59.4% have at least thirty, suggesting that most strata have a sufficient sample size for reliable estimates.
Statistical analysis
Following the MAIHDA methodology, we use multilevel linear regression (or linear mixed effects modelling) to estimate the effect of social strata on the trust in the police, with individual observations (level 1) clustered within social strata (level 2; Evans et al., 2018; 2024). We first fit a null model (Model 1A) which specifies the outcome variable and a random intercept, allowing mean trust in the police to vary by social strata. The model can be written as:
Level 1: yij = β0j + eij
Level 2: β0j = β0 + uj
or as a combined equation: yij = β0 + uj + eij
where uj ∼ N (0, σ2u)
eij ∼ N (0, σ2e)
In this model, yij denotes the trust in the police score for individual i (i = 1,…,nj) in stratum j (j = 1, .., J). β0j is the mean trust in the police score specific to stratum j, and is decomposed into an overall mean β0 (the precision weighted grand mean or overall average value of trust in the police score) and a stratum random effect uj, which measures how different the mean trust in the police score in stratum j is from the overall mean. The uj is assumed to be normally distributed with mean of 0 and variance σ2u. The residual eij measures the deviation of the trust in the police score for individual i in stratum j from their stratum mean, and is also assumed to be normally distributed with mean of 0 and variance σ2e.
We fit Model 1A to estimate individual and stratum-level variances and particularly the Variance Partition Coefficient (VPC). The VPC is defined as the proportion of total individual variance in yij (σ2u + σ2e) that lies between strata: VPC = σ2u / (σ2u + σ2e). VPC values can range from 0 to 1 (but often re-calculated as percentages by multiplying the original value by 100) and the higher the values, the greater the practical significance (Evans et al., 2024). The VPC indicates how strata context influences understanding individual disparities in trust in the police. It measures the intra-stratum correlation or the clustering of individuals’ trust in the police within these strata. A high VPC suggests that individuals within the same stratum have very similar trust in the police scores, which differ significantly from those in other strata. If the VPC were hypothetically 100%, knowing the average trust score for a stratum would reveal each individual’s trust score within it. Conversely, a VPC of 0% would mean all strata are indistinguishable in terms of trust in the police, rendering stratum membership irrelevant for predicting individual trust. In this scenario, no General Contextual Effect (GCE) of the examined intersectional strata exists (Merlo et al., 2018).
We then fit an additive main effects model (Model 1B) to estimate interaction effects for the strata. We add social strata variables as fixed level 2 predictor variables:
yij = β0 + β1X1j + … + βpXpj + uj +eij
uj ∼ N (0, σ2u)
eij ∼ N (0, σ2e)
where x1j,…,xpj denote the p dummy variables and β1,…, βp are their associated regression coefficients of our categorical variables. The summation β0 + β1X1j + … + βpXpj gives the predicted trust in the police score for stratum j based on the additive main effects alone. Critically, Model 1B does not include any fixed interaction parameters. Instead, the stratum random effect uj captures the entirety of the interaction effect for stratum j. As in Model 1A, the uj is assumed to be normally distributed with mean of 0 and between-stratum variance σ2u, which now describes the remaining between-stratum variance after additive effects are controlled for. The residual eij has the same interpretation as in Model 1A.
In addition to VPC, we investigate the relative contribution of additive and interaction effects by calculating the Proportional Change in Variance (PCV): (σ2u (Model 1A) - σ2u (Model 1B)) / σ2u (Model 1A). The PCV captures the extent to which the between-stratum variance reduces between Models 1A and 1B. If we subtract the PCV from 1 (1 – PCV), the resulting value represents the between-stratum variance that remains unexplained after adjusting for additive effects, thus attributing it to interaction effects. Typically, the PCV is multiplied by 100 to be expressed as a percentage. A PCV significantly less than 100% suggests that interaction effects are crucial for accurately describing the observed disparities between strata (Evans et al., 2024). Data management and statistical analysis were conducted using R (version 4.2.3). Full replication data and code are available on GitHub, anonymised for peer-review: https://anonymous.4open.science/r/MAIHDA-Trust-in-the-police-1F9F/README.md
Results
Table 3 presents the results of the linear mixed effects models, which include regression coefficients, 95% confidence intervals (CIs), p-values, within- and between-stratum variances, VPC and PCV values. The general contextual effect of strata in Model 1A, measured by the VPC, was 2.47%. This indicates that 2.47% of the total variance is explained by the clustering of social strata. When the variables that were used to construct the intersectional social strata were added in Model 1B as dummy variables, the VPC decreased to 1.04%. If we compare the between strata-variance from Model 1B to Model 1A by calculating the PCV, we see that more than half of the between-stratum variance observed in Model 1A (59%) is accounted for by the additive main effects. In other words, after accounting for the additive main effects, 41% (100% - 59%) of the original between-stratum variance is still unexplained by the additive main effects, which is an indication of interaction effects (Evans et al., 2024). In short, although the majority of between-strata differences were explained by the additive main effects, the remaining 41% variance unexplained by the main effects shows us that inequalities between the strata exist, and there are significant interactions, which we will revisit later in the paper. Next, we focus on the main effects.
Looking at the coefficients from Model 1B, we can investigate the main effects that explain the trust in the police as 59% of the between-stratum variance is accounted for by these additive main effects. The average non-White stratum has a trust in the police score that is 0.23 lower than the average White stratum, holding all other variables constant. Similarly, compared to social grade AB strata, social grade DE strata have lower trust in the police score by 0.33. In contrast, compared to aged 18-24 strata, aged 50+ strata have greater trust in the police score by 0.25. However, these additive patterns can obscure important findings that become clearer when examining intersections, which involves looking at the average predicted trust in the police scores for each stratum and breaking it down into the portions attributable to additive main effects and interaction effects (Evans et al., 2024).
Table 3: Parameter estimates for linear mixed effects models of trust in police
Model 1A
Model 1B
Social strata variables
Estimates
95% CI
Estimates
95% CI
(Intercept)
3.86 ***
3.79 – 3.93
3.86 ***
3.68 – 4.04
Female
0.02
-0.09 – 0.13
Non-White
-0.23 ***
-0.35 – -0.12
Age: 25-49
0.06
-0.09 – 0.20
Age: 50 +
0.25 **
0.10 – 0.40
Social grade C1
-0.07
-0.21 – 0.07
Social grade C2
-0.03
-0.19 – 0.12
Social grade DE
-0.33 ***
-0.48 – -0.18
Living outside of London
0.11
-0.01 – 0.23
Random Effects: Variances
Individual
2.40
2.40
Stratum
0.06
0.03
VPC
2.47%
1.04%
PCV
58.69%
AIC
43201.97
43199.97
BIC
43224.05
43280.92
N
96
96
Observations
11,609
11,609
* p<0.05 ** p<0.01 *** p<0.001
Figure 1 plots (ranked) predicted stratum random effects (i.e., the uj from Model 1B), which are representations of interaction effects (Evans et al., 2024). In other words, we can observe patterns of higher-then-expected and lower-then-expected predicted trust in the police as the strata level residuals represent the difference between the observed mean trust in the police and expected mean trust in the police for each stratum based on the additive main effects. If the 95% CIs around the estimate of uj for a stratum does not encompass 0 (such as strata 11242 and 21342), it’s a rough indication of statistically significant interaction effects (Evans et al., 2024).
Table 4 also presents the 10 strata with the lowest and highest random effects and interested readers can further identify the strata that have lower-then-expected or higher-then-expected trust in the police scores in Appendix Table 1. In Figure 1 and Table 4, for example, we can see that stratum 21342 (female, White, aged 50+, social grade DE, and living outside of London) has higher-then-expected trust in the police, while stratum 11242 (male, White, aged 25-49, social grade DE, and living outside of London) and stratum 11232 (male, White, aged, 25-49, social grade C2, and living outside of London) have lower-then-expected trust in the police. Stratum 11242 has also one of the lowest predicted trust in the police (3.414, ranked 7th, see Appendix Table 1). That is, the intersecting effects of systems of oppression like ageism, classism and where you live (which might be an indication of economic inequality, limited access to services and resources, and over-policing) on trust in the police compound in a way that exceeds the effects of these systems, separately.
Finally, Figure 2 visually examines the trust in the police by plotting the (ranked) predicted trust in the police scores for all strata from Model 1B. It illustrates the range, spread, and pattern of inequalities between strata (see Appendix Table 1 for predicted scores and 95% CIs for all 96 strata). In Table 5, we also present the 10 strata with the highest and lowest predicted trust in the police scores and their respective 95% CIs. The stratum with the lowest predicted trust in the police score is 22141: Female, non-White, aged 18-24, social grade DE and living in London. The stratum with the highest predicted trust in the police is 21332: Female, White, aged 50+, social grade C2, and living outside of London. Focusing on the 10 strata with the lowest scores, it can be said those who are non-White (7 out of 10), young (6 out of 10), from social grade DE (10 out of 10) and living in London (6 out of 10) have the lowest predicted trust in the police. These results indicate that while the intersection of racism, agism and classism may be an important driver of inequalities in the trust in the police, but for non-White individuals from social grade DE, racism may be the most important factor.
Figure 1: Predicted stratum random effects by stratum
A graph showing a line of lines Description automatically generated with medium confidence
Figure 2: Predicted trust in the police scores by stratum
A graph showing a line of lines Description automatically generated with medium confidence
Table 4: Inspection of 10 lowest and highest ranked interaction effects
Model 1B
Rank
Stratum
n
Mean
95% CI
10 Lowest
Lower
Upper
1
11242*
276
-0.294
-0.500
-0.088
2
11232*
235
-0.28
-0.482
-0.078
3
22211
63
-0.128
-0.389
0.133
4
21122
148
-0.121
-0.317
0.075
5
22342
28
-0.118
-0.385
0.149
6
12232
48
-0.116
-0.379
0.147
7
21221
54
-0.109
-0.366
0.148
8
21112
77
-0.107
-0.368
0.154
9
12312
37
-0.105
-0.375
0.165
10
22231
18
-0.101
-0.391
0.189
10 Highest
1
21342*
753
0.222
0.050
0.394
2
11331
30
0.163
-0.107
0.433
3
12211
47
0.163
-0.100
0.426
4
12222
68
0.156
-0.085
0.397
5
21222
594
0.148
-0.021
0.317
6
21142
63
0.145
-0.118
0.408
7
22242
52
0.139
-0.106
0.384
8
21341
56
0.132
-0.115
0.379
9
12131
14
0.124
-0.180
0.428
10
22212
107
0.12
-0.111
0.351
Sex:1=Male, 2=Female; Ethnicity :1=White, 2=Non-White; Age:1=18-24, 2=25-49, 3=50+; Social grade:1=AB, 2=C1, 3=C2, 4=DE; London:1=Living in London, 2=Living outside of London.
* 95% CI around the estimate of uj (random effect) for this stratum does not encompass 0
Table 5: Inspection of 10 lowest and highest ranked strata for predicted trust in police
Model 1B
Rank
Stratum
n
Mean
95% CI
10 Lowest
Lower
Upper
1
22141
7
3.245
2.871
3.593
2
12241
16
3.292
2.944
3.630
3
12141
12
3.342
2.964
3.677
4
22241
34
3.361
3.044
3.690
5
22142
31
3.375
3.059
3.691
6
12142
15
3.393
3.050
3.723
7
11242
276
3.414
3.217
3.622
8
11141
4
3.477
3.111
3.818
9
12242
42
3.489
3.194
3.770
10
21141
4
3.531
3.165
3.906
10 Highest
1
21332
528
4.274
4.073
4.444
2
21322
713
4.27
4.081
4.443
3
11331
30
4.247
3.931
4.573
4
21312
735
4.185
4.014
4.361
5
21331
30
4.182
3.869
4.521
6
21212
598
4.134
3.965
4.322
7
21222
594
4.128
3.95
4.301
8
21342
753
4.127
3.942
4.315
9
11321
41
4.119
3.833
4.428
10
11311
91
4.114
3.838
4.393
Sex:1=Male, 2=Female; Ethnicity :1=White, 2=Non-White; Age:1=18-24, 2=25-49, 3=50+; Social grade:1=AB, 2=C1, 3=C2, 4=DE; London:1=Living in London, 2=Living outside of London.
Discussion and conclusions
In this section, we reflect on the findings and their policy implications for enhancing trust in the police. In this study, we investigated disparities in the trust in the police between intersectional social identities using a novel statistical approach, the MAIHDA framework (Evans et al., 2018; 2024). This approach has several advantages over the traditional regression models with interaction parameters and for a greater understanding of the methodology interested readers can read Evans et al. (2018; 2024). This approach allowed us to test intersectional effects of all social identities indicating both privilege and disadvantage. To our knowledge, this is the first study to estimate the general contextual effect of social identity on the trust in the police, as most studies which use multilevel models to explore contextual effects consider the effects of geographic clusters, like neighbourhood (Hawdon, 2008; Yesberg et al., 2023) or use interaction parameters (Kule et al., 2019) which have limitations (Evans et al., 2024). The theoretical implication of the study is that using sex, ethnicity, age, social class, and location of residency as separate predictor variables masks the complexity of the oppressing systems, such as sexism, racism, and classism, affecting the trust in the police.
Therefore, by considering intersectional social strata as a level 2 variable in the analysis, we were able to apply intersectional theory and estimate the effect of interlocking systems, rather than individual characteristics, that create inequalities. In particular, we were able to investigate the unexplained between strata variance by additive effects (i.e., random/interaction effects). We found that 41% of the strata-level variation in trust in the police remains unexplained after the main effects are controlled for. This finding supports the intersectional hypothesis that intersection of sex, ethnicity, age, social class, and area of residence as proxies of privilege and disadvantage plays a role in the patterns of inequalities observed in trust in the police.
Visualisation of the strata-level random effects from Model 1B (see Figure 1) enabled us to investigate the variation within intersectional social strata and identify patterns in the intersectional results. For example, while there is no difference between male and females in terms of trust in the police according to the main/fixed effect results presented in Table 3, Figure 1 and Table 4 suggest for example that stratum 11242 (male, White, aged 25-49, social grade DE, and living outside of London) and stratum 11232 (male, White, aged, 25-49, social grade C2, and living outside of London) have lower-then-expected trust in the police. This shows us that these factors/systems are intersecting in a way that negatively influences trust in the police.
Our findings cannot be directly compared with previous studies given the innovative application of the MAIHDA methodology within this paper. However, our analysis confirms that individual factors such as being from an ethnic minority and belonging to a lower social class is associated with reduced trust in the police as in previous studies (Pickering et al., 2024; Foster and Frieden, 2017). This suggests that while intersectionality of multiple dimensions of social identity is critical to understanding inequalities in trust in the police, racism and classism are perhaps the most influential systems of oppression influencing trust in the police negatively.
Berg and Mann (2023) suggest that the intersection of various social identities and characteristics presents significant challenges for policing at various levels. For example, micro-level interactions between police officers and individuals with intersecting social identities can have profound positive or negative effects. The importance of this is emphasised within Objective Four of Pillar One of the Strategic Policing Partnership Board’s Policing Vision 2030 (College of Policing, 2023a). At the meso-level, institutional factors such as police culture, resources, specialised training, and the presence of dedicated teams or programmes can either alleviate or exacerbate negative interactions with individuals possessing intersecting identities. The need for more effective police leadership that can bring about changes in organisational culture and practices to address lack of trust in the police has also been recently prioritised by the College of Policing (2023b). Plans such as the Police Race Action Plan (College of Policing, 2022) which has been developed jointly by the National Police Chiefs’ Council (NPCC) and the College of Policing should be embraced by police leaders as a recent report by Independent Scrutiny & Oversight Board (2024) found that the plan is not applied as it’s envisaged. Finally, macro-level considerations might involve broader structural influences and power dynamics under which the police operate. These influences, shaped by both historical and contemporary contexts such as legislation, policies, political climates, and public expectations, can disproportionately affect certain groups (Berg and Mann, 2023).
Therefore, engaging with communities directly via for example establishment of community outreach programmes, community advisory boards can help identify specific issues affecting trust and better manage micro-level interactions. Programmes could include community policing initiatives (Hawdon, 2008) in which the police officers act as protectors and collaborators, and feedback mechanisms to build trust at the ground level. Police officers should undergo extensive training that would shift a police force model to a police service model (Tyler, 2017), such as unconscious bias (Dario et al., 2019) or procedural justice principles which should result in a cultural change in behaviour (Ghezzi et al., 2021; Van Craen and Skogan, 2017) or awareness, recognition, and sensitivity training (Pickles, 2020; Hutson et al., 2022) to understand and respect the diverse backgrounds of the communities they serve. Training should address the specific intersectional characteristics identified in the current study (e.g., handling interactions with young non-White individuals or White males from lower social grade differently based on their unique challenges). The Independent Office for Police Conduct (IOPC) describes this as “cultural competence” and it means “the ability of individuals and systems to work or respond effectively across cultures, in a way that acknowledges and respects the culture of the person being served” (Williams, 2001 cited in IOPC, 2022, p. 3). IOPC particularly acknowledges the importance of intersectionality in their Equality, Diversity and Inclusion strategy for 2022-25 (IOPC, 2022). They mention that intersectionality is a framework they use, particularly when focusing on marginalised groups and their confidence in policing. However, it should also be noted that recent research (Burnett et al., 2019) found compassion fatigue among police officers and therefore these kinds of trainings might not be effective if not well planned and sourced as the police themselves are vulnerable to stress. Another important barrier might be the police culture that exhibits machismo and/or exclusionary heteronormative values (Pickles, 2020) like in the Met (Casey, 2023).
Another effective policy could be rolling out specialist policing teams and initiatives, as part of which liaison officers are established to support their colleagues including marginalised ones and citizens with intersecting identities, which would eventually help improving police-community relations (Pickles, 2020). However, concerns have been raised about the effectiveness of specialist policing in general due to marginalised individuals not engaging with liaison officers (Fileborn, 2019) and in rural areas particularly due to limited resources (Rural Services Network, 2023).
Scholars also argued that the police workforce should be more diverse (Pickles, 2020; Black and Cari, 2010) and a recent report by Police Foundation (Hales, 2020) shows that diversity among police workforce (particularly PCSOs and police staff) has been increasing but varies across the police forces. Therefore, it is suggested that discussions of police ethnic diversity at a national level are arguably rather meaningless (Hales, 2020). Instead, police force workforces should represent their communities and recruitment should be carried out applying an intersectional lens.
The coexistence of the importance of individual factors and interaction effects highlights the complexity of trust in the police and managing it. Effective policies need to address both or might constitute micro-, meso-, and macro-level engagement with and beyond the police (Berg and Mann, 2023) as Hankivksy and Cormier (2011) argues that “one size fits all” policies do not work. By combining these approaches, policies can be more comprehensive and effective in building trust in the police across different segments of the population.
While enhancing trust in the police faces several embedded challenges which will take time to overcome, further enhancement of the existing evidence base around policing, trust and community engagement is also required. More research is required to understand the importance of local contextual factors in shaping the intersectional nature of trust in the police in relation to specific incidents, neighbourhood policing practices, and the effectiveness of community engagement initiatives. The role of community stakeholders, rather than the direct involvement of police officers, in driving community engagement and trust-enhancing activities also merits further academic attention. Future quantitative research using larger sample sizes can also further unpick the complexities of the interaction between gender, ethnicity, social class and location in shaping disparate trust in the police outcomes across different social groups and neighbourhoods through the inclusion of more refined categories in the construction of the strata contained within empirical analyses.
The findings presented in this study have confirmed the importance of understanding the interplay between different individual characteristics in shaping trust in the police. Implementing initiatives designed to enhance community relations and trust in the police that only target individuals or communities solely based on gender, ethnicity, social class or location are likely to produce only limited improvements. Holistic approaches which acknowledge the interaction of these factors, and which develop bespoke programmes of action that address the interplay of these characteristics in shaping trust in the police are likely to prove more successful.
References
Barton, H., & Beynon, M. J. (2015). Do the citizens of Europe trust their police? International Journal of Emergency Services, 4(1), 65–85.
Black, P. J., & Kari, C. J. (2010). Policing diverse communities: Do gender and minority status make a difference?. Journal of Ethnicity in Criminal Justice, 8(3), 216-229.
Burnett, M. E., Sheard, I., & St Clair-Thompson, H. (2020). The prevalence of compassion fatigue, compassion satisfaction and perceived stress, and their relationships with mental toughness, individual differences and number of self-care actions in a UK police force. Police Practice & Research, 21(4), 383–400.
Cao, L., Frank, J., & Cullen, F. T. (1996). Race, community context and confidence in the police. American Journal of Police, 15(1), 3–22.
Citrin, J., & Stoker, L. (2018). Political trust in a cynical age. Annual Review of Political Science, 21(1), 49–70.
Cobbina, J. E., Conteh, M., & Emrich, C. (2017). Race, gender, and responses to the police among Ferguson residents and protesters. Race & Justice, 9(3), 2–28.
Cochran, J. C., & Warren, P. Y. (2011). Racial, ethnic, and gender differences in perceptions of the police: The salience of officer race within the context of racial profiling. Journal of Contemporary Criminal Justice, 28(2), 206–227.
College of Policing (2022). Police Race Action Plan Improving policing for Black people. Retrieved from https://www.npcc.police.uk/SysSiteAssets/media/downloads/our-work/race-action-plan/police-race-action-plan-improving-policing-for-black-people.pdf
Crenshaw, K. (1991). Mapping the margins: Intersectionality, identity politics, and violence against women of color. Stanford Law Review, 43(6), 1241–1299.
Dario, L. M., Fradella, H. F., Verhagen, M., & Parry, M. M. (2019). Assessing LGBT people’s perceptions of police legitimacy. Journal of Homosexuality, 67(7), 885–915.
Evans, C. R., Leckie, G., Subramanian, S. V., Bell, A., & Merlo, J. (2024). A tutorial for conducting intersectional multilevel analysis of individual heterogeneity and discriminatory accuracy (MAIHDA). SSM-population Health, 26, 101664.
Evans, C. R., Williams, D. R., Onnela, J. P., & Subramanian, S. V. (2018). A multilevel approach to modeling health inequalities at the intersection of multiple social identities. Social Science & Medicine, 203, 64-73.
Farrell, C. (2024). Policing gender, race, and place: A multi-level assessment of stop and frisks. Race & Justice, 14(3), 290–312.
Fileborn, B. (2019). Policing youth and queerness: the experiences and perceptions of young LGBTQ+ people from regional Victoria. Current Issues in Criminal Justice, 31(3), 433-451.
Foster, C., & Frieden, J. (2017). Crisis of trust: Socio-economic determinants of Europeans’ confidence in government. European Union Politics, 18(4), 511–535.
Ghezzi, E. L., Funk, J. A., & Houmanfar, R. A. (2022). Restructuring law enforcement agencies to support prosocial values: A behavior-scientific model for addressing police brutality. Behavior Analysis in Practice, 15(4), 1184-1192.
Han, S., Hwang, E., Nobles, M. R., Basham, S. L., & Piquero, A. R. (2020). Immigrants’ confidence in the police in 34 countries: A multilevel analysis. Police Quarterly, 23(1), 106–137.
Hancock, A. M. (2007). When multiplication doesn’t equal quick addition: Examining intersectionality as a research paradigm. Perspectives on Politics, 5(1), 63–79.
Hankivsky, O., & Cormier, R. (2019). Intersectionality and public policy: Some lessons from existing models. In O. Hankivsky, & J.S. Jordan-Zachery (Eds), The Palgrave handbook of intersectionality in public policy (pp.69-93). Basingstoke: Palgrave Macmillan.
Hawdon, J. (2008). Legitimacy, trust, social capital, and policing styles: A theoretical statement. Police Quarterly, 11(2), 182–201.
Hetherington, M. J. (1998). The political relevance of political trust. American Political Science Review, 92(4), 791–808.
Hurst, Y., McDermott, M. J., & Thomas, D. L. (2005). The attitudes of girls toward the police: Differences by race. Policing: An International Journal of Police Strategies & Management, 28(4), 578–593.
Hutson, T. M., McGhee Hassrick, E., Fernandes, S., Walton, J., Bouvier-Weinberg, K., Radcliffe, A., & Allen-Handy, A. (2022). “I'm just different–that's all–I'm so sorry…”: Black men, ASD and the urgent need for DisCrit Theory in police encounters. Policing: An International Journal, 45(3), 524-537.
Jackson, J., & Bradford, B. (2010). What is trust and confidence in the police? Policing: A Journal of Policy & Practice, 4(3), 241–258.
Kammersgaard, T., Søgaard, T. F., Haller, M. B., Kolind, T., & Hunt, G. (2023). Community policing in Danish “ghetto” areas: Trust and distrust between the police and ethnic minority youth. Criminology & Criminal justice, 23(1), 98–116.
Kule, A., Bumphus, V. W., & Iles, G. (2019). Intersectionality of race, class, and gender in predicting police satisfaction. Journal of Ethnicity in Criminal Justice, 17(4), 321-338.
Lee, J., & Gibbs, J. (2015). Race and attitudes toward police: The mediating effect of social distance. Policing: An International Journal of Police Strategies & Management, 38(2), 314–332.
Levi, M., & Stoker, L. (2000). Political trust and trustworthiness. Annual Review of Political Science, 3(1), 475–507.
Mason, D., Hillenbrand, C., & Money, K. (2014). Are informed citizens more trusting? Transparency of performance data and trust towards a British police force. Journal of Business Ethics, 122, 321-341.
Mbuba, J. M. (2010). Attitudes toward the police: The significance of race and other factors among college students. Journal of Ethnicity in Criminal Justice, 8(3), 201–215.
Merlo, J. (2018). Multilevel analysis of individual heterogeneity and discriminatory accuracy (MAIHDA) within an intersectional framework. Social Science & Medicine, 203, 74-80.
Myhill, A., & Bradford, B. (2012). Can police enhance public confidence by improving quality of service? Results from two surveys in England and Wales. Policing & Society, 22(4), 397–425.
Nofziger, S., & Williams, L. (2005). Perceptions of police and safety in a small town. Police Quarterly, 8(2), 248–270.
Panditharatne, S., Chant, L., Sibley, C. G., & Osborne, D. (2021). At the intersection of disadvantage: Socioeconomic status heightens ethnic group differences in trust in the police. Race & Justice, 11(2), 160–182.
Pickering, S., Dorussen, H., Hansen, M. E., Reifler, J., Scotto, T., Sunahara, Y., & Yen, D. (2024). London, you have a problem with women: trust towards the police in England. Policing & Society, 1–16.
Pickles, J. (2020). Policing hate and bridging communities: A qualitative evaluation of relations between LGBT+ people and the police within the North East of England. Policing & Society, 30(7), 741–759.
Reinhardt, G. Y. (2019). The intersectionality of disasters’ effects on trust in public officials. Social Science Quarterly, 100(7), 2567–2580.
Saarikkomäki, E., Birk Haller, M., Solhjell, R., Alvesalo-Kuusi, A., Kolind, T., Hunt, G., & Burcar Alm, V. (2021). Suspected or protected? Perceptions of procedural justice in ethnic minority youth's descriptions of police relations. Policing & Society, 31(4), 386–401.
Samuels-Wortley, K. (2021). To serve and protect whom? Using composite counter-storytelling to explore black and indigenous youth experiences and perceptions of the police in Canada. Crime & Delinquency, 67(8), 1137–1164.
Shields, S. (2008). Gender: An intersectionality perspective. Sex Roles, 59(2008), 301–311.
Skogan, W. G. (2005). Citizen satisfaction with police encounters. Police Quarterly, 8(3), 298–321.
Skogan, W. G. (2006). Asymmetry in the impact of encounters with police. Policing & Society, 16(2), 99–126.
Soss, J., & Weaver, V. (2017). Police are our government: Politics, political science, and the policing of race–class subjugated communities. Annual Review of Political Science, 20(1), 565–591.
Tyler, T. (2017). Procedural justice and policing: A rush to judgment? Annual Review of Law & Social Science, 13(1), 29–53.
Tyler, T. R. (2005). Policing in black and white: Ethnic group differences in trust and confidence in the police. Police Quarterly, 8(3), 322–342.
Uslaner, E.M. (2002). The moral foundation of trust. New York: Cambridge University Press.
Van Craen, M., & Skogan, W. G. (2017). Achieving fairness in policing: The link between internal and external procedural justice. Police Quarterly, 20(1), 3–23.
Weitzer, R., & Tuch, S. A. (1999). Race, class, and perceptions of discrimination by the police. Crime & Delinquency, 45(4), 494–507.
Yesberg, J., Brunton-Smith, I., & Bradford, B. (2023). Police visibility, trust in police fairness, and collective efficacy: A multilevel Structural Equation Model. European Journal of Criminology, 20(2), 712–737.
I am happy I read about Lost Recovery Masters and l didn’t hesitate to contact them. Lost Recovery Masters were able to recover my $94,000 worth of BTC as of then that I stupidly paid to crypto investment scammers I met on Facebook. They are now my family hacker and I’ll always recommend Lost Recovery Masters to everyone in need of help.
If not for Baba Powers what would my life turn out to be? I want you all to please contact Baba Powers now to get the powerful black mirror from him. I want you all to also BELIEVE AND TRUST HIM because whatever he tells you is the TRUTH and 100% guaranteed. The black mirror makes it happen, attracts abundance. I bought the black mirror from Baba Powers and now, I am super rich and successful with the help of the black mirror. When I first saw the testimonies of Baba Powers of the black mirror, I thought it was a joke but I contacted him to be sure for myself and to my greatest surprise, the black mirror is real. The black mirror is powerful. Check him out on his website to see plenty of amazing testimonies from people about him. These are some of the people that he has helped. Here is his website; Babablackmirrorsofpowers.blogspot.com and here is his email; [email protected] I really can't thank you enough Baba Powers. God bless you, thank you