London, you have a problem with women: trust towards the police in England

ABSTRACT

Following a series of high-profile incidents of violence against women by serving London Metropolitan Police Officers, questions of standards and the public’s confidence in policing are in the spotlight. Over a fifteen-month period between July 2022 and September 2023 using monthly surveys of representative English samples, this study confirms that women, in general, are more trusting in the police than men. This, however, does not hold true in London. Out of nine regions in England, London is the only region where women’s overall trust in the police is lower than men. Lower levels of trust in the police among women in London hold when controls for age, income, political environment and crime levels are considered. In line with existing literature that considers women being more sensitive to cues about trustworthiness, the concerning incidents of sexual violence by police officers against women are likely to further erode trust in police in the capital, which already ranks last among England’s nine regions in citizen trust of the police.

Statement on ethics approval:

Approval was granted for this study to be carried out by the College of Business, Arts and Social Science Research Ethics Committee, Brunel University London. Approval reference: 35290-LR-Jan/2022-37313-1.

KEYWORDS:

• Police
• trust
• gender
• London

Introduction

In early 2021, Sarah Everard disappeared from a walk home after a night out. Soon after her body was discovered, the perpetrator of her rape and murder was found to be a serving police officer in the Metropolitan police. The trial encouraged another woman who was tortured and raped by another serving Met officer to come forward, leading to his conviction. At the time of writing, over 1000 Metropolitan police officers are suspended or on restricted duties for allegations including violence against women and girls, sexual violence and domestic violence (BBC 2023).

According to the Peelian principles upon which the Metropolitan police force is based, policing depends on consent and the avoidance of physical force. Yet at a vigil held for Sarah Everard on Clapham Common, Metropolitan police officers handcuffed and arrested four women for alleged violations of COVID-19 restrictions. This heavy-handed reaction was one incident among many that brought into question the very foundation on which policing in London was built: trust.

Trust in key institutions such as the police is critical. Levi and Stoker (2000) argue that trust is relational with an individual making themselves vulnerable to an institution with the capability to do them harm. Trust overall is not unconditional and can be divided into different categories, most notably generalised and specific trust (Freitag and Bühlmann 2009, Freitag and Traunmüller 2009). Generalised trust is strongly related to desirable outcomes, both in terms of democracy (e.g. Knack 2002, Tavits 2006, Sønderskov 2008, Zmerli and Newton 2008) and economy (e.g. Knack and Keefer 1997, Uslaner 2002). Specific trust is based on a judgement of the situation, is based on previous experience and can vary depending on outside events (Levi and Stoker 2000). In this article, we ask a simple question: what factors influence respondents’ specific trust in policing in England and in London specifically?

The role of trust in policing in London runs deep. London’s Metropolitan Police force was created by an Act of Parliament in 1829, and the Home Office states that from that year on, every new police officer was issued with a list of nine ‘General Instructions’ in which trust play an integral part.¹ Inherent in these are the ideas: that the power of the police depends on public approval; that the police need the willing cooperation of the public; that cooperation diminishes proportionately with the use of force; and that the police should offer friendship equally to all. The rape and murder of Sarah Everard, and the subsequent arrest of women at a vigil in her memory profoundly challenge these key precepts upon which policing in London is built.

In the wake of Sarah Everard's murder, a review into the standards and behaviour of the Metropolitan Police Service was commissioned, conducted by Baroness Casey. It found that ‘far too many Londoners have now lost faith in policing’ (Casey 2023, p. 8). The review went on to find that public trust in the Met fell from 89% in 2016 to a low of 66% in March 2022 (Casey 2023, p. 10). The Casey review highlights that there are systematic and fundamental problems in how policing in the capital of the United Kingdom is run and how its residents and notably its women has been left behind by an organisation that lacks accountability, transparency and above all, trust (Casey 2023). Using a survey of multiple waves of over 8000 combined respondents conducted from July 2022 to September 2023, we find that in the parts of England outside of London, only 44% of women trust the police. However, when we look at London itself, this figure drops to 34%. In other words, Baroness Casey's scathing assessment of the Metropolitan Police Force probably overstates levels of trust in the police. Across England, the region with lowest levels of trust in the police is London. Furthermore, while women generally trust the police more than men do, we find that this is the opposite in London. The critiques raised against English police forces, and especially the Metropolitan police, are linked to officers breaking the law and engaging in heinous crimes which have become occurrences that too often reach the front pages of the newspapers. While negative press for parts of the police force is nothing new, the scope of the problem has meant that there is a scepticism, if not outright distrust, towards the police and policing among groups in British society today. This creates a general problem for a public institution depending on trust to perform duties central to the functioning of society equally across all sub-groups. However, given that policing performance is to some extent dependent on local factors, not least through the police and crime commissioners and mayors having an impact on policing in their constabularies, there is a puzzle to solve, namely, to what extent there is a difference in trust towards the police depending on geography and socio-economic factors.

Our findings on specific trust in the police suggest that respondents who are older have higher levels of trust in the police, while ethnic minority respondents have significantly lower levels of trust. In general, the main effect of being a woman has a positive relationship with trust in the police. But this relationship breaks down with policing in London, where trust in the police among women is lower than anywhere else in the country and is significantly lower than among men. Our findings are robust when controlling for various factors, such as the burglary and violent crime rates in the constituency and the level of support for the Conservative party. When controlling for the impact of generalised trust, our findings remain the same. Overall, our results suggest that there is significant work ahead for police forces generally in England with regards to restoring trust in the police among ethnic minority citizens and in London especially also among women. These results are evidence that this is a serious challenge to the basic principles underpinning the police. For such a central societal institution, this may necessitate stronger focus by the government.

Trust

Trust is a central concept within the social sciences and many central social science theories depend on the concept of trust (Hetherington 1998, Newton 2001). Trust deals with questions concerning the confidence that citizens have in government and political institutions, and their ability to execute their duties effectively (Levi and Stoker 2000). Trust has been shown to influence a wide range of outcomes, from the level of citizens’ engagement in politics (Galston 2001), cooperation with delivery agencies such as the police (Barton and Beynon 2015), to the broader stability of democratic systems (Uslaner 2002). Trust can be classified into two categories: generalised and specific (Newton and Zmerli 2011).

Generalised trust refers to the broad societal level of the trustworthiness of political institutions. It represents an overarching trust in the system of governance and a presumption of good intentions in the absence of specific information about an institution (Newton and Zmerli 2011). This type of trust is often influenced by cultural norms, social experiences, or societal trends, and serves as a social glue that binds citizens to their political system, encouraging participation and cooperation (Rothstein and Stolle 2008; but see also Van der Meer and Ouattara (2019) on the difficulties of making such cross-national comparisons). Generalised trust in political institutions can be essential in forming a stable political environment where citizens are willing to accept decisions made by authorities, even when these decisions may not align with their personal interests. While generalised trust is often used on its own (e.g. Nannestad 2008), it also has different implications (Freitag and Traunmüller 2009) which means considering generalised trust is necessary, even when focusing on specific trust.

Specific trust is grounded in personal experiences, judgements and knowledge about a particular political entity, such as a politician, political party, or a specific government institution (Levi and Stoker 2000, Citrin and Stoker 2018). Specific trust is based on perceptions of competence, integrity, fairness and the degree to which it aligns with an individual's values or expectations. Specific trust is often dynamic, changing with evolving perceptions of performance, credibility and responsiveness (see e.g. Keele 2007). Repeated positive interactions tend to foster trust, while perceived failures or scandals can erode it (see also Hetherington 1998). There is little doubt that both generalised and specific trust in political institutions play central roles in democratic societies. Without trust the relationship between the citizenry and political institutions will break down: losing trust not only influences political behaviour on the one hand but also leads to challenges towards central institutions in a democracy, such as the police, presenting an overall crisis for the health of democracy. In the next section we will consider the relationship between trust and the police.

Specific trust in the police

Specific trust in the police is a fundamental element in the functioning of a democratic society (see Mason et al. 2014, Barton and Beynon 2015). It refers to citizens’ belief in the reliability, integrity and competence of the police force, underpinning their willingness to obey laws and cooperate with police officers (Jackson and Bradford 2010). Specific trust in the police is often built on direct and indirect experiences with the law enforcement system. Positive experiences include police responding promptly to calls for help, treating individuals fairly, and demonstrating competency in crime-solving, shown to foster trust (see also Tyler and Huo 2002). In contrast, negative experiences, such as perceived misconduct, discriminatory practices, or ineffectiveness in ensuring safety, erode trust (e.g. Samuels-Wortley 2021). In fact, research has shown an asymmetrical relationship between trust in the police and personal experience, leading to poor experience exacerbating negative evaluations, while professional service is not given any credit (see also Skogan 2006, Myhill and Bradford 2012).

Trust in the police is also influenced by factors such as representation and community relations. Police forces that reflect the diversity of the communities they serve and engage with tend to earn higher levels of trust (Jackson and Bradford 2010). Initiatives such as community policing, where law enforcement officers build relationships with community members and work collaboratively to solve problems, can foster specific trust (e.g. Torres 2017, Peyton et al. 2019).

However, it is important to note that trust in the police has been shown to vary significantly across different demographic groups. For instance, in many societies, racial and ethnic minorities often report lower levels of trust in the police, reflecting concerns about racial profiling and discriminatory practices (see e.g. Kammersgaard et al. 2023). Similarly, individuals in lower socio-economic strata tend to have lower trust levels possibly due to perceived biases or unequal treatment, both in terms of police and other government institutions (Foster and Frieden 2017). High-profile incidents of police misconduct or systemic issues brought to light can lead to diminished trust, even among individuals who have not personally had negative encounters with the police (e.g. Barnes et al. 2018). Specific trust in the police is vital for maintaining social order and promoting cooperation between citizens and law enforcement. Building and sustaining this trust requires ongoing efforts to ensure fair, competent and accountable policing practices that respect the rights and dignity of all citizens. Understanding the factors that shape trust in the police can help guide these efforts, fostering a law enforcement system that enjoys the confidence of the community it serves. This means that five factors must be considered to achieve a fuller picture of trust: gender, ethnicity, socio-economic status, partisanship and community basis.

Trust and gender

Gender differences in trust have been a subject of interest for years, although the findings vary significantly across cultures and societies, reflecting the complex interplay between innate traits and social conditioning (Newton and Zmerli 2011, Bunting et al. 2021). Men and women may exhibit different patterns of trust due to a combination of social, biological and environmental factors. Socialisation processes often impart different expectations and roles to different genders, which can influence how they perceive and engage in trusting relationships (Coffé and Bolzendahl 2010). Some studies suggest that women tend to exhibit higher levels of trust and are more likely to extend trust to others than men, potentially due to socialisation that encourages empathy, cooperation and relationship-building (see e.g. Etang et al. 2011, Kamas and Preston 2021). Such higher levels of trust seem to be particularly pronounced in the context of interpersonal relationships. Yet this does not necessarily mean women are indiscriminately trusting, as they can be more sensitive to cues of trustworthiness, thus making nuanced decisions about when (or when not) to trust (Neu 2015).

Men, on the other hand, are often socialised to be more competitive and self-reliant, which can translate into more calculated or cautious trust decisions, relying more heavily on evidence of competence or reliability before extending trust, which in turn might lead to lower levels of generalised trust, but higher levels of specific trust once credibility has been established (see also Qiu et al. 2022). In the context of trust in political institutions, gender differences can be influenced by perceived representation and fairness. For instance, in societies where women’s political representation is low, women can exhibit lower levels of political trust compared to men (e.g. Hinojosa et al. 2017).

Trust and ethnicity

Research has been carried out on both the relationship between ethnic diversity and trust and an individual’s ethnicity and their trust levels (see Fennema and Tillie 1999, Uslaner and Conley 2003, Stolle et al. 2008, Hooghe et al. 2009, Lancee and Dronkers 2011, Lolle and Torpe 2011, Tolsma and Van der Meer 2017, 2018). While there is a risk of oversimplifying the differences by merging different ethnic groups together, it is nevertheless the well-established case that there are observable differences in trust levels between non-white and white individuals (e.g. Uslaner and Conley 2003). The evidence, both from the more homogenous ethnic societies of Europe and the heterogenous multicultural societies of the United States and United Kingdom, indicate that white individuals often report higher levels of generalised trust compared to non-white individuals (Stolle et al. 2008, Hooghe et al. 2009). There are multiple reasons for such findings such as societal factors, including historical injustices, systemic biases, and disparities in social, economic, and political spheres. Also in the UK, non-white individuals, particularly those who have experienced racial or ethnic discrimination, often exhibit lower levels of trust towards institutions such as Parliament, the NHS, and not least the police. This lower level of trust can be seen as a rational response to persistent inequalities and experiences of marginalisation. The police may be viewed with suspicion due to perceived systemic bias or unequal treatment, with mistrust seen as a rational response to persistent inequalities and experiences of marginalisation (Van Craen 2013, Van Craen and Skogan 2015, Kammersgaard et al. 2023).

Within ethnic minority communities, there can be high levels of specific trust in community-based institutions, often based on shared experiences, cultural ties, or community solidarity (Uslaner and Conley 2003). It is common for members of marginalised communities to trust one another or specific community institutions more than they trust mainstream society or institutions. Moreover, trust is a dynamic attribute and can change over time as societal attitudes shift and structural inequities are addressed (Levi and Stoker 2000). Efforts to promote social inclusion, equal opportunity and fair treatment can help to build trust among diverse ethnic groups, contributing to a more cohesive and equitable society (Murphy 2013, Pass et al. 2020). While ethnicity can influence trust, it is one of many interrelated factors that shape individual and collective trust patterns. Understanding these dynamics can aid in developing effective social policies, community interventions and strategies for building trust in multi-ethnic societies (Fennema and Tillie 1999).

Trust and socio-economic factors

Socio-economic factors, including age and income, play substantial roles in shaping trust (Hetherington 1998, Anderson and LoTempio 2002). These factors interplay with individual experiences and perceptions, impacting both generalised and specific trust (Uslaner 2000) and help to give us what might be defined as a geography of trust (McKay et al. 2023). Age is a significant variable influencing trust patterns, with some research suggesting that trust increases with age (Bailey and Leon 2019). This might be attributable to a lifecycle effect, where individuals become more trusting as they gain life experiences, develop long-term relationships, and become more settled in their communities, and also need to draw on government institutions such as healthcare and police (Jennings and Stoker 2004). However, other research posits a generational effect, with older generations being more trusting due to the societal norms prevalent during their formative years (Holm and Nystedt 2005). Income level is another central socio-economic factor impacting trust. Generally, individuals with higher incomes have more positive experiences with societal institutions, which generates higher levels of trust (Foster and Frieden 2017). It should be noted that the relationship between income and trust can be bidirectional: while income can influence trust levels, trust can also impact economic prosperity with high levels of trust in a society encouraging economic transactions, fostering investment and stimulating innovation, contributing to income growth (Dincer and Uslaner 2010). Lower-income individuals, on the other hand, might have less trust, particularly in societal institutions, due to perceived or real experiences of inequality and marginalisation and economic hardship. This is especially true if individuals feel the system is stacked against them (Uslaner and Brown 2005, McKay et al. 2021).

Trust and partisanship

Partisanship, or the extent to which individuals identify with and support a particular political party, plays a significant role in shaping trust, especially in the political domain (Citrin and Stoker 2018). The effect of partisanship on trust is complex and multifaceted, deeply intertwined with individual beliefs, values and perceptions of the political landscape (see Carlin and Love 2013, Clarke et al. 2023, Spina 2021). Firstly, partisanship can significantly influence specific trust in political entities; individuals tend to express higher levels of trust in politicians or institutions that align with their partisan identity. This trust is often based on the belief that the favoured party will advance policies that align with their own views and values (Spina 2021). A robust political system aims to maintain trust even when one’s preferred party is not in power. However, deep partisan divides can erode this trust, especially if opposing parties are seen as not just politically different, but as threats to the nation’s well-being (Dalton 2005). In effect an increase in polarisation can lead to decreased generalised political trust (Lee 2022). Moreover, partisanship can influence trust in non-party political institutions, such as the police, if these entities are perceived as biased towards a particular political party or viewpoint; this can erode trust among those who identify with the opposing party (Huang et al. 2013, Van der Meer and Kern 2019, Crandall and Lawlor 2022). Yet it must be understood that the degree to which partisanship impacts trust varies depending on political climate, polarisation and not least individual factors, both ideological and socio-economic.

Trust and communities

Trust within local communities is a foundational aspect of social cohesion and community vitality underpinning the social fabric of a community, influencing social interactions, cooperation and communal initiatives (see also Etang et al. 2011, Lancee and Dronkers 2011, Foster and Frieden 2017). Generalised trust reflects the broad expectation that most community members can be trusted, underpinning a sense of community belonging and willingness to cooperate with others (Stolle et al. 2008). Specific trust, on the other hand, is based on personal relationships and direct experiences with individuals within the community (Levi and Stoker 2000). Trust in the local community contributes significantly to a community’s ability to work together towards shared goals facilitating cooperation in neighbourhood initiatives, voluntary work and communal problem-solving (Fennema and Tillie 1999). High trust levels can lead to more robust community engagement and active participation in community affairs, enhancing collective efficacy. Local trust impacts the relationship between community members and local institutions, such as schools, local government and police (Fennema and Tillie 1999, Peyton et al. 2019). High trust in these institutions can promote cooperation, improve service delivery, and foster a shared sense of community identity and purpose. Yet the level of trust within a local community is influenced by demographic diversity, economic disparities and crime rates, and those shared experiences can all impact trust levels in a positive or negative fashion (Lolle and Torpe 2011). For instance, high levels of social inequality or experiences of crime can erode trust, while shared experiences and collective action can build trust. Local trust is also affected by broader societal trends and issues which might trickle down and impact trust within local communities (Fennema and Tillie 1999). On the other hand, local community trust can serve as a buffer during times of societal upheaval, fostering resilience and cohesion.

Hypotheses

Having presented the theoretical standpoints on trust and the relationship to the police we can now turn to formulating the hypotheses that we will be testing in our analysis.

The first hypothesis relates to gender, where the literature suggests that we should expect women to have higher trust than men. Our first hypothesis is a general gender hypothesis:

H1: Women are more likely to have trust in the police than men.

Ethnic minority groups are generally in the literature seen as having lower trust towards particular institutions, for instance the police, and we therefore hypothesise:

H2: Ethnic minority respondents are less likely to have trust in the police.

Turning to the socio-economic factors we include age and income as direct hypotheses and expect the following relationships to be present:

H3: Older people will have greater levels of trust in the police.

H4: People with higher income will have greater levels of trust in the police.

With regards to partisanship, we expect that respondents who state that they voted for the incumbent Conservative party to have higher trust in the police than other voters, which is why we include the following hypothesis for partisanship:

H5: Conservative respondents are more likely to trust the police.

Finally, given the London hypothesis for gender we also believe that we should test the direct relationship on whether London respondents are different to that of the rest of the country, given the size of an area covered by the Metropolitan Police which leads to the inclusion of the following hypothesis:

H6: London respondents are less likely to trust the police.

However, we acknowledge that this is particularly context dependent and given the problems that the Metropolitan Police have faced with crimes towards women and the actions of their officers, we also believe it reasonable that there would be a specific effect when looking at women in London versus any other group:

H7: Women in London are less likely to have trust in the police than other groups.

In the following section we present the data used to test the hypotheses and we present the operationalisation of the hypothesised relationships with variables available in the data. We further discuss the various control variables that will be included in the analysis and discuss the choice of methods for the analysis, before also touching upon the overall representativeness of the data.

Data and methodology

We asked our respondents how much they trusted the police on a seven-point scale, where 1 means ‘not at all’ and 7 means ‘completely’. YouGov were commissioned to perform a series of surveys. Data were gathered in 15 monthly waves, from July 2022 to September 2023. Respondents were exclusively from England; other parts of the UK were not included in this analysis.

The personal income of respondents is captured, on a 14-point scale, ranging from under £5000 to over £100,000 per annum. Ethnicity is a simple binary variable, where those respondents who do not identify as English/Welsh/Scottish/Northern Irish or British are coded as belonging to an ethnic minority. The London variable is created based on the regions in which respondents live in, identified by YouGov. Respondents living in London (12.1% of all respondents) are coded as 1, with respondents in other regions (North East; North West; Yorkshire and the Humber; East Midlands; West Midlands; East of England; South East and South West) coded as 0.

To examine whether there was a partisan effect, we created a Conservative binary variable, coded as 1 if the respondents voted for an MP from the Conservative party at the 2019 general election, with all other responses coded as 0.

As policing is related to crime, we included constituency-level crime control variables: the burglary and violent crime rates in the constituency (where rate is defined as the total number of incidents divided by the population of the constituency). An additional control was the vote share of the Conservative party at the 2019 general election in the respondent’s constituency. Finally, our interactive term looks at the interaction between women respondents and respondents living in London. Given that our literature review specifically considered the impact of generalised trust, and the research establishing that there is likely a connection between generalised trust and particular trust we have included a control variable for generalised trust. This is measure of seven-points on the extent to which the respondent believes that most people in general can be trusted. By including this as a control variable we ensure that any findings derived from our models will be on the element of specific trust in the police while controlling for the respondent’s level of generalised trust.

Analysis

Trust in the police is naturally a potentially fraught issue. On the one hand trust in the police is necessary for basic societal norms to function and for citizens to feel safe. On the other hand, crimes carried out by rogue and criminal police officers make it difficult to say that trust can be unconditionally given. While police and policing in general might not be traditional political wedge issues, the media exposure of clear crimes and injustices by police means that the issue of trust and policing has the potential of becoming more than a traditional valence issue.

Table 1 presents some descriptive statistics. It shows the mean levels of trust from a seven-point scale, where 1 means ‘not at all’ and 7 means ‘completely’. It shows that there is clear variance in levels of trust in the police across England. When looking at all respondents, we find the lowest levels of trust are to be found in London. This is also the case when we just look at women respondents. As can be seen, women trust the police more in every region, with one concerning exception: London. For ethnic minorities, baseline trust is notably lower (and as will be shown in the next section, this is significantly lower). However, in this case it is not London which is the problematic region, but the North East. More research is needed to piece apart this relationship.

Table 1. Mean levels of trust in the police across England.

Region	All	Men	Women	Ethnic minority
East Midlands	4.11	4.03	4.17	3.58
Yorkshire and the Humber	4.02	3.85	4.16	3.86
North East	3.93	3.68	4.15	3.38
South West	4.06	3.98	4.12	3.6
West Midlands	3.98	3.83	4.08	3.56
South East	4.07	4.06	4.08	3.65
North West	3.93	3.85	4.01	3.66
East of England	4	4.01	4	3.55
London	3.77	3.82	3.72	3.59

Figure 1 presents the time series of trust levels, looking specifically at women. As can be seen, women who live outside of London have consistently higher levels of trust in the police than women who live in London, across the 15 months of our survey.

Figure 1. Levels of trust in the police among London and non-London women. LOESS smoothed; grey areas represent 95% confidence intervals.

For a more formal analysis, we ran multivariate regressions looking at trust in the police. We present five models including a number of control variables which allow us to test the hypotheses put forward and draw conclusions from the results.

Our results show a mixture of the expected relationships, but also present some very interesting and in real-world terms problematic findings. We present five separate models. The first model includes the individual characteristics of the respondents, while in the second model we add measures for rate of burglaries and violent crime in the parliamentary constituencies of the respondents. Our third model includes two variables for partisanship, one related to whether the respondent has reported that they voted for the Conservative party in the 2019 election and a variable of the vote share of the Conservative party in the respondents’ constituency at the 2019 election. In the fourth model we include an interaction between being a woman and living in London in order to test the hypothesis regarding the negative press and general scandal regarding the Metropolitan Police. Finally, while the first four models have focused exclusively on a specific form of trust (trust in the police), our fifth model adds in an extra measure of trust: the extent to which people feel that generally speaking most other people can be trusted (measured on a one to seven scale, with higher indicating more trust). As can be seen from model 5 the inclusion of this control variable does not change either the sign or level of significance of any of the other independent variables, although it does increase the overall fit of the model. Nevertheless, with the inclusion in model 5 it means that we can have a greater confidence in our results, as they appear to be present even when generalised trust is held constant at its mean as a control variable.

Our first hypothesis stated that women would have higher trust in the police. Our results corroborate that finding: across all five models, women have a significantly higher trust than men ranging between .15 and .2 higher on the seven-point scale of the dependent variable. As will be discussed later this finding becomes even more interesting when it is seen in relationship to the interaction between women and residing in London. Our second hypothesis related to ethnicity, and we find support for our hypothesised relationship. Regrettably, ethnic minorities have significantly less trust in the police than white respondents. This is very much in line with what is reported in other research (e.g. Van Craen 2013, Van Craen and Skogan 2015, Pass et al. 2020). It is a finding that is also consistent with general research on ethnicity and trust (e.g. Lolle and Torpe 2011), and should make us pause for thought. Individuals can have positive or negative experience with public institutions and that will naturally influence their opinion of that institution. Bundling all non-white respondents together can also be seen as a very crude measure, but when the results are so stark as is the case here with non-white respondents trusting the police between .32 and .23 less than white respondents this suggests two things. First, there is more going on than just a random individual poor experience with the police; it appears to be systematic. Second, the police forces in England have a significant problem in front of them. Our sample is a nationally representative sample of England and 14% of the respondents are non-white. Our results demonstrate the need for further research into the reasons for why non-white respondents have significantly less trust than white respondents and such research both could and should inform decision-making by the police forces in England.

Our third hypothesis relates to age where the expected positive and significant relationship has been found. The older the respondent the higher the trust, although the actual effect is very small: for each ten-year increase in the age of a respondent we see an increase in trust of between .06 and .08. Our fourth hypothesis expected to find a positive relationship between income and trust. We do not find any significant influence of income on trust in the police. While the finding does not corroborate our expectations, it is also a finding that is not unwelcome, as it does suggest that the trust in the English police is present (or not) irrespective of income, while it is possible that income still could have an impact on overall societal trust.

The fifth hypothesis expects a positive relationship with being Conservative and trust in the police. We find this positive relationship for both Conservative variables that we include. Being a Conservative voter means that the respondents have up to a .29 higher level of trust in the police than non-Conservative voters, while a one unit increase in support for the Conservatives in the constituency of the respondents is associated with an increase in trust towards the police of up to .55. Including only one of these two variables does not change the direction or significance of the results. This does suggest that there is a partisan difference in how much trust is given to the police, which might be seen as problematic, although this could also simply be a general issue where Conservative voters have different trust setting towards the public institutions, especially given the Conservative party has been in government throughout the period covered by the dataset. Our sixth hypothesis stated that London respondents would have less trust in the Police. Notably, in general this is not corroborated by our results. While we did observe that London had the lowest mean level of trust in Table 1, the regression analysis in Table 2 shows that there is no statistically significant relationship. In model 1 there is a slight negative effect, while in model 4 the effect has become a slight positive, but only at the bounds of statistical significance, which given the sample size used in this analysis mean that we should reject the hypothesis as the evidence in favour of accepting it is not strong. A strong relationship is, however, what we find for our seventh and final hypothesis – an interaction between being female and living in London. We do not find this interaction in any of the other regions we analysed.

Table 2. Linear regressions of trust in the police.

	Dependent variable:
	Trust in the police
	(1)	(2)	(3)	(4)	(5)
Women	0.146***	0.145***	0.146***	0.176***	0.198***
	(0.035)	(0.036)	(0.035)	(0.038)	(0.036)
Gross personal income	−0.002	−0.002	−0.003	−0.003	−0.005
	(0.003)	(0.003)	(0.003)	(0.003)	(0.003)
Age	0.008***	0.008***	0.005***	0.005***	0.003***
	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
Ethnic minority	−0.321***	−0.315***	−0.266***	−0.265***	−0.226***
	(0.055)	(0.055)	(0.055)	(0.055)	(0.053)
Lives in London	−0.094*	−0.053	0.013	0.145	0.127
	(0.057)	(0.068)	(0.070)	(0.090)	(0.086)
Burglary rate		−17.278	−8.671	−9.343	−10.574
		(12.835)	(12.957)	(12.957)	(12.457)
Violent crime rate		0.546	0.973	0.990	2.549
		(1.668)	(1.664)	(1.663)	(1.600)
Respondent voted Conservative 2019			0.275***	0.275***	0.293***
			(0.040)	(0.040)	(0.038)
Percentage voting Conservative in constituency			0.440***	0.438***	0.550***
			(0.131)	(0.131)	(0.126)
Interaction: Women * London				−0.251**	−0.256**
				(0.108)	(0.104)
Trust other people					0.271***
					(0.011)
Constant	3.558***	3.612***	3.385***	3.369***	2.400***
	(0.069)	(0.080)	(0.106)	(0.106)	(0.109)
Observations	7821	7821	7801	7801	7800
R^2	0.019	0.020	0.028	0.029	0.103
Adjusted R^2	0.019	0.019	0.027	0.028	0.102

Robust standard errors in parentheses clustered on parliamentary constituency.

Note: *p < 0.1; **p < 0.05; ***p < 0.01.

Had this analysis been done in 2010 it is possible that this interaction would not have been obvious to include. Regrettably, events involving women and the Metropolitan Police, especially the murder of Sarah Everard, the overreaction by the police on vigils related to this, and numerous other negative events means that this interaction is a necessity to include when considering trust in the police. We find that there is significantly less trust by women in London towards the police than other groups. The interaction is between two binary variables making the interpretation quite straightforward, as can be seen in Figure 2.

Figure 2. Interaction of trust in police between being female and living in London

Our first hypothesis found that women in general had higher trust in the police than men; however, it is clear that in London women have a very different attitude and given the sample size this is very unlikely to be random. The various issues have resulted in changes in the leadership of the Metropolitan Police, but the question is whether these changes are nearly enough when these results are considered. When women in a society overall have trust in the police, and then in the national capital half the population report a lower trust, that is a severe problem both for the administrative leadership of the Metropolitan Police and for the political leadership, the Mayor of London. What our findings do not show are the direct reasons for why we observe this lower trust amongst the women of London. However, what we do know is that trust is dependent on context and experiences of past events can influence how trust is given and perceived. The problems that the Metropolitan Police has faced in the past decade and especially in the past 3–4 years are such that there is a monumental task ahead of rebuilding trust, particularly among women, in what is a central institution in a city the size of London. Considering what factors might drive this effect, outside of those we can directly measure and observe within the public are some of those also highlighted by the Casey review. The Casey (2023) report made it clear that the recruitment and vetting in the Met was poor and that the day-to-day management of people and ability to deal with issues raised were equally challenging. In fact, the Casey review went one step further and highlighted the Met police as suffering from ‘optimism bias’, i.e. the organisation was overly focused on signalling a positive message. Such biases can lead to unwelcome problems being hidden and ignored by managers who are focused on telling the positive story. The Casey review (2023) was not the first review of the Met police to highlight a general problem with diversity amongst staff. In fact the Macpherson inquiry (1999) into the murder of Stephen Lawrence also highlighted this.

Yet, the issues raised in our analysis are not altogether unique to London; in fact our findings do not present an altogether positive story of trust in the police in England 2022–2023. Our results point towards severe issues, for instance, ethnic minorities still have lower levels of trust toward the police than white respondents. This is of course a perception, and we are not arguing that this is correct or incorrect, merely that this is what the ethnic minorities report and what they believe, whether this is part of a negativity bias is a question that is beyond the scope of this article, although we note that the results from the US with regards to ethnicity and negativity bias are mixed (see e.g. Wu et al. 2009, Taylor and Lawton 2012), not least when media reporting is considered (Rosenberger and Dierenfeldt 2022). In other words, there is a problem across England in how the police deal with ethnic minorities and with 14% of our nationally representative sample falling in this category, there is much work ahead for the police force to increase trust among a seventh of the population. While our findings did not find an influence for income, it did find it for Conservative voters. We do know that white respondents are more likely to report voting for the Conservative party than ethnic minorities, and the findings might also be driven by this. However, in the ideal scenario we should not be able to see any partisan differences in how much the police are trusted, as it should be equal across all parties, or at least the major players. This is not the case currently and something which is also worth keeping under examination as a police force cannot, whether true or not, be perceived to be more on the side on one partisan group than another; this is completely counter to the fifth of the nine principles on which the Metropolitan Police force was founded. Our results from women respondents in London also provide further evidence that the Metropolitan Police have some almost insurmountable challenges ahead, as their behaviour towards a particular group has led to the members of this group finding themselves let down, which comes across clearly in our findings.

However, our findings also must be considered in light of a caveat. While the results are based on a large number of observations measured across time, there is also no direct measure of personal experience with policing, merely a question asking specifically about trust in the police. In that regard, it is possible that the negativity bias that is generally accepted to be present in views towards the police (see also Skogan 2006, Myhill and Bradford 2012) might be a factor having an impact. Although, as also noted by Myhill and Bradford (2012) direct encounters between the public and the police have the potential to influence the wider public views, but it is not a certainty. In fact, as demonstrated by Peyton et al. (2019) contact between the police and the public in a non-enforcement fashion improves public attitudes towards the police.

Concluding remarks

A police force that is trusted by most and where there are no major differences in trust based on gender, ethnicity or age, should – uncontroversially – be a goal for any society. Our results suggest that in England this is not obtained just yet. First, we show less trust in the police among respondents with an ethnic minority background, suggesting that the issues raised already by the Macpherson (1999) review are still present. Second, overall there women have more trust in the police than men, but this is a result of a divided country. When we consider women living in London, and thus under the remit of the Met police, there is a very different relationship in play. The Met police has a mission to keep London safe for everyone and the ultimate vision for the Met is to be the most trusted police service in the world (met.police.uk 2023). Our findings suggest that the task ahead of the Met police is monumental. The publication of Baroness Casey’s report highlights some of the challenges facing the Met, but likely overestimates the levels of trust. We argue that there are significant socio-economic factors impacting trust towards the Met. Some of the factors are national problems, such as the lower trust towards the police amongst people with an ethnic minority background, while other factors are an issue for the Met alone, such as the clear problem for the Met police in re-establishing trust among women.

Our findings place a further perspective on the findings in the Casey review and suggest that the problem is much more substantial. We have presented a framework for why trust is important in society in general and in the police in particular. The Peelite principles of policing which spread across the world as other police forces were created have been consistently let down by the Met police across many years, meaning that particular sub-groups see a perceived difference in how the Met police treats them. Rebuilding trust is another matter altogether and a recognition that this is not done overnight and through smart messaging, but instead through hard graft and on-site engagement is necessary. Considering our findings in the context of the mission and vision of the Met police leaves one to wonder whether the leadership of the Met is aware of the issues facing the institution, and whether the Mayor of London is accepting that the police force in his city has a problem with women. The fact remains that trust is very much built on personal experiences and if a particular group has problems with an institution, then the best way to rebuild trust is for that institution to work on this in their activities. Whether this is possible for the Met police or whether this requires a more radical approach remains to be seen. Unfortunately, it is not likely that either the general problems for the police in England of lack of trust from ethnic minorities or the specific issue of lack of trust in the Met among women in London will be solved that easily and more research in certainly needed in this crucial element for successful policing.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Japan Society for the Promotion of Science (JPJSJRP 20211704 ); UK Research and Innovation (ES/W011913/1).

Notes

1 While these principles are widely acknowledged in modern policing discourse, there is no evidence that the nine principles were codified as far back as 1829 or that they were created by Robert Peel. The Home Office (2012) does acknowledge that the principles cannot be attributed to Peel, but their claim that they can be attributed to the first Met Commissioners is also unsupported by evidence. Several of the principles are certainly ‘Peelian’, but the most likely source would appear to be Reith (1952, p. 154). See Lentz and Chaires (2007) for further discussion.