A randomised control trial testing the impact of police reassurance contact on hate crime victim withdrawal

Abstract

Hate crime has an extremely detrimental impact on victims and minority communities. Despite this, the policing response is oft criticised and investigations are characterised by high levels of victim withdrawal. Consequently, offenders are rarely prosecuted. This exposes victims to elevated risk of repeat victimisation and undermines trust in communities with historically limited confidence in the police service. This study reports on a randomised control trial testing whether Secondary Reassurance Contact (SRC), a follow up from a neighbourhood police officer not connected with the investigation, can reduce victim withdrawal of support for a prosecution (attrition) in hate crime investigations. Findings indicate that SRC results in a 12–15% reduction in victim withdrawal. The practical implications are considerable and if scaled nationally SRC may result in tens-of-thousands fewer victims withdrawing support for hate crime investigations each year. SRC therefore offers police decision-makers a tangible and deliverable option to improve hate crime outcomes and to bolster confidence amongst minority communities.

Keywords

Hate crime investigation victim withdrawal policing

Introduction

There is limited international consensus on what constitutes ‘hate crime’ (Garland and Funnell, 2016). However, a focus on groups with specific characteristics is consistent across efforts in most jurisdictions to legislate in this sphere (Chakraborti, 2014). In England and Wales the legalistic definition of hate crime, is any criminal offence which is perceived, by the victim or any other person, to be motivated by hostility or prejudice towards someone based on- race, religion, disability, sexual orientation or transgender identity (CPS, 2022). The law in the UK enables enhanced redress for example aggravated sentences, for offences motivated by or demonstrative of ‘hate’ towards these specific groups.

Definitional difficulty (Garland, 2012) and a normative desire to apply hate crime laws to a broad set of practical circumstances has resulted in a number of individual efforts by police forces in England and Wales to operationalise the recording and investigation of hate in often isolated and unique ways (BBC, 2021; Garland and Hodkinson, 2014). The College of Policing (2020) have sought to achieve uniformity by way of national guidance intended to guide forces in the identification, recording and investigation of hate crime. Translating this to practice has not proven to be straightforward with the advocated approach being subject to legal challenge and subsequent amendment (Lloyd, 2022).

Impact

Despite lack of clarity in this area it is posited that hate crime, with its basis in bias, has a considerably more detrimental impact on victims than equivalent non-bias based offending (Iganski, 2001). Victims of hate crime report considerably elevated feelings of loss of confidence, vulnerability, and psychological and physiological side-effects of victimisation (Allen and Zayed, 2022; Williams and Tregidga, 2014). Further, hate crime has a deleterious impact on the wider community (Hall, 2005). It constitutes a ‘signal crime’ (Innes, 2014) impacting negatively on broader perceptions of crime and feelings of safety (Shirlow et al., 2013). Others who are not directly impacted but share the characteristics that were the basis for the targeting of the victim may feel a sense of fear and vulnerability (IACP, 1999). In this way hate crime may be considered pervasive beyond those directly targeted and their associates. Despite this invasive impact on individuals and communities hate crime is grossly underreported (Pezzella et al., 2019) which perhaps limits both understanding of such offending and adequacy of the response.

Minority communities in particular, with habitually lower levels of trust and confidence in the police (Barrett et al., 2014; Madon et al., 2017), are disproportionately impacted by hate crimes. Improving the policing response to hate crime is considered key to securing trust and confidence in those communities that are disproportionately impacted (Chakraborti, 2009a). This is of particular importance given deteriorating public confidence in policing (Brunt, 2023) and frequent discontent at the state response to issues that disproportionately impact minority communities (Bowling et al., 2008).

Victim attrition

The majority of case attrition in the criminal justice lifespan occurs during the police investigative process (Murphy et al., 2022). This is in large part owing to the concept of ‘victim withdrawal’ that is victims deciding that they no longer wish to support the progression of a case through the criminal justice system (Hohl and Stanko, 2015). The literature speaks to prevalence of victim withdrawal amongst non-hate crime types (Davies and Westcott, 2006). There is however a particular preponderance of withdrawal amongst crime types that are inherently personal and intrusive, to include rape (Hansen et al., 2022), other serious sexual offences (Dalton et al., 2022) and domestic abuse (Cretney and Davis, 1997; Sleath and Smith, 2017).

Victims of hate crime report feeling let down by police insensitivity or lack of progression following initial reporting (Williams and Tregidga, 2013). Bowling (1999) found that inaction, failing to keep the victim informed and not appearing to care, were common causes of victim dissatisfaction in hate crime investigations. Hate crime investigations currently take more than twice as long to complete as comparable non-hate crime offences (Allen and Zayed, 2022). Consequently, victims that report hate crime are less satisfied with the response than victims of crime generally and additional focus on responding to the needs of victims is advocated (Walters et al., 2016). With extended periods of investigation and dissatisfaction with the service provided, comes elevated risk of victim attrition with victims withdrawing their support for the investigation (Burman and Brooks-Hay, 2021; Hudson and Paterson, 2020).

Hagerlid and Granström (2023) identified this inability to secure victim ‘cooperation with the investigation’, oft linked to limited trust amongst minority communities, as a key problem with hate crime investigations. Consequently, victims withdrawing support for an investigation by retracting the allegation made, refusing to provide evidence or declining to attend court are prevalent reasons for unsuccessful hate crime prosecutions (Walters et al., 2017).

Responses to secure victim support

Policy makers and law enforcement have, following a range of reviews, sought to find solutions to minimise the risk of victim withdrawal during the investigative process (Hohl and Stanko, 2015; McGuire et al., 2021). This has predominantly drawn upon a victim-centred approach that seeks to promote the needs of the individual victim above, or at least to achieve parity with, the aims of the criminal justice system in achieving prosecutions and convictions (McPhee et al., 2022).

This focus on enhancing victim support and catering for individual victim needs throughout the investigative process are most apparent in the spheres of domestic and sexual abuse. Independent Domestic Violence Advisers (IDVA) are provided in domestic abuse cases. They sit outside of the investigation and their core role is to support and advocate for the victim (Safelives, 2024). McPhee et al. (2022) found that in cases where IDVA support was available arrest, charge and conviction were more likely. Albeit, Ross et al. (2022) found adverse consequences to include repeat victimisation, increased harm and lower conviction rates following IDVA provision at court only. Perhaps this indicates the perils of viewing victim-centred support as a means to support victims at the end of the process only, rather than throughout the criminal justice life course.

Similarly, Independent Sexual Violence Advisers (ISVA) were implemented as a policy response to support victims in sexual offence cases. ISVA support to victims is considered essential to secure their ongoing support and progression of cases to conviction (Hester and Lilley, 2015). Victim-centric approaches, as exemplified by IDVA and ISVAs, with overt focus on responsiveness to individual victim needs, rather than the requirements of the criminal justice system, are considered vital to indirectly secure victim support for prosecution (HMICFRS, 2021) and optimal opportunity for criminal justice outcomes (Hester and Lilley, 2018).

Secondary responder programmes are follow-up visits by police (often in conjunction with support agencies) to victims subsequent to their having reported a crime. They aim to offer support to victims in the immediate aftermath of offending (College of Policing, 2015). In the context of domestic abuse, this approach was found to increase victim reporting and uptake of services, perhaps indicative of increased victim satisfaction and confidence (Davis et al., 2008; Petersen et al., 2022).

Victim support- positive outcomes and deterrence

Victim support for the investigative process has consistently been demonstrated to be a key predictor of case progression (Anders and Christopher, 2011; King and Growette Bostaph, 2024) and positive criminal justice outcomes (McFadzien et al., 2020). Inability by law enforcement to secure victim support for hate crime investigations is therefore extremely limiting when it comes to achieving positive criminal justice outcomes for example conviction. It may go some way to explaining why charge and conviction rates for hate crimes are particularly low (Chakraborti, 2018). This is problematic as such outcomes may be key to prevent harm to the victim or others who share the characteristics that have been targeted by the perpetrator. By limiting the viability of prosecution in hate crime cases, preventative incapacitation (Robinson, 2019) is somewhat undermined. This is of particular import in cases perpetrated by ideologically motivated offenders rather than those who may have committed hate-based offending in specific contextual situations (Iganski et al., 2011). Inability to exert legal control over motivated offenders in such cases may place victims at risk given elevated levels of repeat victimisation in the context of hate crime (Corcoran et al., 2016).

Similarly, victim support for prosecution is a key determinant of suspect arrest (Richards and Harinam, 2020). This is of relevance to the concept of specific deterrence (Maxwell et al., 2002; Sherman and Berk, 1984) generated predominantly from the certainty of punishment (Grogger, 1991). Arrest is perhaps the most obvious and visceral representation of this certainty. Nagin (2013) suggests that more specifically it is certainty of ‘apprehension probability’ that is the likelihood of being caught, rather than necessarily certainty of suffering subsequent legal sanction, that serves to deliver deterrence. With certainty of apprehension being key to prevention and victim support for prosecution being crucial to facilitate said apprehension, it is apparent that bolstering victim support is necessary to deter future hate crime victimisation.

Importance & gaps in the evidence-base

Recorded hate crime has seen considerable increase in recent years (Home Office, 2022). It is posited that societal changes to include COVID (Olajide, 2022), internet access (Chan et al., 2016) and terrorism (Hanes and Machin, 2014) have driven hate crime prevalence. Of note a number of factors of particular contemporary relevance to include economic instability and political capitalisation on anti-immigrant sentiment serve to fuel hate crimes (Chakraborti, 2018). In England and Wales, 93% of hate crimes reported do not result in a suspect being charged or summonsed to court and victim withdrawal of support is by some way the most common outcome in cases that do not progress via the criminal justice system (Home Office, 2023). It is considered that this failure to secure victim support, a key predictor of arrest and prosecution means that victims are at risk from further offending, especially given elevated levels of repeat victimisation with hate crimes.

It is noted that despite increasing prevalence and high levels of victim attrition, there is limited empirical research that directly tests specific police responses to hate crime. The majority of the hate crime literature considers: the broad policing challenges against the historical context (Chakraborti, 2009b); public attitudes towards hate crime (Bacon et al., 2021); descriptive work detailing how various guises of hate crime manifest (King and Sutton, 2013) and police officer and force perspectives on dealing with hate crime (Trickett and Hamilton 2016).

There has to date been no empirical assessment of tactics adopted by the police in an effort to secure victim support for investigation and prosecution. Consequently, potentially analogous empirical research is drawn upon. This is derived predominantly from efforts to secure victim support in the context of domestic abuse and sexual offences. This dearth of empirical assessment of specific interventions applied to hate crime is perhaps reflective of the comparative infancy of hate crime scholarship in the UK (Chakraborti, 2009b). This gap in the literature is remiss and Walters et al. (2016) suggest that given few hate crime interventions are evidence-based, researchers should work to undertake robust evaluations of programmes in this space.

Criticism of the police response to hate crime suggests that failure to secure victim support is in part because victims oft do not feel listened to, do not consider their report is being taken seriously and are not treated with due empathy. This manifests in the perception of the criminal justice system as ‘slow, intimidating and incomprehensible’ (Chakraborti, 2018). Key then to ensuring victim satisfaction is a timely and appropriate response that demonstrates that the matter is being taken seriously (Williams and Tregidga, 2013). Kielinger and Paterson (2007: 203) noted that police services need “to be aware of the connection between improving the quality of their investigation and providing reassurance to the individual victims of hate incidents”. Put simply, the policing response must take steps to reassure victims, secure their support and by extension optimise opportunities to identify and deal with offenders (Iganski et al., 2005). If realised, this would further bolster victim confidence by addressing another prominent frustration amongst hate crime victims: failure by the criminal justice system to bring offenders to justice (Chakraborti, 2018).

Intervention

Given limited extant evidence on effective means for police to prevent victim withdrawal of support for hate crime investigations, we sought to address this gap. Upon reviewing potentially analogous interventions utilised in relation to other crime types, secondary responder programmes in the context of domestic abuse were noted to increase victim reporting and uptake of services, perhaps indicative of increased victim satisfaction and confidence (Davis et al., 2008; Petersen et al., 2022). Consequently, we resolved to test whether a similar secondary response following report of hate crime served to increase victim satisfaction and accordingly reduce victim withdrawal of support in the investigative process.

Methodology

This study was undertaken within the UK’s largest police service, the Metropolitan Police Service (MPS). A randomised control trial (RCT) method was used. Sample size was determined by way of a priori power analysis (Vo and James, 2012), using G*Power (Faul et al., 2009). In line with convention this aimed to ensure statistical power >.8, at significance level p < .05 (Lacey 2012; Vo and James 2012). Based on this sample size n = 250 was determined.

All hate crimes reported on South West Basic Command Unit (SW BCU) were identified from the application of a hate crime flag on the MPS crime recording system (CRIS). 250 reports were randomly assigned to treatment (secondary reassurance contact) or control (no secondary reassurance contact/ordinary service). This was done using a random number generator to randomly assign treatment/control to the rows 1-250 of an excel spreadsheet. As cases entered the experiment by way of a ‘trickle flow’ process (Ariel et al., 2022) they were added to the spreadsheet in chronological order to said numbers 1-250. Consequently, they were assigned to the treatment/control condition randomly associated with that number. In this way 125 case were assigned to each condition. These cases were drawn from the period October 2022 – January 2023.

If a case was assigned to receive secondary reassurance contact it was referred for a visit to the relevant neighbourhood policing team corresponding with the victim’s home address. Referral was achieved via CRIS notification (code added to bring crime to the attention of neighbourhood policing teams with relevant geographical responsibility) and follow-up via email to the relevant officers. Officers completed the contact and noted completion on the CRIS.

Variables

Independent variable

The piloted intervention (coined ‘secondary reassurance contact’ (SRC)) was contact from a local neighbourhood police constable or Police Community Support Officer (PCSO) with geographic responsibility for the area that the victim lived in. This took place either in person or via phone depending on the circumstances and victim wishes. This was distinct and separate from any contact from officers investigating the reported offence. This form of contact in these circumstances does not habitually take place within the MPS. Ordinarily the only contact a victim receives is from the officer investigating the reported offence. Evidence suggested that key to maintaining victim support is demonstrating an understanding, supportive approach that is flexible to victim need (Farrell and Lockwood, 2023). This is what the SRC aimed to do. To recognise what had happened to the individual, to acknowledge the impact and to demonstrate that the police service took it seriously and was committed to assist.

Given the broader impact of hate crime on the community and the inherent link with trust and confidence we sought to situate delivery within pre-existing neighbourhood policing teams (Longstaff et al., 2015) embedded within communities. The officers delivering the SRC were given loose guidance that primarily focused on provision of reassurance to victims. A conscious decision was made not to be overly prescriptive regarding the format or content of the SRC. This endeavoured to ensure the contact was victim-led, responsive to victim need and to allow officers to use their discretion to bring their local contextual knowledge and relationships to bear. It is accepted that this limits the ability to speak to the precise mechanism behind any effect.

Dependent variable

The aim was to determine whether the SRC impacted on victim attrition. Did the victim remain supportive of the investigation throughout? Victim attrition was considered to have occurred when a case concluded with ‘outcome codes 14 or 16’- this is when a case is closed on the basis that the victim does not support police action. Where a case was closed using any other outcome code, the case was counted as there not having been victim attrition. Outcome codes referenced are determined by the Home Office (see Table 2.1 in Home Office (2023) for details). Outcome codes for specific offence investigations were derived from the MPS crime recording system (CRIS). This allowed for binary coding of each case as ‘victim attrition’ or ‘no victim attrition’.

CRIS was used to monitor and extract the following:

(1) Whether or not SRC had been delivered for each case in the treatment group. If contact was not made an escalation process was followed through the relevant officer’s supervisory structure.

(2) Whether or not SRC had not been made for each case in the control group.

(3) The outcome code for each case.

This data was then used to conduct the analysis detailed below.

Implementation

Senior agreement and buy-in was considered necessary to signal organisational importance and to facilitate effective implementation. Initial discussions were undertaken through the SW BCU Hate Crime Improvement Group and agreed with the neighbourhood strand Senior Leadership Team. It was considered that buy-in and understanding of the officers delivering this pilot and their supervisors was key to effective implementation. Officers and supervisors were briefed via email and this was followed with a number of online, interactive sessions.

The following are relevant points from officer briefing:

• Officers were not briefed prescriptively regarding how they should approach the contact or what they should discuss. The aim was to test the effect of the contact itself. It was also considered that this allowed the officers to adapt to the specifics of the case and victim. The only suggested actions were completion of risk assessment and support referrals, if this had not already been completed.

• It was suggested that in person contact was preferable but that the wishes of the victim should be followed so some contact was likely to take place by phone. This was considered preferable to insisting on in person contact as in some cases it may be contrary to victim’s wishes or place victims in unintended difficult situations/danger.

• Whilst separate from any ongoing investigation it was considered prudent to give some direction on what neighbourhood officers should do if a victim wished to provide a statement etc. This was included in the briefing.

Briefings and launch were timed to coincide with National Hate Crime Awareness Week in order to build momentum.

Treatment delivery

The intention was to deliver SRC to 125 cases, 50% of the total sample. Treatment was considered to have been delivered if officers recorded on CRIS that they had spoken to the victim in person or over the phone. ‘Not treated’ included when officers did not endeavour to make contact and when they did try to contact the victim but could not make contact in any way that the victim could conceivably have been aware of. ‘Treated’ and ‘not treated’ were defined in the same way for both treatment and control groups.

Treatment fidelity

In the treatment group 60% of non-excluded cases were treated. The pilot failed to deliver the treatment in 40% of cases. Of the non-excluded cases in the control group there was 98.4% fidelity. Incorrect contact or treatment cross-contamination occurred in less than 2% of cases.

Analysis

Exclusion criteria

Cases were excluded from analysis if: they were duplicate reports, victim details were missing, the case was transferred to another police force or the report was re-classified as not being a hate crime contrary to initial classification. In line with this exclusion criteria 16 cases were excluded from analysis in the treatment condition. The same exclusion criteria were applied to the control and 17 cases were excluded on this basis. The sample post-exclusion is detailed at Table 1.

Table 1.

Sample post-exclusion.

Treatment	Control	Total
109	108	217

Analytical approach

The failure to deliver treatment to 40% of the treatment condition cases causes considerable issues in drawing causal conclusions. Consequently, due consideration was given to the optimal analytical approach to mitigate against incomplete treatment delivery.

‘Intention-to-treat’ analysis (ITT) involves analysis of all cases in their intended treatment/control group, irrespective of whether they received the relevant treatment. The strength of this approach is that it maintains the benefits of randomisation- namely the equivalence of groups and mitigation of confounding variables. It has particular use in medical trials when subsets of patients with certain characteristics may be more likely not to comply with treatment. It demonstrates what is likely to happen in practice. However, it involves analysing the 40% of treatment cases where a SRC wasn’t delivered (and to a much lesser extent the <2% control cases where it was) within the intended group. This involves including cases to assess whether the intervention had an effect when there is no conceivable mechanism explaining how it could given the intervention wasn’t actually delivered. This likely underestimates any magnitude of treatment effect in the case that SRC was effective at delivering improved outcomes.

Conversely ‘As-treated’ analysis (AT) includes all cases but they are analysed as per the treatment they actually received. The weakness of this approach is that it works contrary to the randomisation process, the basis of the RCT. Equivalence of groups is compromised. Whilst there are no obvious differences between cases where treatment should have been delivered and was, and those where it should have been delivered but wasn’t, there is no way to know for sure. Using this method may conceivably, in certain cases introduce bias and give the impression of causation where none exists (McCoy, 2017). Results should therefore be interpreted with caution.

Lee et al. (1991) detail the significant benefits of ITT when compared to a treatment analysis. However, of importance, this study isn’t analogous to assignment of a medication. In that case non-compliance often relates to subject/patient decision-making. Patients who comply with treatment regimens and those who do not may differ with regard to certain characteristics that also impact the outcomes measured (Smith et al., 2021). A significant confounding factor skewing the results obtained. However, in this study, when non-compliance was explored it overwhelmingly related to officer implementation that is officers not attempting the SRC, rather than trying to deliver it and being unable to contact victims or victims declining the contact. The vast majority of cases that were not treated, despite being randomised to treatment condition, had little to do with victim choice. This reduces (but does not eliminate) the likelihood of differential characteristics between compliant/not compliant cases when compared to other contexts.

Hernán and Robins (2017) discuss the risk of failing to identify significant effects by over-reliance on ITT in pragmatic trials when poor adherence to treatment condition is attributable to a factor that could be readily controlled in a clinical setting for example easily treatable side-effects. It is arguable that this is analogous to officer compliance in the case of SRC: in a fully operational setting officer compliance can be mandated at a force policy level. Another point of differentiation between this and a medical context (as well as many others) relates to defining whether or not treatment has been delivered. It may be problematic to utilise treatment analysis where treatment is a matter of degree for example number of doses, with partial compliance. That is not the case in this study with SRC constituting a singular, discrete, easily measurable contact.

In summary, it can be said that ITT assesses the impact of assigning a secondary reassurance contact and AT the impact of receiving it (Smith et al., 2021). The aim of this study was to assess the impact of receiving the contact and for the reasons detailed above AT is not considered wholly inappropriate. ITT will be reported and analysed given it is the preferred default method as it reduces bias and reports what actually happened from assignment (Lee et al., 1991). AT will also be reported as it speaks to any impact of actually receiving a SRC (Ellenberg, 1996). Gyani et al. (2023) discuss the validity of this approach to optimise robustness where treatment fidelity is weak but can be readily improved in an operational environment.

Analytical methods

Three elements are calculated and all are reported for both ITT and AT. The first is a chi square test (Gordon, 2018) which assesses whether the variables are independent that is are not associated with one another. If there is no association then SRC cannot be said to have had any impact on victim withdrawal. Cramer’s V (McHugh, 2018) is also reported indicating strength of any association. Secondly a calculation is undertaken to gauge both ‘absolute’ and ‘relative’ risk (Andrade, 2015) that is what is the risk of victim withdrawal with a SRC relative to current practice when there is no SRC undertaken. Thirdly, the ‘Number Needed to Treat’ is calculated (Cook and Sackett, 1995). In the context of this study, this gives insight into the number of cases that would likely need to receive SRC in order for one fewer victim to withdraw than would be the case with the business as usual policing response.

Table 2.

Intention-to treat.

	Treatment	Control	Total
Outcome 14 or 16	49	62	111
Other outcome	60	46	106
Total	109	108	217

Results

Intention-to-treat

Intention-to-treat outcomes are as per Table 2. A chi-square test of independence was performed to examine the relationship between secondary reassurance contact and recording of outcome 14 or 16. Chi-Square statistic is χ² (1, N = 217) = 3.367, p = .066. and the corresponding Cramer’s V = 0.12, which indicates a small effect size.

Absolute risk of victim attrition in the treatment group is 0.45. Amongst those assigned to control group the absolute risk is 0.57. The Relative Risk is 0.78. The risk of victim attrition is reduced by 22% in the treatment group relative to the control.

Based on this it is estimated that the Number Needed to Treat (NNT) is 8. For every 8 cases that receive secondary reassurance contact, as opposed to the business as usual response, it is anticipated that one fewer case would suffer from victim withdrawal.

Table 3.

As-treated.

	Secondary reassurance contact	No secondary reassurance contact	Total
Outcome 14 or 16	29	84	113
Other outcome	40	64	104
Total	69	148	217

As-treated

As-treated outcomes as per Table 3. A chi-square test of independence was performed to examine the relationship between secondary reassurance contact and recording of outcome 14 or 16. Chi-Square statistic is χ² (1, N = 217) = 4.09, p = .043 and the corresponding Cramer’s V = 0.14, which indicates a small effect size.

Absolute risk of victim attrition in those who received secondary reassurance contact is 0.42. Amongst those who did not receive victim contact the absolute risk is 0.57. The Relative Risk is 0.74. The risk of victim attrition is reduced by 26% in the treatment group relative to the control.

Based on this it is estimated that NNT is 6.8. For every 6.8 cases that received secondary reassurance contact, instead of the business as usual response it is anticipated that one fewer case would suffer from victim withdrawal.

Discussion

Limitations and areas for future research

Before considering the policy implications of the findings, it is important to recognise the limitations of this study and to consider how future endeavours may seek to address these. The primary limitation is the poor compliance within the treatment group. It must be acknowledged that this and subsequent marginal failure to reach significance with ITT analysis mean that the internal validity of the findings must be treated with caution. Consequently, future studies seeking to augment these findings may benefit from a concerted focus on implementation, perhaps running a pre-trial phase to embed the practice and drawing on the implementation literature (Fixsen et al., 2005; del Pozo et al., 2024).

It should also be highlighted that this study is based on a small sample of hate crimes drawn from one area, within a single police force, over a short time-period. It cannot therefore be assumed to have external validity and to be generalisable. Larger replications across geography and time would be beneficial to give police leaders and policy-makers confidence that the findings would hold true when applied elsewhere.

Similarly, given this study was not prescriptive about the specific approach officers took when conducting a SRC it cannot speak, to any degree of specificity, on differential effects of differing approaches for example contact in person versus contact by phone etc. This may mean that the outcome differs dependant on the approach of the officers tasked with delivering SRC. Future efforts may wish to be more prescriptive with a view to testing whether different mechanisms of delivery have differential impact. It may also be possible to test SRC in a much more granular way. This could include assessing differential impact on different forms of hate crime for example racial, disability etc. The form of the SRC itself could also be built upon and tested for example joint deployment with third sector support groups etc. By testing various iterations, it may be possible to speak to optimisation of the SRC in a way this study cannot.

Finally, this study tests whether SRC reduces victim withdrawal. As mentioned reducing victim withdrawal is not considered an end in its own right. Rather, it is considered a proxy for victim satisfaction with the police response and as a mechanism to improve criminal justice outcomes. Future studies may seek to test the relationship between SRC and satisfaction directly. This would require a direct survey or similar with hate crime victims. This was not possible with this study. Similarly, the impact of SRC on charge (or other positive outcome) rates could be assessed. Given the relative rarity of suspects being charged it was not possible to do this meaningfully within the limits of this study. A much larger number of cases than were available in this study would be needed to achieve suitable power in order to detect any effect.

Implications

Limitations notwithstanding, it is considered that there are considerable implications for policy and practice from the findings of this study. There was a negative association between SRC and victim withdrawal that is SRC was associated with victims remaining supportive of the police investigation. This association was statistically significant in the as-treated analysis and only marginally failed to reach significance in the intention-to-treat analysis despite poor treatment delivery. The effect size was meaningful. Within the sample, victims withdrew support for the investigation when a case was assigned a SRC in 42–45% of cases. Comparatively, victims withdrew support for hate crime investigations in the case of business as usual in 57% of cases. This is a reduction of victim withdrawal of between 12 and 15 percentage points with SRC.

This effect size has practical significance. 145,214 hate crimes were recorded in England and Wales in 2022-23 (Home Office, 2023). If SRC was scaled nationally and similar effects were noted across geographies, a 12-15 percentage point reduction would equate to approximately 17,425-21,782 fewer victims withdrawing support for hate crime investigations each year. When considered in the context of the literature, this insight may be instructive for policy makers when considering potential implications for trust and confidence amongst minority communities and the potential implications for improving victim experience and criminal justice outcomes in relation to hate crime.

Findings estimate that for every 6.8-8 victims of hate crime that receive a SRC, one additional case would retain victim support that would have been withdrawn were it not for SRC. Against the backdrop of competing demands so prevalent in the context of contemporary law enforcement (Ludwig et al., 2017), this provides some insight into the potential resources required from police forces to deliver this service and the return on that investment.

On the subject of cost-benefit, it is worth noting that during this trial SRC was completed using existing processes and without additional budget that is officers completed SRC within their ordinary working hours. Therefore, when considered alongside the reduced victim attrition it could be argued that SRC represents a cost effective and efficient use of police resources to maintain victim engagement in hate crime investigations. Whilst it is conceivable that the failure to deliver SRC to such a high proportion of the treatment group may be indicative of insufficient capacity, it is of note that since this trial was conducted the policing area in question is now delivering SRC for hate crime as part of business as usual. This has seen close to 100% delivery based on full incorporation into staff daily working routines and an updated computer system that enables direct tasking. Nonetheless adequate capacity is something that police leaders must consider if seeking to further test or implement SRC.

Conclusion

This study sought to begin addressing a gap in the extant evidence-base, namely whether direct police practices could serve to reduce the considerable victim attrition in hate crime investigations. Not only is this failure to secure victim support indicative of low levels of trust in policing amongst vulnerable minority communities but it is well documented in the solvability literature as a barrier to identifying offenders and bringing them to justice. The specific practice assessed is secondary reassurance contact (SRC), an additional victim contact by neighbourhood officers intended to reassure the victim and demonstrate that the police are taking the report seriously.

This study found that there is a 12–15% reduction in victim attrition in hate crime investigations when a SRC is made by a police officer/PCSO outside of the investigation. This may offer police leaders and policy-makers an evidence-base on which to increase victim satisfaction with the policing response and by extension to ameliorate low levels of trust in policing within minority communities. Further, SRC may provide a practical means, within the gift of police leaders, to increase positive criminal justice outcomes bringing more offenders to justice. To this end, findings that SRC serves to reduce victim attrition can reasonably be described as promising. Whilst further research is required to bolster the internal and external validity of findings, this study should as a minimum justify further exploration of how SRC may be optimised and whether it may also serve to reduce victim withdrawal in other crime types that suffer from high attrition levels.

Footnotes

Declaration of conflicting interests

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

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Dan Whitten

References

Allen

Zayed

(2022) Hate crime statistics. Available at: https://researchbriefings.files.parliament.uk/documents/CBP-8537/CBP-8537.pdf (Accessed 9th August 2023).

Anders

Christopher

(2011) A socioecological model of rape survivors’ decisions to aid in case prosecution. Psychology of Women Quarterly 35(1): 92–106.

Andrade

(2015) Understanding relative risk, odds ratio, and related terms: as simple as it can get. Journal of Clinical Psychiatry 76(7): e857–e861.

Ariel

Bland

Sutherland

(2022) Experimental Designs. Sage.

Bacon

May

Charlesford

(2021) Understanding public attitudes to hate: developing and testing a UK version of the hate crime beliefs scale. Journal of Interpersonal Violence 36(23–24): NP13365–NP13390.

Barrett

Fletcher

Patel

(2014) Black minority ethnic communities and levels of satisfaction with policing: findings from a study in the north of England. Criminology and Criminal Justice 14(2): 196–215.

BBC (2021) Greater Manchester plan to tackle abuse of women unveiled. Greater Manchester plan to tackle abuse of women unveiled - BBC News (Accessed 2nd July 2024).

Bowling

(1999) Violent Racism: Victimization, Policing, and Social Context. USA: Oxford University Press.

Bowling

Parmar

Phillips

(2008) Policing minority ethnic communities. Handbook of policing 2: 611–641.

10.

Brunt

(2023) ‘Hanging by a thread’: public trust in police has fallen to its lowest level ever. In: Watchdog Warns. Sky News.

11.

Burman

Brooks-Hay

(2021) Delays in Trials: The Implications for Victim-Survivors of Rape and Serious Sexual Assault: an Update. The Scottish Centre for Crime & Justice.

12.

Chakraborti

(2009a) A glass half full? Assessing progress in the policing of hate crime. Policing: Journal of Policy Practice 3(2): 121–128.

13.

Chakraborti

(2009b) Crimes against the “other”: conceptual, operational, and empirical challenges for hate studies. Journal of Hate Studies 8: 9.

14.

Chakraborti

(2014) Framing the boundaries of hate crime 1. In: The Routledge International Handbook on Hate Crime. Routledge, pp. 13–23.

15.

Chakraborti

(2018) Responding to hate crime: escalating problems, continued failings. Criminology and Criminal Justice 18(4): 387–404.

16.

Chan

Ghose

Seamans

(2016) The internet and racial hate crime. MIS Quarterly 40: 381–404.

17.

College of Policing (2015) Second responder programmes to prevent domestic abuse. Available at: https://www.college.police.uk/research/crime-reduction-toolkit/second-responder-programmes (Accessed 2nd July 2024).

18.

College of Policing (2020). Responding to hate. Responding to hate | College of Policing (Accessed 2nd July 2024).

19.

Cook

Sackett

(1995) The number needed to treat: a clinically useful measure of treatment effect. BMJ 310(6977): 452–454.

20.

Corcoran

Lader

Smith

(2016) Hate crime, England and Wales. Statistical Bulletin 5: 15.

21.

CPS (2022) Hate crime. https://www.cps.gov.uk/crime-info/hate-crime (Accessed 9th August 2023).

22.

Cretney

Davis

(1997) Prosecuting domestic assault: victims failing courts, or courts failing victims? The Howard Journal of Criminal Justice 36(2): 146–157.

23.

Dalton

Barrett

Horvath

, et al. (2022) A systematic literature review of specialist policing of rape and serious sexual offences. International Criminology 2(3): 230–252.

24.

Davies

Westcott

(2006) Preventing withdrawal of complaints and psychological support for victims. Practical psychology for forensic investigations and prosecutions 25: 183–202.

25.

Davis

Weisburd

Taylor

(2008) Effects of second responder programs on repeat incidents of family abuse: a systematic review. Campbell Systematic Reviews 4(1): 1–38.

26.

del Pozo

Belenko

Pivovarova

, et al. (2024) Using Implementation Science to Improve Evidence-Based Policing: An Introduction for Researchers and Practitioners. Police Quarterly.

27.

Ellenberg

(1996) Intent-to-treat analysis versus as-treated analysis. Drug Information Journal 30(2): 535–544.

28.

Farrell

Lockwood

(2023) Addressing hate crime in the 21st Century: trends, threats, and opportunities for intervention. Annual Review of Criminology 6: 107–130.

29.

Faul

Erdfelder

Buchner

, et al. (2009) Statistical power analyses using G* Power 3.1: tests for correlation and regression analyses. Behavior Research Methods 41(4): 1149–1160.

30.

Fixsen

Naoom

Blase

, et al. (2005) A conceptual view of implementation’and ‘conclusions and recommendations. Implementation Research: A Synthesis of The Literature 11(22): 67–79.

31.

Garland

(2012) Difficulties in defining hate crime victimization. International Review of Victimology 18(1): 25–37.

32.

Garland

Funnell

(2016) Defining hate crime internationally: issues and conundrums. In: The Globalisation of Hate. Internationalising Hate Crime, pp. 15–30.

33.

Garland

Hodkinson

(2014) F** king freak! What the hell do you think you look like?’ Experiences of targeted victimization among goths and developing notions of hate crime. British Journal of Criminology 54(4): 613–631.

34.

Gordon

(2018) Chi- square test. In: Frey

(ed) The Sage Encyclopedia of Educational Research, Measurement, and Evaluation. Thousand Oaks, CA: Sage Publications, pp. 269–271.

35.

Grogger

(1991) Certainty vs. severity of punishment. Economic Inquiry 29(2): 297–309.

36.

Gyani

Chappell

Tebbutt

, et al. (2023) Using a randomised controlled trial to test the effectiveness of a simplified notice of police bail. Cambridge Journal of Evidence-Based Policing 7(1): 3.

37.

Hagerlid

Granström

(2023) Hate crime investigation and sentencing in Sweden: what have we learned in the past 20 Years? European Journal on Criminal Policy and Research: 1–18.

38.

Hall

(2005) Hate Crime: Crime and Society Series. Cullompton: Willian Publishing.

39.

Hanes

Machin

(2014) Hate crime in the wake of terror attacks: evidence from 7/7 and 9/11. Journal of Contemporary Criminal Justice 30(3): 247–267.

40.

Hansen

Ravn

Hansen

(2022) Recorded withdrawal from the police investigation among victims of rape: a mixed method approach to identifying case characteristics and police documented reasons. Nordic Psychology 74(1): 3–15.

41.

Hernán

Robins

(2017) Per-protocol analyses of pragmatic trials. New England Journal of Medicine 377(14): 1391–1398.

42.

Hester

Lilley

(2015) More than Support to Court: ISVAs in Teeside. Bristol: University of Bristol in association with the Northern Rock Foundation.

43.

Hester

Lilley

(2018) More than support to court: rape victims and specialist sexual violence services. International Review of Victimology 24(3): 313–328.

44.

HMICFRS (2021) Police response to violence against women and girls Final inspection report. Available at: https://assets-hmicfrs.justiceinspectorates.gov.uk/uploads/police-response-to-violence-against-women-and-girls-final-inspection-report.pdf (Accessed 2nd July 2024).

45.

Hohl

Stanko

(2015) Complaints of rape and the criminal justice system: fresh evidence on the attrition problem in England and Wales. European Journal of Criminology 12(3): 324–341.

46.

Home Office (2022) Hate crime, England and Wales, 2021 to 2022. Available at: https://www.gov.uk/government/statistics/hate-crime-england-and-wales-2021-to-2022/hate-crime-england-and-wales-2021-to-2022 (Accessed 9th August 2023).

47.

Home Office (2023) Hate crime, England and Wales, 2022 to 2023 (2nd ed). Available at: https://www.gov.uk/government/statistics/hate-crime-england-and-wales-2022-to-2023/hate-crime-england-and-wales-2022-to-2023 (Accessed 8th July 2024).

48.

Hudson

Paterson

(2020) Implementing police-led responses to hate crime: a case study of one English northern town. British Journal of Community Justice 16(1): 21–40.

49.

IACP (1999) Responding to hate crimes:A police officer's guide to investigation and prevention. Available at: https://files.eric.ed.gov/fulltext/ED442024.pdf (Accessed 2nd July 2024).

50.

Iganski

(2001) Hate crimes hurt more. American Behavioral Scientist 45(4): 626–638.

51.

Iganski

Kielinger

Paterson

(2005) Hate Crimes against London’s Jews: An Analysis of Incidents Recorded by the Metropolitan Police Service 2001-2004. Institute for Jewish Policy Research.

52.

Iganski

Smith

Dixon

, et al. (2011) Rehabilitation of hate crime offenders: research report. Scotlands: Equality and Human Rights Commission.

53.

Innes

(2014) Signal crimes. Social Reactions to Crime, Disorder, and Control.

54.

Kielinger

Paterson

(2007) Policing hate crime in London. American Behavioral Scientist 51(2): 196–204.

55.

King

Growette Bostaph

(2024) Victim participation in sexual assault cases: the importance of investigative actions and victim support. Crime & Delinquency: 00111287241264215.

56.

King

Sutton

(2013) High times for hate crimes: explaining the temporal clustering of hate-motivated offending. Criminology 51(4): 871–894.

57.

Lacey

(2012) Significance level, concept of. In: Salkind

(ed) Encyclopedia of Research Design. Thousand Oaks CA: Sage Publications, p. 1367.

58.

Lee

Ellenberg

Hirtz

, et al. (1991) Analysis of clinical trials by treatment actually received: is it really an option? Statistics in Medicine 10(10): 1595–1605.

59.

Lloyd

(2022) Miller V College of Policing: Social Media, ‘non-Crime Hate Incidents’, and the Right to Freedom of Expression. Mountford Chambers - London Barristers Chambers (Accessed 2nd July 2024).

60.

Longstaff

Willer

Chapman

, et al. (2015) Neighbourhood Policing: Past, Present and Future. The Police Foundation, pp. 25–31.

61.

Ludwig

McLean

Norton

(2017) Measuring Police Effectiveness. Chartered Institute of Public Finance & Accountancy.

62.

Madon

Murphy

Sargeant

(2017) Promoting police legitimacy among disengaged minority groups: does procedural justice matter more? Criminology and Criminal Justice 17(5): 624–642.

63.

Maxwell

Garner

Fagan

(2002) The preventive effects of arrest on intimate partner violence: research, policy and theory. Criminology & Public Policy 2(1): 51–80.

64.

McCoy

(2017) Understanding the intention-to-treat principle in randomized controlled trials. Western Journal of Emergency Medicine 18(6): 1075–1078.

65.

McFadzien

Pughsley

Featherstone

, et al. (2020) The evidence-based investigative tool (EBIT): a legitimacy-conscious statistical triage process for high-volume crimes. Cambridge Journal of Evidence-Based Policing 4: 218–232.

66.

McGuire

Evans

Kane

, et al. (2021) Victim and witness retraction and disengagement: a systematic review of contributing factors. Evidence-based policing and community crime prevention: 313–359.

67.

McHugh

(2018) Cramér’s V coefficient. In: Frey

(ed) The Sage Encyclopedia of Educational Research, Measurement, and Evaluation. Thousand Oaks, CA: Sage Publications, Inc, 417–418.

68.

McPhee

Hester

Bates

, et al. (2022) Criminal justice responses to domestic violence and abuse in England: an analysis of case attrition and inequalities using police data. Policing and Society 32(8): 963–980.

69.

Murphy

Hine

Yesberg

, et al. (2022) Lessons from London: A contemporary examination of the factors affecting attrition among rape complaints. Psychology, Crime and Law 28(1): 82–114.

70.

Nagin

(2013) Deterrence in the twenty-first century. Crime and Justice 42(1): 199–263.

71.

Olajide

(2022) The state of hate: trends in hate crime over the past decade. Available from: https://www.crestadvisory.com/post/the-state-of-hate-trends-in-hate-crime-over-the-past-decade.

72.

Petersen

Davis

Weisburd

, et al. (2022) Effects of second responder programs on repeat incidents of family abuse: an updated systematic review and meta‐analysis. Campbell systematic reviews 18(1): e1217.

73.

Pezzella

Fetzer

Keller

(2019) The dark figure of hate crime underreporting. American Behavioral Scientist: 000276421882384.

74.

Richards

Harinam

(2020) Tracking police arrests of intimate partner domestic abuse suspects in London: A situational factors analysis. Cambridge Journal of Evidence-Based Policing 4: 103–113.

75.

Robinson

(2019) Punishing dangerousness: cloaking preventive detention as criminal justice. In: Deterrence. Routledge, pp. 234–262.

76.

Ross

Sebire

Strang

(2022) Tracking repeat Victimisation after domestic abuse cases are heard with and without independent domestic violence advisors (Idvas) in an English magistrate’s court. Cambridge Journal of Evidence-Based Policing 6(1): 54–68.

77.

Safelives (2024) What is an IDVA? Available at: https://safelives.org.uk/about-domestic-abuse/domestic-abuse-response-in-the-uk/ (Accessed 2nd July 2024).

78.

Sherman

Berk

(1984) The specific deterrent effects of arrest for domestic assault. American Sociological Review 49(2): 261–272.

79.

Shirlow

Taylor

Merrilees

, et al. (2013) Hate crime: record or perception? Space and Polity 17(2): 237–252.

80.

Sleath

Smith

(2017) Understanding the factors that predict victim retraction in police reported allegations of intimate partner violence. Psychology of violence 7(1): 140–149.

81.

Smith

Coffman

Hudgens

(2021) Interpreting the results of intention-to-treat, per-protocol, and as-treated analyses of clinical trials. JAMA 326(5): 433–434.

82.

Trickett

Hamilton

(2016) Hate crime training of police officers in Nottingham: a critical review. Nottingham Trent University.

83.

James

(2012) Power. In: Salkind

(ed) Encyclopedia of Research Design. Thousand Oaks CA: Sage Publications, pp. 1066–1077.

84.

Walters

Brown

Wiedlitzka

(2016) Preventing Hate Crime: Emerging Practices and Recommendations for the Improved Management of Criminal Justice Interventions. Brighton: University of Sussex.

85.

Walters

Wiedlitzka

Owusu-Bempah

, et al. (2017) Hate crime and the legal process: options for law reform. Available at SSRN 3427984.

86.

Williams

Tregidga

(2013) All Wales hate crime research project. Available from: https://orca.cardiff.ac.uk/id/eprint/60690/13/TimeforJustice-AllWalesHateCrimeProject.pdf (Accessed 1st July 2024).

87.

Williams

Tregidga

(2014) Hate crime victimization in Wales: psychological and physical impacts across seven hate crime victim types. British Journal of Criminology 54(5): 946–967.