Sage Journals: Discover world-class research

Abstract

The COVID-19 pandemic and associated lockdowns have provided an unprecedented opportunity to study how such situational factors affect police recorded domestic abuse. This article presents findings from a large, representative study of the effect of the introduction and lifting of lockdowns on the volume and nature of domestic abuse recorded by seven English police forces within the first 12 months of the pandemic. The results suggest that lockdowns and the pandemic context did not create the domestic abuse crisis, and that the crisis does not go away when lockdown restrictions lift. Lockdowns interact with and amplify underlying patterns of domestic abuse. Notable differences between police forces suggest that local contexts and local police force practices play a role, with implications beyond pandemic contexts.

Keywords

COVID-19 domestic abuse pandemic policing

Introduction

The COVID-19 pandemic has put a public spotlight on the prevalence and harm caused by domestic abuse. At the onset of the pandemic media reporting and televised government announcements drew attention to how ‘stay at home’ restrictions or ‘lockdowns’ aimed at containing the spread of coronavirus had the unintentional consequence of trapping domestic abuse victims in their homes with their abuser, and how this was resulting in a worldwide surge in domestic violence.¹ Victim support organisations highlighted how lockdowns give domestic abusers opportunities to (further) isolate, control and abuse victims in their own home (End Violence Against Women Coalition, 2020). Victims had fewer options for safely seeking help. Lockdown also meant fewer opportunities for third parties, such as social workers, health workers, schools, family and friends to spot the abuse and potentially provide victims with support and routes to safety (Elliott et al., 2022). Various studies have attempted to measure the impact of lockdowns on domestic abuse, most commonly by using police administrative data (Kourti et al., 2023; Piquero et al., 2021).

Yet, police crime recorded domestic abuse cannot be taken to be representative of the extent of domestic abuse in the wider population. Barriers to reporting abuse to the police are multiple and even greater for victims from minoritised backgrounds, resulting in significant and disproportionate underreporting (Burman et al., 2004; Donovan and Hester, 2011; Gregory et al., 2021; Wolf et al., 2003). Consequently, domestic abuse coming to police attention is unlikely to be representative of the volume or nature of domestic abuse occurring within a population. The relationship between domestic abuse in the population and its recording in police data is further obscured by police failings to record a large proportion of crime in general (Her Majesty’s Inspectorate of Constabulary (HMIC), 2014a, 2014b) and particularly when it comes to domestic abuse (Myhill and Johnson, 2016). In sum, police recorded domestic abuse is an amalgam of offending behaviour in the population, (under-)reporting of domestic abuse to police and police crime recording practices. The three elements are difficult to disentangle within police data, and it is possible that the pandemic affected all three in different ways, further complicating any inference from police recorded domestic abuse to how COVID-19 lockdowns affected on domestic abuse in the population.

The focus of this article is on police recorded domestic abuse without attempting to draw inferences about the nature and extent of potential lockdown effects on domestic abuse in the wider population. Police recorded abuse, and the potential effects of the pandemic upon it, is worthy of studying in its own right. The domestic abuse coming to police attention is the abuse for which some form of police action has been requested, and a police intervention can take place. The most obvious form of police intervention is a criminal investigation, though this is not always wanted by the victim, or appropriate (Hoyle and Sanders, 2000). Police action may be limited to an immediate intervention to stop an attack in progress, so-called ‘positive action’. Positive action can consist of an arrest, removal of the abuser from the address, or other forms of safeguarding. Tracking the volume of domestic abuse coming to police attention, including seasonal, current events or pandemic related volume changes, provides data to argue for sufficient police resources to meet that demand. Understanding the nature of police recorded domestic abuse is also essential in designing effective police interventions responsive to the diverse needs and circumstances of victims from all backgrounds and contexts. Furthermore, during COVID-19 lockdowns, the police were the only domestic abuse safeguarding service consistently providing a face-to-face service, with social services and third sector support organisations required to adhere to ‘stay at home’ rules and ceasing all in-person support for some or all lockdowns (Bates and Struthers, 2022). Understanding police recorded abuse during the pandemic is thus of particular interest.

This article uses a census sampling approach to include all domestic abuse related crimes recorded in seven English police forces for the period of March 2018 to March 2021 to test the impact of the start and lifting of COVID-19 lockdowns on the volume and nature of police recorded abuse. It further uses data from all Domestic Abuse Stalking and Harassment (DASH) risk assessments carried out with domestic abuse victim in two of the police forces. The data allow measuring the lockdown impact on victim–survivor experience of the abuse, as disclosed to police. The remainder of this article has four sections. The first reviews the existing literature on police recorded domestic abuse during the COVID-19 pandemic and places this into the context of a wider understanding of domestic abuse police recording practices. The second section explains the empirical study and the third reports the results. The final section concludes with a discussion of the findings and their implications for our understanding of police recorded domestic abuse.

Background

The effect of COVID-19 restrictions on the volume of domestic abuse is unclear

Empirical studies of the impact of COVID-19 restrictions on domestic abuse published to date are inconclusive. Most police data based studies find the initial COVID-19 lockdown weeks to be associated with an increase in domestic abuse, though others evidence a decrease or find no significant change (see Kourti et al., 2023; Piquero et al., 2021, for systematic reviews). Leslie and Wilson (2020) estimate a 7.5% increase in calls for police service from March to May 2020 in 14 large US cities. Most of the increase was concentrated in the first 5 weeks of lockdown, and comparable to increases typically observed on ‘a home team upset or hot day’, that is, the spikes typically recorded after large sporting events or during the summer months when domestic abuse reports reach a seasonal high (Leslie and Wilson, 2020: 7). McCrary and Sanga (2021), using a slightly different set of US cities, shorter study time frame and different statistical model found a 16% increase in the first 4 weeks of lockdown and a return to the predicted pre-lockdown linear trend by the end of April 2020. Perez-Vincent et al. (2021) analysed calls for service to Buenos Aires Police, Argentina, within the first 6 weeks of the first national lockdown and compared it to victimisation survey data. They found calls for service increased by 28% following the introduction of mobility restrictions, with reports of psychological violence increasing the most. In contrast, other studies have concluded lockdowns had not significantly changed the volume of domestic abuse (Campedelli et al., 2021; Payne et al., 2022). Yet, Campedelli et al. (2021) only analysed police crime data from the initial weeks of COVID-19 restrictions in Los Angeles before full lockdown came into effect. Payne et al.’s (2022) Australian study is limited to data on breaches of domestic violence orders, accounting for a very small proportion of domestic abuse, and as such, is not representative of all police recorded domestic abuse.

In England, Ivandić et al. (2020) found a statistically significant increase in police calls for service for part of the lockdown, but no significant increase in domestic abuse related crime. Furthermore, once separating between calls for service made by domestic abuse victims and calls made by third parties (e.g. neighbours) they concluded that their observed increase in calls for service was entirely driven by increased third party reporting, particularly in areas of high population density – Calls from victims themselves had in fact decreased in the first weeks of lockdown. Leslie and Wilson (2020) concluded the opposite in their US-based study, here third party reporting did not account for their observed increase, nor did any socio-demographic variables. Ivandić et al.’s (2020) study further found distinguishing between current intimate partner, ex-partner and family abuse revealed a statistically significant 8% increase in current partner abuse was off-set by a 11% decrease in ex-partner abuse, giving the appearance of a ‘null-effect’ at the aggregate level. This finding points to the potential pitfalls of analysing highly aggregated data that do not distinguish between reports made by the victim or someone else, and averages over current and ex-partner abuse.

During lockdowns, opportunities to abuse may be vastly different for perpetrators locked down with the victim in the same household, compared to ex-partner perpetrators where lockdown may have reduced physical access to the victim. Similarly, in a lockdown context, victims of current abuse may have had fewer opportunities for safely contacting the police than victims of ex-partner abuse (Boxall and Morgan, 2021; Gregory and Williamson, 2022; Pfitzner et al., 2022).

In sum, the majority of quantitative studies find lockdowns to be associated with an increase in calls for police recorded domestic abuse, others find a decrease, or no statistically significant effect. There is some evidence of a differential impact of lockdowns on third party and victim calls for police service, though the direction of effects is inconsistent between studies and to date, there is little replication of such sub-group analyses. Importantly, existing studies are typically limited to the first weeks of the pandemic, and only one (Ivandić et al., 2020) disaggregates the data to distinguish between current and ex-partner abuse contexts, despite their immediate relevance within a lockdown context. Existing studies of the impact of COVID-19 lockdowns on domestic abuse generally differ in national context, the type of data used for the analysis, analytical approach, and the overall picture that emerges is inconclusive. Specifically, to date, there exists no large-scale study of police recorded domestic abuse in England and Wales that compasses data from all COVID-19 lockdowns, tests the replicability of lockdown effects from one lockdown to another, and distinguishes between current and ex-partner contexts.

Police data provide a particular lens on domestic abuse

Police data, both calls for service and crime records, form the basis of the majority of studies on the impact of COVID-19 lockdowns on domestic abuse published to date (Kourti et al., 2023; Piquero et al., 2021). Police data provide a picture of the volume and nature of domestic abuse for which a police intervention has been sought by a victim, a third party or even an officer, for example, when domestic abuse comes to light as part of another police interaction (see Lovett et al., 2022). Yet, most domestic abuse is not coming to police attention. Data from the Crime Survey for England and Wales self-completion module on intimate partner violence suggests that only 21% of victims report the abuse to the police (Office for National Statistics (ONS), 2016). The survey findings further suggest the most common reasons for not reporting are the victim believing the abuse was ‘too trivial’ or ‘not worth reporting’ (43%), domestic abuse being perceived as a ‘private matter and not the business of the police’ (37%) and believing the police could not help (25%; ONS, 2016). Bates et al. (2021) in their analysis of domestic homicides and suspected victim suicides study during the pandemic found that victims from Black, Asian and minority ethnic (BAME) groups were less likely to have been previously known to police or other agencies, suggesting greater barriers to reporting for ethnic minority victims. This also means that domestic abuse experienced by victims from minoritised backgrounds is under-represented in police data. Police under-recording further reduces the visibility of domestic abuse in police statistics. Myhill and Johnson’s (2016) study shows officers applying considerable discretion when deciding whether to record domestic abuse, resulting in the under- and mis-recording of domestic abuse (see also Hoyle, 1998). Poor data recording practice in England and Wales extends beyond domestic abuse related crimes, and has resulted in police crime statistics having its National Statistics designation removed by the Office for Statistics Regulation in 2014, and yet to be regained.² Consequently, police data do not adequately capture the full extent and the nature of domestic abuse perpetrated and experienced within a population. Police data however provide a picture of the volume and nature of domestic abuse to which police are responding.

COVID-19 restrictions made living with domestic abuse and help-seeking more difficult

For many victims, lockdowns made ongoing abuse worse. Qualitative studies showing the impact of lockdowns on the nature and lived experience of domestic abuse suggest that lockdowns were associated with an escalation of domestic abuse, with some perpetrators exploiting lockdowns to increase, diversify or mask the abuse (Brodie et al., 2023). Victims had to navigate increased fear for their safety and additional barriers to help-seeking, particularly when being locked down with their abuser (Boxall and Morgan, 2021; Gregory and Williamson, 2022; Pfitzner et al., 2022). This finding chimes with Ivandić et al. (2020) finding that victims’ calls to police had decreased, once victim calls for service were analysed separately from third party calls. This highlights the need to distinguish between current partners and ex-partners, as ex-partners may be less likely to be co-habiting with the victim during lockdowns. Moore et al.’s (2021) study of police referrals to a specialist domestic abuse service found that while the overall volume of referrals remained stable, there was an increase in high-risk cases, suggesting that during lockdowns victims might only be seeking help in the most severe circumstances. Lyons and Brewer (2021) analysed online forum posts written by female victim–survivor of male perpetrated intimate partner abuse. Their findings suggest that not only did victims experience an increase in the severity and frequency of the abuse, but the general distress and psychological harm caused by the abuse was heightened because of the pandemic and sense of no escape from abuse during lockdowns (Lyons and Brewer, 2021: 5). In sum, qualitative studies appear to be more consistent than quantitative studies in their findings and conclusions about how lockdown has affected victims’ lived experience of domestic abuse and help-seeking behaviour. They suggest that the lockdown context has resulted in an increase in the severity of domestic abuse, and a decrease in help-seeking, with victims placing a higher threshold on themselves for when to attempt to seek help from police or others.

This article expands the existing evidence base by addressing the following three research questions:

Can we generalise ‘lockdown effects’ across all three national lockdowns in 2020–2021 and across police force areas within the same jurisdiction (England)?

Have lockdowns affected differently on police recorded domestic abuse perpetrated by current partners and ex-partners?

Have lockdowns affected victim’s assessments of the severity of domestic abuse as disclosed to police?

To address the research questions, this study uses all police recorded domestic abuse crime recorded by seven English police forces between 2018 and 2021 and victim responses to all DASH risk assessments completed in two police force areas recorded between 2019 and 2021.

Methodology

Data

In England and Wales, domestic abuse is not a criminal offence in its own right. The exception is the ‘controlling or coercive behaviour’ offence introduced in 2015. However, the use of the offence is yet to receive significant uptake in police practice (see Brennan and Myhill, 2021). As per Home Office Crime Counting rules, domestic abuse related crimes should appear in police recorded data as a combination of the relevant general crime code (e.g. common assault, grievous bodily harm, stalking, harassment, homicide, to name a few) and a so-called ‘domestic abuse flag’ applied to the crime record to indicate that the crime took place between parties in some form of domestic relationship (Home Office, 2023). Domestic relationships include current intimate partners, former intimate partners and relatives, including immediate as well as extended family members. The dataset in this study comprises all domestic abuse-flagged crimes recorded by seven English police forces between 1 March 2018 and 31 March 2021. After removal of duplicates and non-domestic crimes, the total sample size is 636,673 domestic abuse crimes. Because all domestic abuse crimes recorded within the study period are included, the data are representative of the seven police force areas for the study period. The seven forces areas vary in size from small to very large, urban to rural, and cover some of the most and least deprived areas in the country. Force areas also vary in local lockdown periods, as well as in levels of domestic abuse as a percentage of total crime recorded by that force.

Analytic approach and variables

Replicating the empirical strategy of Perez-Vincent et al. (2021) and Leslie and Wilson (2020), difference-in-differences (DiD) linear regression is used to estimate the impact of the start and lifting of each of the three national lockdowns occurring between March 2020 and March 2021 on police recorded domestic abuse.

DiD estimates the difference in the domestic abuse crime volume before and after the same day of the year (e.g. the start date of the first lockdown) within the treatment years (here 2020 and 2021) compared to the control years (here 2018 and 2019). Leslie and Wilson (2020) and Perez-Vincent et al. (2021) demonstrate how simply comparing the number of domestic abuse crimes immediately before and after the start of a lockdown without using a DiD approach results in substantial overestimation of lockdown effects (or in some instances, underestimation). This is the result of the confounding effect of the well-documented seasonal patterns and long-term trends in the reporting and recording of domestic abuse (ONS, 2020), rendering it necessary to control for such long term and seasonal patterns in the analysis. For a fuller discussion of DiD and its application to quasi-experimental research designs, see Wing et al. (2018).

Key variables and selection of lockdown dates

The dependent variable in this linear regression DiD model is the daily count of domestic abuse-flagged crime records within a given police force. In the first model, the key explanatory variable of interest is the interaction effect between the treatment group dummy variable (coded as 0 = control for the years 2018 and 2019 and 1 = treatment for the years 2020 and 2021) and the before and after treatment dummy variable for the first national lockdown coded as 0 = before for the days before the 23rd of March (1 January to 22 March) and coded as 1 = for 23rd March until the last day of the first national lockdown (3 July). Lockdown dates were taken from the Institute for Government (2021) documentation of official lockdown dates set by the UK government. While the 23rd of March 2020 was the official start of the first national lockdown, there is ambiguity over whether the English population already observed a de-facto lockdown in the 7 days preceding the official start date. On the evening of the 16th of March, the Prime Minister announced in a televised address ‘now is the time for everyone to stop non-essential contact and travel’. From the 17th of March onwards, many businesses immediately moved to a home working, restaurants closed, and Google mobility data recorded a significant majority of the population ceasing trips outside their residential area (Langton et al., 2021). To test the sensitivity of results to the empirical start date of lockdown-like public behaviour, the analysis was re-run specifying the start of lockdown to fall 2, 4 and 7 days earlier than the official lockdown start date. Estimates did not change statistically significantly or substantially, consequently the results reported in this article are based on the official lockdown dates.

Additional explanatory variables

The remaining explanatory variables in the model control for seasonal and long-term trends, all specified as categorical variables: ‘day-of-the-week’ (Monday to Sunday) controls for systematic differences in the number of crimes recorded on different days of the week. ‘Week-of-the-year’ (Weeks 1–52) controls for seasonal effects and a year variable (i.e. 2018, 2019, 2020, 2021) controls for longer term trends.

To test whether the first national lockdown affected differently on different types of domestic abuse the analysis was repeated specifying as the dependent variable the daily number of crimes recorded for current partners only, ex-partners only, and family members only. Finally, analyses were repeated specifying as dependent variable the percentage of daily crimes involving a female victim, and the percentage of daily domestic abuse crimes involving BAME.

All analyses were then repeated for the second national lockdown (5th November to 1st December 2020) and third national lockdown (4th January to study period end, 31st March 2021), as well as the lifting periods between these lockdowns, with dummy variable coded following the above logic. The third national lockdown was not fully lifted until after the study period ended (23rd June 2021), consequently the model is only an estimation of the effect of the third lockdown up to the end of our study period (i.e. 31st March 2021).

To complicate matters, legislation enabling local lockdowns was introduced on the 4th of July 2020 and a new three-tier system of COVID restrictions came into force on the 14th of October 2020. To account for local lockdowns and tier restrictions akin to lockdowns, separate models are estimated for each police force area, adjusting the start or end date of lockdowns where force areas entered local ‘Tier 3’ or ‘Tier 4’ ahead of national lockdowns, or immediately entered ‘Tier 3’ or ‘Tier 4’ restrictions after the end of national lockdowns.³ No force area in this study experienced local lockdowns not immediately preceded or followed by a period of national lockdown. Tier 3 restrictions banned households mixing indoors and restricted outdoor mixing to six persons, mandated working from home where possible, the closure of all hospitality and personal care businesses, and ruled against unnecessary travel. Tier 4 restrictions equated to full lockdown with the additional closure of all non-essential shops (Brown and Kirk-Wade, 2021). In three of the seven police force areas Tier 3 or Tier 4 restrictions were in place for the entire period between the second or third national lockdown. Consequently, the period between the start of the second lockdown and the end of the third lockdown was treated as one lockdown period within the analyses.

Finally, two of the seven forces (Force 2 and Force 3) provided the answers victims have given to officers in response to the DASH questionnaire, used as part of the mandatory risk assessment in domestic incidents in England and Wales. The DASH consists of a total of 27 questions about a range of aspects of potential past and ongoing abuse, how the abuse is experienced by the victim-survivor and its impact on to the victim-survivor, risk factors relating to the abuser, as well as range of situational risk factors, such as recent pregnancy, childbirth or recent separation (see Myhill and Hohl, 2019, for further details on the DASH). For the purposes of this article, four items from the DASH questionnaire were used to test the effect of lockdowns on potential escalation of the abuse as experienced by the victim, and the impact of the abuse on victim mental health. For purposes of this study only select items measuring these aspects, a comprehensive analysis is provided elsewhere (Johnson and Hohl, 2023).

The four DASH variables here are as follows:

Victim experience of escalation of the abuse in terms of its severity (‘Is the abuse getting worse?’) or frequency (‘Is the abuse happening more often?’). Both variables coded as ‘1’ if the officer recorded the victim to have responded ‘yes’ to this question and coded ‘0’ if the victim responded ‘no’, refused to answer the question, no answer was sought or recorded for the question. The dataset does not allow distinguishing between these reasons for a ‘0’ being recorded in the dataset (for a fuller discussion, see Myhill and Hohl, 2019).

Victim experience of the impact of the abuse coded as ‘1’ if the officer recorded the victim responding in the affirmative to the DASH question ‘Are you feeling depressed or suicidal?’ and coded ‘0’ otherwise.

Recent separation/attempts to separate from the abuser was coded ‘1’ if officer noted the victim disclosing that they had recently separated or attempted to separate from the abuse and coded ‘0’ otherwise. Separation is a known trigger of escalation (Monckton-Smith, 2021).

Findings

Figure 1 plots the domestic abuse volume over the 3-year study period. It is important to note that the volume of domestic abuse in the pre-pandemic years of 2018 and 2019 was already high and increasing, accounting for 1 in 10 police recorded crimes (Her Majesty’s Inspectorate of Constabulary Fire and Rescue Services (HMICFRS), 2019). The 2020 pandemic year follows the well-known seasonal pattern of higher volumes during the late spring and summer months, and a peak at Christmas.

Figure 1.

Domestic abuse crime in seven English police forces March 2018 to March 2021.

Yet, the summer increase and Christmas spike are far more pronounced in 2020 than in the pre-pandemic years 2018 and 2019, and the long-term increase in domestic abuse crime appears to be accelerating during the pandemic years 2020 and 2021. The year-on-year increase in police recorded domestic abuse crime has more than doubled since the first national lockdown (5.1%–14.6%, Table 1). One must be cautious in drawing conclusions about the impact of lockdowns on the volume of domestic abuse based on descriptive statistics alone. It risks erroneously attributing long-term trends and seasonal patterns to lockdowns and thereby over- or underestimating lockdown effects.

Table 1.

Descriptive statistics for domestic abuse crime dataset and DASH dataset 2018–2021.

	2018		2019		2020		2021
	1–22 March	23 March to 31 December	1 January to 22 March	23 March to 31 December	1 January to 22 March	23 March to 31 December	1 January to 22 March	23–31 March^a
All domestic abuse crimes (mean weekly count)	3301	3646	3700	3870	3888	4361	4457	3693
% change to same period previous year	–	–	–	6.1	5.1	12.7	14.6	–
Relationship profile
Current intimate partner (%)	34.1	34.3	35.1	34.3	34.0	34.3	34.1	–
Ex-intimate partner (%)	37.3	37.5	36.5	36.0	35.6	34.0	34.4	–
Family member (%)	28.6	28.2	28.4	29.7	30.4	31.7	31.5	–
Demographic profile
Female victim (%)	75.1	75.9	75.8	75.3	74.8	73.9	73.2	–
Male suspect (%)	77.1	77.7	77.5	76.9	76.4	76.0	74.0	–
Mean age victim	36.0	35.8	35.8	36.3	36.5	36.5	36.7	–
Mean age suspect	34.3	34.3	34.4	34.6	34.9	35.1	35.1	–
BAME victim (%)	7.4	8.0	8.0	8.1	8.4	8.1	7.9	–
BAME suspect (%)	9.0	9.5	9.4	9.4	9.6	9.2	9.2	–
DASH risk assessment forms (mean weekly count)	949	958	948	981	938	1100	1017	1072
% change to same period previous year	–	–	–	2.4	–1.0	12.1	8.4	–
Recent separation/attempt to separate	36.3	34.7	34.7	35.9	35.4	31.5	30.1	–
Abuse increasing in severity	17.2	15.6	15.0	15.8	16.1	16.5	18.1	–
Abuse increasing in frequency	17.8	16.8	16.7	17.1	17.9	17.8	19.3	–
Abuser excessive jealous or controlling	15.3	14.9	14.5	14.2	14.9	14.1	15.0	–
Victim feeling depressed or suicidal	12.3	12.1	12.9	13.5	13.9	13.4	14.7	–

BAME: black, Asian or minority ethnic; DASH: domestic abuse stalking and harassment.

DASH data only available for Forces 2 and 3.

Not comparable to the full 23 March to 31 December period due to strong seasonal patterns.

Figure 1 further suggests a temporary dip in domestic abuse crime volume in the weeks on either side of the official start of the first national lockdown on the 23rd of March 2020, yet this pattern is not replicated in the second and third national lockdowns. There is a noticeable rise in volume during the months of June and July when the first lockdown was gradually eased with first schools reopening (2nd June) followed by non-essential shops (15th June) and the Prime Ministers’ 23rd of June announcement of most lockdown restrictions coming to an end by the start of July. This period of stepwise lockdown easing coincided with the typical period of a seasonal summer uptick in domestic abuse, making necessary the disentangling of seasonal effects from potential lockdown effects.

Table 2 shows the results of the DiD regressions estimating the impact of lockdowns on domestic abuse crime, controlling for year (annual trends), week-of-the-year (seasonal patterns) and day-of-the-week (weekday patterns).

Table 2.

Difference in differences regression ‘lockdown effect’ estimates.

	Force 1			Force 2			Force 3			Force 4			Force 5			Force 6			Force 7
	b***	beta	R-squared	b***	beta	R-squared	b***	beta	R-squared	b***	beta	R-squared	b***	beta	R-squared	b***	beta	R-squared	b***	beta	R-squared
Start lockdown 1
All domestic abuse	0.015	0.001	20.7%	−3.128*	−0.135	29.3%	5.463*	0.186	20.3%	−3.639	−0.110	44.4%	6.133	0.108	40.4%	−10.438*	−0.403	46.6%	−2.388	−0.036	77.9%
Intimate partner abuse	−1.437	−0.138	13.7%	–	–	–	4.907*	0.144	32.3%	−3.147	−0.131	26.9%	1.312	0.032	30.8%	–	–	–	−1.406	−0.030	77.1%
Current partner abuse	0.456	0.068	17.3%	–	–	–	–	–	–	1.167	0.263	29.7%	4.652*	0.147	34.8%	–	–	–	−2.821*	−0.091	70.6%
Ex-partner abuse	−1.893**	−0.232	11.6%	–	–	–	–	–	–	−4.315***	−0.238	29.0%	−3.341*	−0.135	23.9%	–	–	–	1.415	0.064	56.6%
Family abuse	1.596*	0.153	16.6%	–	–	–	2.666**	0.190	25.3%	−0.767	−0.052	51.7%	6.569***	0.287	35.5%	–	–	–	0.369	0.017	61.1%
% female	−0.041	−0.125	9.0%	0.006	0.032	8.3%	−0.007	−0.052	10.4%	0.000	−0.001	9.9%	−0.009	−0.113	15.2%	2.058	0.439	17.9%	−1.226	−0.097	14.8%
% BAME	−0.002	−0.046	6.4%	−0.002	−0.036	10.2%	−0.005	−0.095	9.7%	0.005	0.076	7.5%	0.022**	0.244	13.4%	–	–	–	−1.351	−0.090	14.5%
Risk grading
High	−1.460**	−0.205	9.8%	−0.160	−0.023	8.7%	0.974	0.070	54.2%	−0.384	−0.053	15.8%	−0.305	−0.038	20.3%	–	–	–	−1.185	−0.092	13.6%
Medium	3.493***	0.284	26.9%	−3.471**	−0.180	30.9%	2.375**	0.128	57.9%	0.543	0.045	35.7%	−2.916*	−0.110	−11.0%	–	–	–	−1.591	−0.053	65.9%
Standard	−2.019**	−0.164	45.2%	0.110	0.020	9.8%	4.187*	0.120	42.5%	−0.122	−0.007	49.7%	9.378**	0.210	35.6%	–	–	–	0.414	0.011	71.7%
Lifting lockdown 1
All domestic abuse	0.936	0.049	28.1%	−4.673**	−0.180	34.3%	16.312***	0.577	27.7%	−3.785	−0.106	46.4%	14.732***	0.218	48.5%	−0.574	−0.021	44.2%	9.442***	0.124	79.9%
Intimate partner abuse	−0.701	−0.063	19.0%	4.525*	0.129	44.0%	−1.218	−0.045	27.9%	7.083*	0.149	39.0%	–	–	–	7.457***	0.139	78.4%
Current partner abuse	0.385	0.054	21.1%	–	–	–	–	–	–	1.810	0.116	33.1%	2.444	0.068	42.5%	–	–	–	4.333**	0.122	72.4%
Ex-partner abuse	−1.086	−0.137	12.1%	–	–	–	–	–	–	−3.028*	−0.145	25.8%	4.638*	0.171	28.4%	–	–	–	3.124*	0.129	60.4%
Family abuse	1.302	0.122	24.0%	–	–	–	4.206***	0.309	30.4%	−2.593**	−0.172	53.4%	6.777***	0.276	37.8%	–	–	–	3.276**	0.135	63.5%
% female	−0.015	−0.047	10.6%	0.006	0.027	7.0%	−0.027**	−0.203	16.4%	−0.005	−0.036	13.7%	−0.006	−0.065	12.6%	−1.048	−0.180	16.6%	−0.628	−0.052	11.7%
% BAME	−0.003	−0.058	8.7%	−0.002	−0.041	10.1%	−0.009*	−0.165	9.8%	0.006	0.088	6.5%	0.015*	0.160	15.6%	–	–	–	−2.607*	−0.176	13.6%
Risk grading
High	−0.667	−0.092	8.8%	−0.630	−0.089	8.5%	2.640***	0.181	57.5%	0.059	0.008	12.5%	–0.896	–0.108	15.0%	–	–	–	2.166*	0.151	28.5%
Medium	3.373***	0.264	26.5%	−6.369***	−0.301	33.5%	1.890*	0.098	62.2%	0.568	0.044	39.9%	3.117	0.098	57.5%	–	–	–	1.633	0.048	68.5%
Standard	−1.770*	−0.134	46.3%	0.953***	0.148	18.3%	4.166*	0.122	53.6%	1.046	0.051	57.4%	12.519***	0.249	40.4%	–	–	–	5.629**	0.141	72.9%
Start lockdown 2
All domestic abuse	−2.957	−0.131	16.3%	−10.347	−0.470	31.5%	11.683	0.432	16.2%	−15.166***	−0.432	38.6%	2.362	0.041	34.9%	18.192**	0.666	32.4%	23.042***	0.268	85.3%
Intimate partner abuse	−3.735***	−0.280	11.0%	3.638	0.110	37.0%	−9.523***	−0.358	23.2%	1.610	0.038	30.5%	–	–	–	13.887**	0.230	83.6%
Current partner abuse	−1.420	−0.156	10.3%	–	–	–	–	–	–	−2.033	−0.121	32.4%	0.131	0.004	41.6%	–	–	–	14.026***	0.337	79.3%
Ex-partner abuse	−2.315**	−0.221	10.9%	–	–	–	–	–	–	–7.489***	−0.362	26.0%	1.479	0.063	19.3%	–	–	–	−0.139	−0.006	64.1%
Family abuse	0.621	0.046	19.7%	–	–	–	6.609	0.488	23.5%	−5.449***	−0.358	47.7%	0.958	0.044	21.5%	–	–	–	8.463**	0.308	70.4%
% female	−0.041	−0.125	9.0%	−0.019	−0.093	8.0%	0.069	0.497	18.2%	0.010	0.066	12.0%	0.003	0.039	7.9%	3.254*	0.540	24.9%	−0.984	−0.077	20.4%
% BAME	−0.002	−0.046	6.4%	0.023	0.415	8.8%	0.004	0.078	10.5%	0.009	0.121	6.6%	0.013	0.126	13.2%	–	–	–	0.801	0.055	12.9%
Risk grading
High	−1.460**	−0.205	9.8%	−0.676	−0.105	9.0%	4.106	0.276	56.4%	−1.080	−0.140	17.8%	−0.488	−0.057	16.7%	–	–	–	5.497*	0.359	32.5%
Medium	3.493***	0.284	26.9%	−11.495*	0.021	30.3%	−0.890	−0.051	56.8%	−1.393	−0.113	35.0%	−4.273*	−0.159	42.0%	–	–	–	5.400	0.140	73.9%
Standard	−2.019**	−0.164	45.2%	0.751	0.129	8.0%	7.074	0.206	49.9%	−3.180**	−0.172	46.5%	7.080*	0.164	31.6%	–	–	–	12.344**	0.281	78.7%
Lifting lockdown 2
All domestic abuse	−2.179	−0.097	21.1%	–	–	–	–	–	–	–10.265***	−0.216	44.1%	2.348	0.031	32.9%	–	–	–	–	–	–
Intimate partner abuse	−2.121*	−0.087	18.4%	–	–	–	–	–	–	−8.029**	−0.227	30.0%	−1.394	−0.026	24.6%	–	–	–	–	–	–
Current partner abuse	−0.779	−0.134	10.4%	–	–	–	–	–	–	–3.246*	−0.150	33.4%	−0.547	−0.013	36.4%	–	–	–	–	–	–
Ex-partner abuse	−1.342	−0.134	10.4%	–	–	–	–	–	–	−4.783**	−0.187	30.0%	−0.847	−0.027	15.2%	–	–	–	–	–	–
Family abuse	−0.006	0.000	21.6%	–	–	–	–	–	–	−2.153	−0.105	51.4%	2.478	0.086	31.4%	–	–	–	–	–	–
% female	−0.037	−0.084	8.3%	–	–	–	–	–	–	−0.030	−0.159	18.9%	−0.005	−0.048	11.7%	–	–	–	–	–	–
% BAME	−0.006	−0.078	9.7%	–	–	–	–	–	–	0.002	0.018	6.5%	0.009	0.072	11.4%	–	–	–	–	–	–
Risk grading
High	−0.953	−0.099	9.0%	–	–	–	–	–	–	−0.332	−0.036	16.0%	−0.254	−0.025	14.7%	–	–	–	–	–	–
Medium	0.427	0.026	18.8%	–	–	–	–	–	–	−2.521*	−0.167	39.7%	−3.934	−0.115	38.9%	–	–	–	–	–	–
Standard	−1.653	−0.125	29.8%	–	–	–	–	–	–	−3.352	−0.137	51.0%	6.096	0.109	30.0%	–	–	–	–	–	–
Lockdown 3 start
All domestic abuse	−2.304*	−0.082	24.1%	–	–	–	–	–	–	18.709***	0.373	37.6%	41.986***	0.475	44.9%	–	–	–	–	–	–
Intimate partner abuse	−0.098	−0.009	16.9%	–	–	–	–	–	–	4.737**	0.128	21.6%	14.831***	0.237	32.1%	–	–	–	–	–	–
Current partner abuse	−0.779	−0.087	18.4%	–	–	–	–	–	–	1.392	0.062	30.0%	−0.010	0.000	38.8%	–	–	–	–	–	–
Ex-partner abuse	−0.886*	−0.074	9.1%	–	–	–	–	–	–	3.344***	0.116	19.0%	14.840***	0.372	28.3%	–	–	–	–	–	–
Family abuse	−0.700	−0.043	18.2%	–	–	–	–	–	–	13.774***	0.593	48.0%	12.252***	0.359	32.5%	–	–	–	–	–	–
% female	−0.026	−0.055	7.9%	–	–	–	–	–	–	−0.056***	−0.264	11.5%	−0.070***	−0.516	25.6%	–	–	–	–	–	–
% BAME	−0.003	−0.035	5.5%	–	–	–	–	–	–	0.004	0.044	5.4%	−0.020***	−0.145	10.6%	–	–	–	–	–	–
Risk grading
High	−0.847	−0.079	6.6%	–	–	–	–	–	–	−3.135***	−0.280	16.7%	3.066***	0.248	15.1%	–	–	–	–	–	–
Medium	4.543***	0.234	22.3%	–	–	–	–	–	–	7.264***	0.404	36.7%	25.062***	0.593	50.9%	–	–	–	–	–	–
Standard	−6.000***	−0.310	42.2%	–	–	–	–	–	–	13.369***	0.465	49.5%	2.243	0.034	36.3%	–	–	–	–	–	–

BAME: Black, Asian or minority ethnic.

Control variables not displayed: year, day-of-the-week, week-of-the-year.

p < 0.05; **p < 0.01; ***p < 0.001.

The results suggest that lockdowns have affected domestic abuse recorded crime differently in different police force areas and had differential effects on different types of domestic relationships. In four of the seven police forces, the start of the first national lockdown did not result in a statistically significant change in the volume of domestic abuse crime recorded over the entire first lockdown period, once controlling for annual and seasonal effects. Only one force recorded a small statistically significant increase of an average 5.5 additional daily reports (p < 0.05) while two other forces saw a statistically significant negative effect of on average of 3.1 (p < 0.05) to 10.4 (p < 0.05) fewer daily domestic abuse crimes attributable to the start of the lockdown net of annual and seasonal effects. Within forces, the direction of the lockdown effect remained consistent across subsequent lockdowns in five of the seven forces. Exceptions are Force 4 (medium-sized force covering a large urban area with very high levels of deprivation as well as some rural areas) and Force 6 (small force covering urban and rural areas, including some with high levels of deprivation). In Force 4, the third lockdown resulted in a statistically significant increase of 18.7 additional daily reports (p < 0.001), while the second lockdown had resulted in a statistically significant decrease of 15.2 daily reports (p < 0.001). The first lockdown had no statistically significant effect. Force 6 observed the reverse pattern, with the first lockdown resulting in a statistically significant decrease of 10.4 fewer domestic abuse crimes recorded per day (p < 0.05), followed by a statistically significant 18.2 average increase in domestic abuse crime recording (p < 0.001).

Turning to the nature of police recorded domestic abuse, the results provide evidence of a differential impact of lockdowns on abuse by current partners, ex-partners and family members. In all but one force, the first national lockdown resulted in an increase in current partner abuse (0.5–4.9 additional daily reports) while ex-partner abuse decreased net of annual and seasonal effects (1.4–4.3 fewer daily reports). Family abuse also tended to increase during the first lockdown (1.6–6.6 additional recordings per day). Furthermore, in two out of three forces, the decrease in ex-partner abuse recording was roughly equal in size to the combined increase in the police recording of current partner and family abuse, resulting in the appearance of a statistical null effect of the lockdown on domestic abuse crime recording at the aggregate level. This finding underlines the importance of differentiating between current and ex-partner abuse, as well as family abuse, when considering how, and why, context may affect abusive relationships. The pattern repeated, albeit less pronounced, during the subsequent second lockdown: current ex-partner abuse reduced (b = –2.3 in Force 1 and b = –7.5 in Force 4, both p < 0.01) while current partner abuse remained unchanged. In Force 7, ex-partner abuse remained unchanged while current partner abuse increased significantly (b = 14.0, p < 0.001), controlling for annual and seasonal effects. In Forces 3 and 5, the second lockdown had no statistically significant effect on the volume or profile of police recorded domestic abuse for any available relationship variable. In the third lockdown, the pattern reversed. Ex-partner abuse increased (b = 3.3, p < 0.001 in Force 4 and b = 14.8, p < 0.001 in Force 5) while current partner abuse remained unchanged. Next, the impact of the lifting of lockdowns on police recorded domestic abuse was examined. The results suggest that the effect of the first national lockdown on police recorded domestic abuse crime continued after restrictions lifted, with an accelerating increase in Force 5 (b = 6.1, n.s. to b = 14.7, p < 0.001) and Force 7 (b = –2.4, not significant (n.s) to b = 9.4, p < 0.001), a continued decrease in recorded domestic abuse in Force 2 (b = –3.1, p < 0.05 to b = –4.7, p < 0.001), and plateauing in the remaining four forces. This pattern is replicated during the brief lifting of the second national lockdown. Finally, the demographic profile of police recorded domestic abuse, the gender and ethnic profile of victims did not change systematically.

Turning to changes in the abuse as experienced by victims and disclosed to police during the DASH interview, Table 1 shows a decline in recent separations or attempts to separate after the beginning of lockdowns (35.4% January to March 2020 to 30.1% January to March 2021), victims reporting the abuse increasing in severity (16.1%–18.1%) and frequency (17.9%–19.3%). Furthermore, there is an increase in victims feeling depressed or having suicidal thoughts (13.9%–14.7%), though this trend seems to predate the pandemic already rising from 12.3%–13.5% between early 2018 and early 2019.

Table 3 reports the results of DiD regression estimating the effect of lockdown starts and lifts on the DASH variables, net of annual and seasonal effects. In both police forces, the start of the first national lockdown resulted in a statistically significant increase in the number of DASHs completed (and additional 4.9 DASHs per day in Force 2 and 7.7 additional DASHs per day in Force 3), with the increase continuing after restrictions lifted, b = 14.1 (p < 0.00) in Force 2 and b = 8.6 (p < 0.01) in Force 3. As the pandemic continued to unfold and England entered a second national lockdown, victims disclosing the abuse happening more often (b = 4.0, p < 0.01) and disclosing the abuse is getting worse (b = 4.6, p < 0.01) increased substantially in Force 3, net of annual and seasonal effects. Victims disclosing feeling depressed or having suicidal thoughts also increased by 5.7 percent points (b = 5.7, p ⩽ 0.01). The findings were not replicated in Force 2.

Table 3.

Difference in differences regressions DASH data.

	Force 2			Force 3
	b***	beta	R–squared (%)	b***	beta	R–squared (%)
Total number of completed DASH forms
Lockdown 1	4.905***	0.216	40.8	7.675**	0.187	29.8
Lifting lockdown 1	14.108***	0.502	60.5	8.643***	0.209	42.5
Lockdown 2	9.456	0.418	48.5	6.161	0.163	33.8
Is the abuse happening more often?
Lockdown 1	−0.822	−0.142	16.6	0.15	0.024	11.4
Lifting lockdown 1	−0.654	−0.108	12.3	0.956	0.139	17.3
Lockdown 2	−2.886	−0.527	14.6	4.032*	0.576	14.7
Is the abuse getting worse?
Lockdown 1	−0.41	−0.076	14.7	−0.365	−0.049	16.7
Lifting lockdown 1	−0.241	−0.043	14.6	1.736	0.219	24.5
Lockdown 2	−2.589	−0.494	14.1	4.598*	0.575	18.2
Are you feeling depressed or having suicidal thoughts?
Lockdown 1	−0.629	−0.115	20.0	−2**	−0.194	20.5
Lifting lockdown 1	−0.541	−0.099	18.8	2.323**	0.218	23.1
Lockdown 2	−1.189	−0.22	19.3	5.66*	0.528	21.8
Have you separated or tried to separate from your abuser within the past year?
Lockdown 1	0.967	0.099	14.2	−0.013	−0.001	15.5
Lifting lockdown 1	0.484	0.045	18.3	0.52	0.025	19.6
Lockdown 2	0.191	0.02	15.3	−0.178	−0.009	16.9

DASH: domestic abuse stalking and harassment.

Control variables not display, year, day-of-the-week, week-of-the-year.

p < 0.05; **p < 0.01; ***p < 0.001.

The victim recently separating from the abuser or attempting to separate from the abuser is a known trigger of escalation of abuse (Monckton-Smith, 2021) and a precursor to 43% of domestic feminicides (Femicide Census, 2020). Figure 2 plots the weekly number of separations disclosed in the DASH risk assessment. The plot shows fewer victims separated in the first week of lockdown than during any other week over the study period in Force 3 (Force 3: 270 disclosed separations compared to 320 during the same week in 2019 and in 2020, 10 separations in the weeks 1–7 April compared to 13 during the same week the previous year), and spikes in separations towards the end of the first lockdown and immediately after the first lockdown at levels normally only seen during the first week of the New Year (Force 3: 446 separations in the weeks 1–7 July 2020 compared to 406 during the first week of January 2020; Force 2: 19 in the week 24–30 June 2020 compared to 14 in the first week of January 2020). This finding is indicative of lockdowns keeping victims in abusive relationships for longer, with separations delayed but not cancelled, and occurring when lockdown lifts are announced. However, the DiD regressions did not find a statistically effect of lockdowns on separations, possibly due to regression estimates averaging over the entirety of a lockdown (or between lockdown) period.

Figure 2.

Separations/attempts to separate from the abuser as disclosed in the DASH.

Discussion

In only three of the seven police force areas did lockdowns have a statistically significant effect on the volume of police recorded abuse. While one force recorded a statistically significant increase, the two others recorded a statistically significant decrease, once controlling for long-term trends and seasonal effects. Furthermore, effect sizes are small relative to the absolute volume of police recorded domestic abuse and relative to the 56% rise in domestic abuse related crime recording between 2017 and 2020 and 181% rise between 2014 and 2020 (HMICFRS, 2021: 9–10, though note the change in recording practices). Placing lockdown effects into this wider context makes evident that the pandemic did not create a domestic abuse crisis. Rather, the domestic abuse crisis predates the pandemic, with no evidence to suggest its ending when COVID restrictions ended.

The results further provide evidence that while earlier lockdowns were useful indicators of the effect of subsequent lockdowns in some force areas, there is limited or, and in some areas, no predictability of the impact of future lockdowns based on past lockdowns. This echoes findings of international meta-analyses of the impact of the pandemic on domestic abuse concluding that lockdowns were associated with an increase in domestic abuse in many jurisdictions, while also concluding inconsistency of effect directions and sizes between studies, local, and national contexts (Kourti et al., 2023; Piquero et al., 2021). The limited transferability of findings from one lockdown to another may be linked to the change in lockdowns as the pandemic progressed, even if their main characteristic – stay at home restrictions – was the same in all three national lockdowns. For example, schools closed during first and third national lockdown, but remained open during the second national lockdown. With each lockdown, further exemptions were introduced, such as the ability for single households to form a ‘support bubble’ with one other household, the introduction of outdoor mixing with one and then up to six persons from another household. Within weeks of the first lockdown, the government and public sector agencies sought to reassure domestic abuse victims through media and advertisement campaigns that breaking ‘stay at home’ restrictions to flee abuse was lawful, and that help remained available (e.g. free rail travel). Furthermore, public compliance with lockdown restrictions steadily reduced over the course of the pandemic (Ganslmeier et al., 2022).

Crucially, and in line with Ivandić et al. (2020) findings, the lockdown restrictions appear to have impacted differently on current and ex-partner abuse, with family and current partner abuse increasing during the first two lockdowns while ex-partner abuse decreased in some police forces, given the appearance of a ‘null effect’ of lockdowns when aggregating across relationship types. It may be speculated that ex-partner abuse reduced during the first and second lockdown due to fewer opportunities for in-person offending, much like other crime types (Langton et al., 2021). By the third national lockdown, lockdown habituation and declining compliance (see Ganslmeier et al., 2022) may have rendered lockdown rules ineffective barriers to ex-partner offending. Furthermore, one must be cautious in inferring a reduction in ‘true’ levels of ex-partner abuse from a reduction in police recorded ex-partner domestic abuse. Buil-Gil et al. (2021) demonstrate an increase in online crime during the pandemic that persists beyond the lifting of lockdowns. It is conceivable that ex-partner abuse may have similarly shifted online, for example, through social media facilitated or image-based sexual abuse (‘revenge porn’) both of which are rarely reported to the police (McGlynn et al., 2017).

Surprisingly, there is no evidence of a significant change in reporting from minority ethnic victims, despite the potential of interlocking equalities compounding the risk of domestic abuse. The adequacy of the DASH as a risk assessment tool, and frontline officer domestic abuse risk assessment is contested (Medina Ariza et al., 2016), however analysing victim-survivor responses to individual DASH questions provides some insight into how victim assess and experience the abuse. The significant increase in the number DASH forms completed during the first lockdown might be explained by the decline in all other forms of crime freeing officers up to respond to domestic abuse incidents quicker and more thoroughly than previously, with many officers acutely aware of the heightened pressures and dangers of domestic abuse during lockdowns (HMICFRS, 2021: 19). In one of the two police forces that provided DASH data, there is clear evidence of victim disclosing an escalation of the abuse (increasing in severity and/or frequency), with a corresponding statistically significant and substantial increase in victim disclosing feeling depressed or suicidal. This finding corroborates, in police data, victim-survivor research evidencing how abusers intensify the abuse with fewer opportunities for victims to seek help, and worsening impact on victim mental health (Boxall and Morgan, 2021; Elliott et al., 2022; Gregory and Williamson, 2022).

The findings further reveal differences between police forces in the effect of lockdowns on police recorded domestic abuse, and consequently, limited transferability of findings from one force area to another. This study does not permit pinpointing its cause(s). There are, to date, no other published studies that address this question directly. Studies in other areas provide tentative evidence of local variations in how the COVID-19 pandemic and lockdowns affected on a wide range of other aspects of peoples’ lives, including the risks and outcomes of financial hardship (Cross et al., 2022), propensity to access health care including a dramatic fall in Accidents & Emergencies (A&E) department visits (Thornton, 2020), and general mental well-being (Etheridge and Spantig, 2022). There is also emerging evidence of local variations in how police forces adapted their response to domestic abuse within the lockdown context which might, at least in part, explain the observed variation. Walklate et al. (2022) found some forces used local media and placed posters in supermarkets and pharmacies to raise awareness of domestic abuse and the availability of police. Forces also graded domestic abuse calls as highest priority more frequently, and introduced new online routes for victims to access police, communicate and share evidence with police, such as webchats or video calls. Some forces increased their perpetrator focus by charging more suspects, progressing cases quicker and making more perpetrator interventions. There is, to date, no evidence of changes in police recording practices during lockdowns, though the only published report on the matter, a police inspection report (HMICFRS, 2021), only covers the first months of the pandemic.

Finally, the poor quality of police administrative data puts limitations on this study (as well as limitations in police forces own ability to make strategic use of their data). This includes little or no systematic recording of police ‘positive action’ linked to individual crimes and incidents. Positive action includes arrests, use of Domestic Violence Protection Notices/Orders, and the removal of the abuser from a shared home, among others (College of Policing, 2018). Data quality is compromised because of high levels of inconsistent, incomplete or incorrect data entries by officers. The problem is well-documented in the literature (Bland and Ariel, 2020; HMIC, 2014b). In this study, poor data quality became evident through the data cleaning process and the cross-checking of variables, and the author discussing each variable in detail with the police analysts who provided the datasets. Where data quality appeard compromised to the extent that inclusion of the variable in question would likely produce unreliable, potentially misleading or effectively uninterpretable results these were excluded from the study. As a result, the analysis is less comprehensive than planned. Furthermore, because police systems and recording practices vary between forces, results may not be fully comparable between forces.

Conclusion

Most crime types decreased significantly during the pandemic (Buil-Gil et al., 2021). In contrast, this study finds that the overall police recorded domestic abuse volume was remarkably insensitive to the pandemic context. When changes in police recorded abuse were observed, these effects were inconsistent in direction, size and significance across lockdowns within police force areas, and varied between force areas. The variation in the statistical findings between police force areas point to local contexts and police force practices potentially playing a role in what comes to police attention and what is being recorded as such. This study does not permit identifying the causes of this observed variation. Further research is required to unpack the extent to which the differences are the result of differences in the prevalence of domestic abuse, help-seeking by victims from police for domestic abuse, or police crime recording, or a combination of all three.

COVID-19 lockdowns and the pandemic context did not create a domestic abuse crisis as has been reported across media outlets, and domestic abuse does not go away when lockdown restrictions lift. This finding runs counter to the prevailing media narrative but is unsurprising when placed in the context of feminist theories of domestic abuse. These have long argued that domestic abuse, like other forms of male violence against women and girls, is rooted in deeper structures of gendered inequality (Walby, 1989; Walklate, 2021). The DASH data analysis showed statistically significant increases in victims experiencing an escalation of the abuse and in feeling depressed or suicidal from the second national lockdown onwards. As such, lockdowns are like football matches and Christmas: they interact with and amplify underlying patterns of domestic abuse, triggering spikes in the severity or frequency of abuse which then come to police attention (see also Brooks-Hay and Lombard, 2018; Williamson et al., 2020). Anticipating such surges in police demand for domestic abuse is important for police forces to allocate enough resources to their domestic abuse response. Furthermore, the findings revealed differential impacts of lockdowns on family abuse, ex-partner and current intimate partner abuse, with implications beyond pandemic contexts. Police responses to domestic abuse must be alive to differing risks and dynamics in different abuse contexts, something Robinson et al. (2018) suggest is not yet the case.

In sum, this study provides empirical evidence that extreme changes in a nations’ circumstance, in this instance, a series of national lockdowns in the context of the COVID-19 pandemic, can result in a worsening of the lived experience of the severity of domestic abuse and its negative impact on victims, as disclosed to police as part of the DASH risk assessment. Yet, this large, seven-force study found only small, inconsistent, and often not statistically significant effects of lockdowns on the police recorded volume of domestic abuse-related crimes. Lockdowns, like other events, can produce situationally induced spikes, but they do not lead to a fundamental shift in domestic abuse.

Footnotes

Acknowledgements

The author would like to thank Kelly Johnson for their discussions of the study findings. She would also like to thank Andy Myhill and both anonymous reviewers for their helpful comments.

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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: I would like to thank the URKI-ESRC for funding the research (grant reference number ES/V007033/1), this study would not have been possible without it.

ORCID iD

Katrin Hohl

Notes

Author biography

Katrin Hohl is Professor of Criminology and Criminal Justice at City, University of London. Her research centres on police responses to sexual violence and domestic abuse.

References

Bates

Struthers

(2022) Domestic abuse identification and increased complexity. In: Domestic abuse, safeguarding and covid-19 final project conference, 10 March 2022, Available at: https://domesticabuse-safeguarding-covid19.co.uk/wp-content/uploads/2022/03/DASC-10th-March-2022-slidepack.pdf

Bates

Hoeger

Stoneman

M-J

, et al. (2021) Domestic homicides and suspected victim suicides during the covid-19 pandemic 2020–2021. Available at: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/1013128/Domestic_homicides_and_suspected_victim_suicides_during_the_Covid-19_Pandemic_2020-2021.pdf

Bland

Ariel

(2020) Targeting Domestic Abuse with Police Data. Cham: Springer.

Boxall

Morgan

(2021) Who is most at risk of physical and sexual partner violence and coercive control during the COVID-19 pandemic? Trends & Issues in Crime and Criminal Justice 618: 1–19.

Brennan

Myhill

(2021) Coercive control: Patterns in crimes, arrests and outcomes for a new domestic abuse offence. The British Journal of Criminology 62(2): 468–483.

Brodie

Hawkins

MacLean

, et al. (2023) ‘Abusers are using COVID to enhance abuse’: Domestic abuse helpline workers’ perspectives on the impact of COVID-19 restrictions on those living with domestic abuse. Journal of Social and Personal Relationships 40: 2375–2399.

Brooks-Hay

Lombard

(2018) ‘Home game’: Domestic abuse and football. Journal of Gender-Based Violence 2(1): 93–108.

Brown

Kirk-Wade

(2021) Coronavirus: A History of English Lockdown Laws (Briefing Paper 9068). London: House of Commons Library. Available at: https://commonslibrary.parliament.uk/research-briefings/cbp-9068/

Buil-Gil

Miró-Llinares

Moneva

, et al. (2021) Cybercrime and shifts in opportunities during COVID-19: A preliminary analysis in the UK. European Societies 23(1): 47–59.

10.

Burman

Smailes

Chantler

(2004) ‘Culture’ as a barrier to service provision and delivery: Domestic violence services for minoritized women. Critical Social Policy 24(3): 332–357.

11.

Campedelli

Aziani

Favarin

(2021) Exploring the immediate effects of COVID-19 containment policies on crime: An empirical analysis of the short-term aftermath in Los Angeles. American Journal of Criminal Justice 46(5): 704–727.

12.

College of Policing (2018) Arrest and other positive approaches. Available at: https://www.college.police.uk/app/major-investigation-and-public-protection/domestic-abuse/arrest-and-other-positive-approaches (accessed 26 June 2023).

13.

Cross

Evans

MacLeavy

, et al. (2022) Analysing the socio-economic impacts of COVID-19: A new regional geography or pandemic enhanced inequalities? Regional Studies, Regional Science 9(1): 461–485.

14.

Donovan

Hester

(2011) Seeking help from the enemy: Help-seeking strategies of those in same-sex relationships who have experienced domestic abuse. Child and Family Law Quarterly 23(1): 26–40.

15.

Elliott

Bardwell

Kamke

, et al. (2022) Survivors’ concerns during the Covid-19 pandemic: Qualitative insights from the national sexual assault online hotline. Journal of Interpersonal Violence 38: 84–107.

16.

End Violence Against Women Coalition (2020) Rapidly compiled – Initial briefing on the COVID-19 pandemic and the duty to prevent violence against women & girls. Available at: https://www.endviolenceagainstwomen.org.uk/wp-content/uploads/EVAW-Coalition-Briefing-on-COVID19-Pandemic-and-Duty-to-Prevent-VAWG-April-2020-FINAL.pdf

17.

Etheridge

Spantig

(2022) The gender gap in mental well-being at the onset of the Covid-19 pandemic: Evidence from the UK. European Economic Review 145: 104114.

18.

Femicide Census (2020) Femicide census report 2020. Available at: https://www.femicidecensus.org/wp-content/uploads/2022/02/010998-2020-Femicide-Report_V2.pdf

19.

Ganslmeier

Van Parys

Vlandas

(2022) Compliance with the first UK covid-19 lockdown and the compounding effects of weather. Scientific Reports 12: 3821–3920.

20.

Gregory

Williamson

(2022) ‘I think it just made everything very much more intense’: A qualitative secondary analysis exploring the role of friends and family providing support to survivors of domestic abuse during the COVID-19 pandemic. Journal of Family Violence 37: 991–1004.

21.

Gregory

Nnawulezi

Sullivan

(2021) Understanding how domestic violence shelter rules may influence survivor empowerment. Journal of Interpersonal Violence 36(1–2): NP402–NP423.

22.

Her Majesty’s Inspectorate of Constabulary Fire and Rescue Services (HMICFRS) (2019) The police response to domestic abuse: An update report. Available at: https://www.justiceinspectorates.gov.uk///hmicfrs/wp-content/uploads/the-police-response-to-domestic-abuse-an-update-report.pdf

23.

Her Majesty’s Inspectorate of Constabulary Fire and Rescue Services (HMICFRS) (2021) Review of policing domestic abuse during the pandemic. Available at: https://www.justiceinspectorates.gov.uk/hmicfrs/wp-content/uploads/review-of-policing-domestic-abuse-during-the-pandemic-2021.pdf

24.

Her Majesty’s Inspectorate of Constabulary (HMIC) (2014a) Crime-recording: Making the victim count: The final report of an inspection of crime data integrity in police forces in England and Wales. Available at: https://www.justiceinspectorates.gov.uk/hmicfrs/wp-content/uploads/crime-recording-making-the-victim-count.pdf

25.

Her Majesty’s Inspectorate of Constabulary (HMIC) (2014b) Crime data integrity: Literature review of crime-recording by Ms Patricia Mayhew. Available at: https://www.justiceinspectorates.gov.uk/hmicfrs/wp-content/uploads/crime-data-integrity-literature-review.pdf

26.

Home Office (2023) Home Office Counting Rules for Recorded Crime. London: Home Office. Available at: https://www.gov.uk/government/publications/counting-rules-for-recorded-crime

27.

House of Commons (2021) Domestic abuse and Covid-19: A year into the pandemic, 11 May. Available at: https://commonslibrary.parliament.uk/domestic-abuse-and-covid-19-a-year-into-the-pandemic/

28.

Hoyle

(1998) Negotiating Domestic Violence: Police, Criminal Justice, and Victims. Oxford: Oxford University Press.

29.

Hoyle

Sanders

(2000) Police response to domestic violence. The British Journal of Criminology 40(1): 14–36.

30.

Institute for Government (2021) Timeline of UK coronavirus lockdowns and measures, March 2020 to December 2021. Available at: https://www.instituteforgovernment.org.uk/sites/default/files/timeline-coronavirus-lockdown-december-2021.pdf

31.

Ivandić

Kirchmaier

Linton

(2020) Changing patterns of domestic abuse during Covid-19 lockdown. Centre for Economic Performance Discussion Paper, no. 1729. London: London School of Economics and Social Science. Available at: http://eprints.lse.ac.uk/108483/1/dp1729.pdf

32.

Johnson

Hohl

(2023) Police responses to domestic abuse during the Covid-19 pandemic: Positive action and police legitimacy. Policing: A Journal of Policy and Practice 17: paac108.

33.

Kourti

Stavridou

Panagouli

, et al. (2023) Domestic violence during the COVID-19 pandemic: A systematic review. Trauma, Violence, & Abuse 24: 719–745.

34.

Langton

Dixon

Farrell

(2021) Six months in: Pandemic crime trends in England and Wales. Crime Science 10(1): 6.

35.

Leslie

Wilson

(2020) Sheltering in place and domestic violence: Evidence from calls for service during COVID-19. Journal of Public Economics 189: 104241.

36.

Lovett

Hales

Kelly

, et al. (2022) Appendix 11: Pillar 5 – Data and performance in operation Soteria bluestone year 1 report. Available at: https://www.gov.uk/government/publications/operation-soteria-year-one-report

37.

Lyons

Brewer

(2021) Experiences of intimate partner violence during lockdown and the COVID-19 pandemic. Journal of Family Violence 37: 969–977.

38.

McCrary

Sanga

(2021) The impact of the coronavirus lockdown on domestic violence. American Law and Economics Review 23(1): 137–163.

39.

McGlynn

Rackley

Houghton

(2017) Beyond ‘revenge porn’: The continuum of image-based sexual abuse. Feminist Legal Studies 25: 25–46.

40.

Medina Ariza

Robinson

Myhill

(2016) Cheaper, faster, better: Expectations and achievements in police risk assessment of domestic abuse. Policing: A Journal of Policy and Practice 10(4): 341–350.

41.

Monckton-Smith

(2021) In Control: Dangerous Relationships and How They End in Murder. London: Bloomsbury Publishing.

42.

Moore

Buckley

Howarth

, et al. (2021) Police referrals for domestic abuse before and during the first COVID-19 lockdown: An analysis of routine data from one specialist service in South Wales. Journal of Public Health 44: e252–e259.

43.

Myhill

Hohl

(2019) The ‘Golden thread’: Coercive control and risk assessment for domestic violence. Journal of Interpersonal Violence 34(21–22): 4477–4497.

44.

Myhill

Johnson

(2016) Police use of discretion in response to domestic violence. Criminology & Criminal Justice 16(1): 3–20.

45.

Office for National Statistics (ONS) (2016) Intimate personal violence and partner abuse. Available at: https://www.ons.gov.uk/peoplepopulationandcommunity/crimeandjustice/compendium/focusonviolentcrimeandsexualoffences/yearendingmarch2015/chapter4intimatepersonalviolenceandpartnerabuse#police-and-the-criminal-justice-system

46.

Office for National Statistics (ONS) (2020) Domestic abuse: Prevalence and trends, England and Wales: Year ending March 2021. Available at: https://www.ons.gov.uk/peoplepopulationandcommunity/crimeandjustice/articles/domesticabuseprevalenceandtrendsenglandandwales/yearendingmarch2021

47.

Payne

Morgan

Piquero

(2022) COVID-19 and social distancing measures in Queensland, Australia, are associated with short-term decreases in recorded violent crime. Journal of Experimental Criminology 18(1): 89–113.

48.

Perez-Vincent

Schargrodsky

García Mejía

(2021) Crime under lockdown: The impact of COVID-19 on citizen security in the city of Buenos Aires. Criminology & Public Policy 20(3): 463–492.

49.

Pfitzner

Fitz-Gibbon

Meyer

(2022) Responding to women experiencing domestic and family violence during the COVID-19 pandemic: Exploring experiences and impacts of remote service delivery in Australia. Child & Family Social Work 27(1): 30–40.

50.

Piquero

Jennings

Jemison

, et al. (2021) Domestic violence during the COVID-19 pandemic – evidence from a systematic review and meta-analysis. Journal of Criminal Justice 74: 101806.

51.

Robinson

Pinchevsky

Guthrie

(2018) A small constellation: Risk factors informing police perceptions of domestic abuse. Policing and Society 28(2): 189–204.

52.

The Guardian (2020) Revealed: Surge in domestic violence during Covid-19 crisis, 12 April. Available at: https://www.theguardian.com/society/2020/apr/12/domestic-violence-surges-seven-hundred-per-cent-uk-coronavirus

53.

The New York Times (2020) A new covid-19 crisis: Domestic violence rises worldwide, 6 April. Available at: https://www.nytimes.com/2020/04/06/world/coronavirus-domestic-violence.html

54.

Thornton

(2020) Covid-19: A&E visits in England fall by 25% in week after lockdown. The BMJ 369: m1401.

55.

Walby

(1989) Theorising patriarchy. Sociology 23(2): 213–234.

56.

Walklate

(2021) Criminological futures and gendered violence(s): Lessons from the global pandemic for criminology. Australian & New Zealand Journal of Criminology 54(1): 47–59.

57.

Walklate

Godfrey

Richardson

(2022) Changes and continuities in police responses to domestic abuse in England and Wales during the Covid-19 ‘lockdown’. Policing and Society 32(2): 221–233.

58.

Williamson

Lombard

Brooks-Hay

(2020) Domestic violence and abuse, coronavirus, and the media narrative. Journal of Gender-Based Violence 4(2): 289–294.

59.

Wing

Simon

Bello-Gomez

(2018) Designing difference in difference studies: Best practices for public health policy research. Annual Review of Public Health 39(1): 453–469.

60.

Wolf

Hobart

, et al. (2003) Barriers to seeking police help for intimate partner violence. Journal of Family Violence 18(2): 121–129.

The impact of the COVID-19 pandemic on police recorded domestic abuse: Empirical evidence from seven English police forces

Abstract

Keywords

Introduction

Background

The effect of COVID-19 restrictions on the volume of domestic abuse is unclear

Police data provide a particular lens on domestic abuse

COVID-19 restrictions made living with domestic abuse and help-seeking more difficult

Methodology

Data

Analytic approach and variables

Key variables and selection of lockdown dates

Additional explanatory variables

Findings

Discussion

Conclusion

Footnotes

Acknowledgements

Declaration of Conflicting Interests

Funding

ORCID iD

Notes

Author biography

References