Abstract
Danish studies of intimate partner violence (ipv) using police data are scarce, in part because access to records had been limited. The present study reduces critical gaps in the scholarly literature by examining ipv offending patterns in Denmark, using nearly 10,000 ipv incidents reported to the North Zealand Police, Denmark (2015–2019). We explore a common framework for analysing ipv, by observing (a) frequency, (b) severity, (c) intermittency, (d) escalation, and (e) concentrations of ipv. Harm is estimated using the Danish Crime Harm Index, which is based on the sentencing guidelines as an objective rod for estimating severity. Findings support the gender-based explanation for ipv, with males causing considerably more and higher harm than female offenders. Furthermore, the likelihood of re-offending only predicable not for 1/3 of the ipv offender population and rarely for high-harm incidents as they usually have no prior or no subsequent contact with the police. While there is a tendency towards escalation of harm between contacts to the police for all offenders, no such consistent pattern is discernible for ipv offenders who cause serious harm to their victims. Implications for policy and future research are discussed.
Introduction
Intimate partner violence (ipv) is a key part of the crime policy agenda in Denmark. This pressed the Danish Police to consider efficient management of ipv incidents, offenders, and victims, with a pronounced focus on risk, protection, and investigation (Danish National Police 2020b). Yet while ipv research in Denmark has made strides with regard to understanding victimisation patterns (Helweg-Larsen 2012; Helweg-Larsen and Frederiksen 2007, 2008; Ottosen and Østergaard 2018), less is known about ipv offenders and their patterns of behaviour (cf. Leth 2009; Rye 2018). We are unaware of systematic analyses of the offending patterns of ipv offenders in Denmark; what is their frequency of offending? How severe are the reported ipv incidents? Can we reasonably predict ipv offending using police records? These questions are largely unanswered, and these gaps enclave police targeting of ipv offenders to an ad hoc, case-by-case strategy, rather than an evidence-based approach to the problem.
Previous research utilised large samples of police records to analyse patterns of ipv offending and victimisation (Barnham et al. 2017; Bland 2014; Bland and Ariel 2015; Dudfield et al 2017; Kerr et al. 2017; Sherman et al. 2017). This growing body of evidence has yielded actionable intelligence based on the frequency, intermittency, escalation, severity, and concentration of ipv offenses (Bland and Ariel 2020). Given the lack of published evidence on offenders based on police records in Denmark, there is a critical need for a systematic assessment of the problem. The present study is an attempt to utilise the same analytical techniques on Danish data.
Literature Review
Measurement of ipv
Several analytical approaches exist to study ipv offending. The most common framework to analyse ipv includes observing frequency, severity, intermittency, escalation, and concentrations of ipv (Bland and Ariel 2020).
Frequency. Regularity of offending is a strong predictor of future offending, which is why understanding its patterns is important for prevention purposes. Frequency is the measurement of repeat offending and therefore allows us to categorise between different types of offenders based on the habituality of their criminal behaviour. Barnham (2016), exploring 140,998 partner abuse incidents over a six-year period, found that 37% of all incidents involved a repeat offender. Very similarly, Bland (2014) explored 36,742 domestic abuse incidents recorded in a five-year period in local England & Wales police force and found that 65% of the identified offenders only had one recorded incident in their criminal history. Moreover, 28% of offenders had 2 – 4 recorded crimes, but 7% had more than 5 recorded incidents. The concentration of multiple incidents within 7% of the overall population can be particularly useful for a prevention strategy, as it allows the police to target the felonious few who cause the most harm (Sherman 2019). Of course, 7% would still comprise of more than 2,600 offenders to potentially target – an impossible target for any one police department – but nevertheless an improvement over treating all known ipv offenders as posing the same risk level of recidivism.
Severity. Severity and harm take centre stage in recent research on ipv (Sherman 2017). All abuse is damaging, but there are variations between different types of assaults, threats, or violence perpetrators inflict on their victims. Severity scores can be used as indexes that enable a more nuanced classification of offenders, which can then applied to typologize ipv pathologies – for example coercion, pathological, sadism, or homicidal (see Elisha, Idisis, Timor & Addad 2010; Dobash, Dobash, Cavanagh, & Medina-Ariza 2007). Reserach shows that there are phenomenological differences between domestic homicide and non-lethal domestic abuse in terms of the ‘emotions that trigger it, the circumstances that led up to it, and the state of mind that characterises it’ (Goussinsky & Yassour-Borochowitz, 2012:553).
Thus, not all ipv is created equal, and identifying variations between the types of harm seems important for the development of valid ipv policy. Sherman et al. (2016) and others have recently called to observe harm scores rather than counting crime incidents. Their argument is compelling: there is a broad spectrum of abuse, and the level of harm the victim suffers varies – between victims and even between incidents suffered by the same victim. Therefore, giving all ipv events the same weight is crude and we should avoid the imprecision offered by a system that does not distinguish between different levels of harm. Importantly, however, the variance that exists in severity of the abuse can be quantified. Different approaches were developed to estimate harm (e.g., Babyak, Alavi, Collins, Halladay, and Tapper 2009; Boivin, 2014; Curtis-Ham and Walton 2017; Kärrholm, Neyroud, and Smaaland 2020; Sherman, Neyroud, and Neyroud 2016; Wolfgang 1985; Wolfgang, Figlio, Tracy, and Singer 1985).
Mathematically, severity measures can use any number of scales. Earlier works used surveys to gauge from citizens their own views of what crimes are considered more serouis than others (e.g., Wolfgang 1985) – however these are subjective and therefore carry both conceptual and methodological limitations. More objective rods use the legal system to classify crime; in the United States the ‘mildest’ crime types are referred to as infractions, followed by more severe crimes known as misdemeanours, followed by the most “serious” crimes called felonies. Aggravating or mitigating circumstances are also ordinal scales of severity, which often looks at both the characteristics of the offence as well as of the offender and the victim (Amirault and Beauregard 2014). More sophisticated classifications systems use such as the aforementioned Cambridge Crime Harm Index (cchi; Sherman et al. 2016; Barnham et al. 2017; Bland and Ariel 2015; Kerr et al. 2017; Sherman et al. 2016)), the UK’s experimental crime severity index (ons 2020), or the official Canadian Crime Severity Index (ccsi), are all examples thereof (see also Bendlin and Sheridan 2019; Bonomi et al. 2006; Straus et al. 1996). Each system has its own inherent limitations (see for example Sarnecki 2021), but we concur that counting all crimes as having equal weights would lead to inefficient results.
Intermittency. Related to the question of frequency is the temporal intervals at which incidents occur – every day, week, month, etc. – as an indication of persistence or desistance of ipv (Blumstein et al. 1988; Fagan 1989; Mele 2009; Piquero et al. 2006). Intermittency is usually measured as the time passed between calls for service to the police (Bland and Ariel 2020). Interestingly, there is usually a declining time interval between reported victimisations (which is separate from the question of escalation of crime severity). For example, Lloyd et al. (1994) reported that following the first domestic incident, 35% of households called the police again within 35 days, and after the second incident, 45% of the households called again within 35 days. Similarly, Mele (2009) reported the number of victimisations decreases from 62 days between the first and second victimization to15 days between the 9th and 10th victimisation (Mele 2009). Kerr et al. (2017) reported a similar decline, with 270 days separating the first and second incidents and 46 days separating the 9th and 10th incidents.
Escalation. One of the most heated areas of research in ips research is the matter of increased harm over time, and how to measure it. Whilst some scholar reported more acute and aggressive violence becomes over time (Pagelow, 1981; Walker, 2006), others show escalaory patterns for some dyads, but not in others (Kerr, Whyte and Srang 2016; Piquero et al., 2006), whilst more recent research debunks the genearlised escalatory hypothesis (Barnham, Barnes and Sherman 2016; Bland and Ariel 2015). There is, however, strong argument against an escalation hypothesis insofar as homicidal ipv is concerned: most domestic homicides occur without any contact with the police, escalatory or otherwise, which debunks the escalation hypothesis (Bridger, Strang, Parkinson, & Sherman, 2017; Button, Angel, & Sherman, 2017; Dobash, Dobash, Cavanagh, & Medina-Ariza, 2007; Chalkley & Strang, 2017; Goussinsky and Yassour-Borochowitz 2012; McPhedran and Baker 2012: 967; Sherman, Schmidt, Rogan, & DeRiso 1991; Sherman & Strang, 1996; Thornton, 2017). At the same time, escalation does occur (see Piquero et al. 2006), but it is admittedly difficult to predict which dyads will experience increases in harm and which will not (Bland and Ariel 2020).
Concentrations. Studies have found considerable ipv concentration at the level of individual offenders, victims, and couple dyads. This is perhaps unsurprising, as ipv is subject to a pareto curve distribution, where a small number of units – e.g., offenders – account for an outsized proportion of harm. For example, Barnham et al. (2017) found that 3% of ipv offenders generate 90% of all crime harm, while Bland and Ariel (2015) reported that 1.7% of couple dyads accounted for 80% of all crime harm.
ipv Research in Denmark
Over the past two decades, Danish ipv studies have generally focused on the victims of intimate physical, sexual, and psychological abuse. Offenders are usually glossed over. Moreover, this research primarily relies on surveys, qualitative interviews, and data from public health agencies (Helweg-Larsen 2012; Helweg-Larsen and Frederiksen 2007; Plauborg et al. 2012; Ottosen et al. 2018). To this extent, research has generally focused on violence against women and interventions that have been directed at assisting female victims (Bertelsen et al. 2019; Helweg-Larsen and Frederiksen 2007, 2008; Helweg-Larsen and Kruse 2003).
Nevertheless, the body of research remains informative, and lessons are presently available about ipv offending. Danish studies focused on the frequency and severity ipv. This is in accordance with governmental efforts to analyse these characteristics for the benefit of tailoring preventative approaches for ipv victims. The findings typically demonstrate the presence of a “power few” phenomenon as stipulated earlier (Sherman, 2007), where a small group of victims experience greater and more severe ipv incidents (Balvig and Kyvsgaard 2006; Deen et al. 2018; Helweg-Larsen 2012; Heinskou et al. 2017; Plauborg et al. 2012).
Furthermore, Danish ipv research has produced mixed results on the normalcy or exceptionalism of escalation. This is primarily due to variability in data sources. Jensen and Nielsen (2005) surveying 968 women from 34 different crisis shelters in Denmark and found that many ipv victims experience high levels of victimisation, more severe forms of violence, greater harm escalation, and decreasing intermittency between incidents. For example, 33% of female victims experienced violence on a daily basis when in a relationship for less than a year. In contrast, studies using a mixture of survey samples and agency-based data found low frequency and severity amongst the majority of ipv victims (Helweg-Larsen 2012). These variations in terms of both scope as well as type of ipv between self-reported and official statistics are a global phenomenon (e.g., Brown, Cohen, Johnson, and Salzinger 1998; Cooper and Obolenskaya 2021). We return to these methodological considerations in the discussion chapter.
However, the primary gap in the literature is the lack of studies utilising Danish police records to understand ipv patterns using the offender or the offence as the units of analysis. The non-Danish studies reviewed above paid close attention to frequency, severity, intermittency, escalation, and concentrations of ipv based on police records. Our aim is to follow this approach in this study and fill the gap in the literature with information extrapolated from official records.
As it relates to ipv research in Europe, Costa et al. (2015), examining ipv across six European cities, found that men and women generally experienced ipv as both victims and perpetrators with few significant sex-based differences. Moreover, ipv was predominant across European urban centres. Reichel (2017), analysing the determinants of ipv in the European Union, found a higher prevalence of ipv among couples with a low economic status. Furthermore, women suffer more often from violence if they do not have an equal say about household income. Barbier, Chariot, and Lefevre (2020) conducted a European-wide survey of 40357 ever-partnered women to determine the prevalence of ipv. The researchers found that prevalence of physical, sexual and psychological ipv were respectively 20%, 8.4%, and 48.5%. They conclude that the lifetime prevalence of ipv in EU is high and likely to be underestimated. Specific trajectories and profiles of perpetrators should be characterized to ground the interventions.
This study differs somewhat from the current European ipv literature as it goes beyond examining the prevalence of ipv and toward measurements of intermittency, conditional probability, and power few. Moreover, no European study of ipv has considered ipv as it relates to harm. This added dimension gives a more rounded perspective of ipv and aids in a more accurate allocation of resources.
Methods
Setting. We examine 9,889 unique ipv cases committed by 4,847 unique offenders between 2015 and 2019 in North Zealand, Denmark. The North Zealand Police District covers thirteen municipalities in the Danish capital of Copenhagen. Whereas nine municipalities are in predominantly rural areas, four are urban zones that constitute the Greater Copenhagen city area (Danish National Police 2021; Statistics Denmark 2021).
Data. we use incident records from the Denmark Police Case Management System employed by the Danish Police. Data include information on the location of the incident, crime or non-crime classification, date and time, report type, incident summary, crime scene type, and operational grouping. We also gained access to the gender, age, and verdict type (if available) of the offender.
We note that in order to be included in our data, a list of 50 search words were used against the police records to identify ipv incidents. These include terms such as “ex-boyfriend,” “domestic dispute,” and “violence”. This approach is not without limitations, as false positives (incidents tagged as ipv incidents but are not) and false negatives (incidents not tagged as ipv incidents but are) were present and manually omitted; search words failed to tag 5% of all ipv incidents.
Measurements. Bland and Ariel (2020) noted that ipv in the English context “is not a crime classification in its own right” and could involve any type of crime. This is the same in Denmark. For the purposes of this paper, ipv is defined as “physical violence against a current or former partner or attempted violence or threats of violence” (Danish National Police 2017). We are cognisant that domestic abuse can include other types of exploitation, including psychological, economic, social, coercive, or otherwise (see Lévesque and Rousseau 2021; Nicolaidis and Paranjape 2009; Tjaden 2000), however as these categories are missing from police records in Denmark, we are forced to omit these from the analysis, and discuss this limitation in the discussion chapter. Nevertheless, offenders are persons assigned the status of “accused”, “suspect”, or “person of interest” linked to an ipv incident, and they were all included in the cohort.
In this study. a “repeat” ipv offender are those who have recorded an ipv incident within two years of their first offence recorded within the study period. These include both crime and non-crime incidents.
Finally, following Bland and Ariel (2015), this study defines escalation as a two-legged concept: harm escalation and intermittency. The first concept tracks the harm measure over time while the second measures speed of the next ipv relative to the first.
Analytical Approach. We employed several approaches to analyse ipv data. first, descriptive statistics are used the summarise the data, taking stock of the frequency and severity of ipv incidents. Conditional probability is then used to ascertain the likelihood of future ipv offending. Here, the conditional probability is based on the number of incidents reported to the police and the likelihood of reporting a new incident to the police when x number of previous incidents have already been reported (see Bland 2014: 38). Intermittency is determined by calculating the number of days between the first ten incidents for all offenders and high-harm offenders.
We consider harm as a function of the Danish Crime Harm Index (‘dchi’; Andersen 2017). The dchi assigns each crime category a score based on the recommended number of days in prison that first-time offenders would receive if the case were to progress to court. The number of imprisonable days is based on the Danish Sentencing Guidelines, which make the dchi a relatively objective measure of severity (however cf. Sarnecki 2020).
The trajectory of intermittency is analysed by analysing upward or downward average tendencies. The analysis uses the reported date of each incident to calculate the number of days between calls to the police.
Finally, concentration trends were calculated as a cumulative distribution function; high harm offenders are classified as offenders generating 80% of the total crime harm for all offenders with a crime harm score of 100 dchi value or more. The same approach was applied in previous studies that looked at pareto curves associated with domestic abuse (e.g., Barnham et al. 2017; Kerr et al. 2017).
Findings
Sample Characteristics
Between 2015 and 2019, there was a total 9,889 ipv incidents and 4,847 ipv offenders. Furthermore, 69% (6,832) and 39% (3,057) of ipv incidents were crime and non-crimed events, respectively. On average, there were approximately two reported incidents against every known offender. Figure 1 displays the number of incidents and offenders for each year. The results demonstrate an upward trajectory for both incidents and offenders. This increase is possibly due to changes in public policy and management of ipv by Danish law enforcement: since 2014, Danish officials have spent increasing resources on improving the detection rate of ipv as well as increasing positive action taken against offenders. Alterations to recording practices, exposure to violence, and the willingness to report by victims and witnesses may also be contributing factors in the year-to-year variations – which, collectively, imply that these figures should necessarily imply the ipv has varied over time, but rather extraneous factors associated with recording artefacts.
Yearly distribution of ipv incidents and offenders
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
Yearly distribution of ipv incidents and offenders
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
Yearly distribution of ipv incidents and offenders
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
Still, violent crimes accounted for 46.2% of all crime incidents (Penal Code offences) while specific laws governing abuse such as economic and sex crimes accounted 30%, 13.8%, 5.4% and 4.6% of all crime incidents, respectively. These categories can be broken down further: Assault causing bodily harm and death threats were the most common violent crimes, while restraining order violations, unwarranted access to personal property, and unwarranted sharing of personal property were the predominate special law crimes. Vandalism and theft were the most frequently occurring property crimes, while fraud and rape were the most prevalent property and sex crimes, respectively.
Male offenders generated more ipv incidents and greater crime harm relative to female offenders. About 74% of offenders were male, and while men were involved in 2.2 ipv incidents, female offenders were involved in 1.5 incidents, on average. In terms of harm, male offenders committed an average harm score of 54.6 dchi, while female offenders had an average value of 17.1 dchi.
A similar pattern is observable when we accounted for the age of offenders. Figures 2a and 2b present the count and harm of ipv incidents when accounting for the age and gender of offenders, respectively. As sown in Figure 2a, male and female offenders follow similar age-crime curves, albeit at differing frequencies. Indeed, the number of ipv incidents peaks between the ages of 25 and 39 but decreases rapidly after age 59 for both genders. This pattern persists when examining crime harm as well (Figure 2b). Male offenders generated far higher levels of crime harm between the ages of 55 and 94 relative to female offenders.
(a) ipv incidents by age and gender. (b) ipv harm by age and gender
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
(a) ipv incidents by age and gender. (b) ipv harm by age and gender
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
(a) ipv incidents by age and gender. (b) ipv harm by age and gender
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
Table 1 presents the distribution of incidents amongst all offenders by frequency of offending. The results demonstrate that 65.7% of all identified offenders had only one recorded incident between 2015 and 2019. As such, the majority of ipv offenders were not repeat offenders, committing only a single offense throughout the period of study – even though as noted earlier the average number of offences associated with each offender was approximately two, and 16.3% and 7.3% of all offenders were involved in two and three incidents, respectively. In general, these results demonstrate that the overwhelming majority of ipv offenders (94.9%) had a low to medium frequency of offending. Offenders with a high frequency of offending (five or more incidents) only accounted for 5.1% of the cohort under examination.
Repeat Offending and Conditional Probability of Repeat Offending
With regards to repeat offenders, 33% (n = 940) of all offenders apprehended between 2015 and 2017 (n = 2,850) were involved in a second ipv incident within two years of their first arrest. Moreover, many of these offenders reoffended in the same year as their first offense and did not reappear in police records. Within this cohort of 940 repeat offenders, only 35 continue to reoffend in each of years following their first offense. These offenders were involved in 15 ipv incidents, on average, within the period of study.
Figure 3a presents the conditional probability of further ipv incidents for the first 16 incidents between 2015 and 2019. Whereas offenders with one ipv incident have a 34% likelihood of being involved in a 2nd incident, the likelihood of involvement in 3rd ipv incident following the 2nd incident jumps to 52%. To this extent, the conditional probability of further incidents gradually increases until the 9th incident where offenders have an 83% probability of having a 10th incident. Following this, the conditional probability of repeat offending becomes increasingly volatile due to the low number of offenders with more than ten reported incidents.
(a) Conditional probability of crime and non-crime incidents. (b) Conditional probability of crime-only incidents
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
(a) Conditional probability of crime and non-crime incidents. (b) Conditional probability of crime-only incidents
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
(a) Conditional probability of crime and non-crime incidents. (b) Conditional probability of crime-only incidents
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
When accounting for crimed ipv incidents (as opposed to non-crimed incidents), offenders with a single incident have a 31% probability of committing a 2nd offence (see figure 3b). This increases to 47% and 54% from the 2nd to 3rd and 3rd to 4th ipv incidents, respectively. In fact, the likelihood of a further incident increases until the 9th incident where offenders have an 85% likelihood of committing a 10th incident. The suggests a high likelihood of repeat offending for medium to high frequency offenders.
Escalation of Severity Between Sequential Offending
Escalation within all offenders. Figure 4a presents the mean dchi scores in the first ten incidents for all ipv offenders. The results demonstrate a pattern of escalating harm between the 1st and 4th incidents, with the average harm score increasing from 21.0 dchi to 34.0 dchi. Between the 5th and 9th incidents, the average crime harm score decreases and fluctuates between 11 dchi and 20 dchi before escalating to 35 dchi with the tenth incident. Again, this inconsistency is due primarily to the low number of offenders with 5 or more ipv offenses.
(a) Average dchi score in first 10 incidents (all offenders). (b) Average dchi score in first 10 incidents (high-harm offenders)
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
(a) Average dchi score in first 10 incidents (all offenders). (b) Average dchi score in first 10 incidents (high-harm offenders)
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
(a) Average dchi score in first 10 incidents (all offenders). (b) Average dchi score in first 10 incidents (high-harm offenders)
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
Escalation within the high-harm offenders. Figure 4b presents the average dchi scores in the first ten incidents for high-harm offenders who generated 80% of all crime harm with a crime harm score of 100 or higher. The results show decreasing harm between the 1st and 5th incident where the average crime harm score fell from 234 dchi to 45 dchi. Between the sixth and eight incidents, the average harm score hovered around 50 dchi until increasing from 25 to 83 between the 9th and 10th incidents. This pattern of de-escalating harm is attributable to the large proportion of one-incident-high-harm offenders. Out of the 20 offenders with the highest crime harm scores, ten were one-incident offenders. Among the offenders generating 80% of the crime harm among high harm offenders (the power few of the power few), half were one- or two-incident-high-harm offenders.
Intermittency
The data suggest a decreasing intermittency within the first ten incidents, as the number of days between calls gradually declines as the number of incidents increases – i.e., the offences occur more frequently (Figure 5a). Between the 1st and 2nd incident the number of days between calls is 166 days or 5.5 months. In comparison, there were 46 days between the 9th and 10th incidents.
(a)Average intermittency in first 10 incidents (all offenders). (b) Average intermittency in first 10 incidents (high harm offenders)
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
(a)Average intermittency in first 10 incidents (all offenders). (b) Average intermittency in first 10 incidents (high harm offenders)
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
(a)Average intermittency in first 10 incidents (all offenders). (b) Average intermittency in first 10 incidents (high harm offenders)
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
The intermittency pattern differs when we examine high-harm offenders (see Figure 5b). There is greater fluctuation in intermittency despite a downward trend. Between the 1st and 5th incident the number of days between calls decreased from 104 days to 38 days. The number of days increased between the 5th and 6th calls and only fell below the previous 40-day low between the 9th and 10th incident. Overall, these results suggest that ipv offenders typically increase the speed of reoffending, thereby escalating their offending activity and potentially generating more harm within a shorter period.
Concentration of Harm
Based on the cumulative distribution function, 1.7% (86) of all offenders generated 50% of the total reported ipv crime harm between 2015 and 2019 (Figure 6). Furthermore, 6.8% of all offenders accounted for 80% of the total crime harm. As such, ipv offending in North Zealand is subject to a power law distribution, with 331 out of 4,847 offenders were responsible for the nearly all ipv crime harm.
Power curve for ipv offending harm amongst all offenders
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
Power curve for ipv offending harm amongst all offenders
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
Power curve for ipv offending harm amongst all offenders
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
Following Bland (2014), we divided all high harm offenders (those who were responsible for 80% of crime harm) into three groups: “never called before” (1 ipv incident), “intermediate” (2 to 4 ipv incidents), and “chronic” (5 or more ipv incidents). Those who never called before represented 1.9% of all offenders as were responsible for 27.6% of the total ipv crime harm. In comparison, intermediates accounted for 3% of all offenders and generated 33.6% of all crime harm while chronic ipv offenders represented 1.9% of all offenders and accounted for 18.8% of the total crime harm.
Rural-Urban Divide
Finally, the nine rural municipalities of North Zealand contain 61.5% (362,834) of the district’s population and covers 87.1% of the district’s area. In comparison, the four urban municipalities possess 38.5% of the population and covers 12.9% of the district’s area. The average number of incidents per offender is 2.0 in rural municipalities and 1.9 in urban municipalities. Moreover, the average harm score per offender is 46.5 in rural areas and 33.2 in urban areas. In general, the total number of incidents and offenders was twice as high in rural areas while the total crime harm score was three times higher in these areas. Furthermore, the ipv incidents with the highest harm were predominately committed in rural municipalities; this includes 75% of all homicides, 86.4% of all attempted homicides, and 73.6% of all domestic rapes.
Figure 7 presents the average harm scores in the first ten incidents for all offenders by rural-urban classification. The results suggest both similarities and differences concerning harm escalation. In rural areas, there are two rounds of harm escalation between the first and fourth incidents and between the fifth and eighth incidents. The average harm score increases from 21 dchi to 39 dchi between the first and fourth incident then increases from 14 dchi to 26 dchi between the fifth and eighth incident. Between the ninth and tenth incident there is a rapid increase in harm. Despite the two rounds of harm escalation there is no consistent upward tendency in the rural areas. In fact, the overall trajectory of harm in rural areas seems to be downward. Urban areas demonstrate escalating crime harm between the first three incidents followed by steady decrease. The average harm score increases from 16 to 29 dchi between the first and third incident, decreases from 22 to 7 dchi between the fourth and sixth incident, and then hover around 10 dchi between the seventh and tenth incidents. These results underscore the differences between the severity of offending in rural and urban areas although neither area shows any consistent upward trend in harm escalation.
Average harm scores in first 10 incidents for all offenders: rural versus urban
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
Average harm scores in first 10 incidents for all offenders: rural versus urban
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
Average harm scores in first 10 incidents for all offenders: rural versus urban
Citation: European Journal of Crime, Criminal Law and Criminal Justice 30, 3-4 (2022) ; 10.1163/15718174-bja10036
Discussion
The North Zealand Police identified 9,889 incidents and 4,847 offenders in the five-year period from 2015 through 2019. Access to official police ipv data reaffirmed gender-based theories of ipv, as male offenders have considerably higher levels of crime frequency and crime harm relative to female offenders. Male offenders also have a longer age-crime harm trajectory than female offenders.
The conditional probability of further ipv incidents gradually increases until the 9th incident where offenders have an 83% probability of having a 10th incident. These findings suggest a strong predictability of ‘heavy repeaters’, which in turn can assist the police in preventing repeat victimisation. The identification of a small group of offenders with a high frequency of offending supports findings from earlier studies (Bland 2014). To this extent, 35 offenders were present in each year of this study, which makes them a predicable cohort. However, such a policy would be adequate for a selected few, as 65.7% of offenders had only one recorded incident between 2015 and 2019 and low to medium frequency offenders’ account for 94.9% of all identified offenders. These findings are not unique and have been shown in previous studies using the same analytical approach (e.g., Bland and Ariel, 2015; Barnham et al. 2017; Kerr et al. 2017), which strengthens the ubiquity of the findings.
With regards to harm escalation or de-escalation, the results demonstrate both increasing and decreasing harm in the first ten incidents, but no overall tendency towards harm escalation. As such, our findings resemble Piquero et al.’s (2006) findings of a heterogeneous population of abuse, with some but not all dyads showing escalatory tendencies. We are not in a position to ascertain what covariates are associated with each upwards or downward harm trend, which is a topic that should be covered in future research. On the other hand, our results suggest decreasing intermittencies for all offenders and high harm offenders – although these high-harm offenders show signs of fluctuation, which is likely driven by a statistical underpowered sample.
Not surprisingly, a small group of offenders represent the ‘power few’ cohort, generating the highest amounts of ipv harm. One the one hand, this is encouraging as it informs a preventative approach with a limited number of offenders. The trouble, however, is that most of these high-harm offenders are unpredictable using police ipv records: they do not appear in the system prior to the one high-harm incident for which they were charged and more than a third of them never appear again. Indeed, while there is tendency towards escalation of harm for all offenders, no such consistent pattern is discernible for high harm offenders. These results imply that using police records – and by implication the dchi which is based on these official data – are not particularly useful to identify offenders that are most at risk of committing the most harmful ipv against their partners. More research is needed, preferably with alternative data sources such as self-reported crime, to look at these issues more closely.
Directly linked to the question of predictability, we note that, ordinarily, risk assessment tools rely heavily on factors associated with escalation in offending frequency and severity (Danish National Police 2017). However, the results of our study demonstrate that escalation is far from a consistent feature in ipv incidents, despite evidence of decreasing intermittency between calls for medium to high frequency offenders. In general, the results of this study do not rule out the potential harm reduction benefits of risk assessments tools based on police data. However, the police data do demonstrate both the limitations of using escalation as a central risk factor and the need for the continued evaluation of risk assessment tools.
Importantly, the inordinate amount of crime harm caused by one-incident high harm offenders has two implications. First, law enforcement should increase their reliance on information provided by victims, particularly from those with low levels of prior police contact. The fact that the victim complained about stalking or low-level property damage – incidents with relatively low dchi scores – should not be interpreted as an incident with low risk of escalation. In fact, studies relying on domestic abuse hotline data suggest that “abusers [have] a history of trauma, alcohol/drug problems, mental illness, and homicidal and suicidal ideations” (Hines et al. 2007:63) – all attributes or behaviours that score nil or low scores on any crime harm index. Therefore, insofar as predictability is concerned, dchi is not the only – and seemingly not the strongest – factor.
Finally, as it relates to data quality, the Danish police should strive to tag all incidents with ipv search keys in order correctly identify ipv offenders. Currently, very few police districts in Denmark have a strategic overview of ipv incidents and offenders. This hampers data analysis and subsequent resources allocation. Improvements on this front would also benefit the National Danish Police and their ability to affect public policy on a national level by showing the importance of data quality.
With regard piv research in Europe, this study goes beyond examinations of prevalence and public opinion and toward measurements of intermittency, conditional probability, and power few. Importantly, it also introduces a harm measurement which has not been attempted in prior European studies of ipv. This study aid greater depth of to ipv research within European criminology. As it relates to future research on this topic, researchers should aim at replicating these findings across other jurisdictions, testing the generalizability of escalation, intermittency, conditional probability, and power few. Furthermore, researchers might also conduct experiments around to the power few of offenders to determine which interventions might reduce the level of harm these offenders account for.
References
Amirault, J., & Beauregard, E. (2014). The impact of aggravating and mitigating factors on the sentence severity of sex offenders: An exploration and comparison of differences between offending groups. Criminal Justice Policy Review, 25(1), 78–104.
Andersen, H.A. (2017) ‘How Measurement Matters – The Creation of a Danish Crime Harm Index and Its Initial Application to National Police Data’, (Unpublished M.St. Thesis in Applied Criminology and Police Management, University of Cambridge).
Babyak, C., Alavi, A., Collins, K., Halladay, A., & Tapper, D. (2009). The Methodology of the Police-Reported Crime Severity Index. Statistics Canada, Household Surveys Methods Division. hsmd-2009006E/F. Ottawa.
Balvig, F. and Kyvsgaard, B. (2006) Vold og overgreb mod kvinder, København: Justitsministeriet.
Barbier, A., Chariot, P., and Lefevre, T. (2020). Describing Intimate partner violence against ever-partnered women in Europe and perpetrator’s characteristics. Results from the Violence against women EU-wide survey. European Journal of Public Health, 30(5), 123–134.
Barnham, L. (2016) ‘Targeting perpetrators of partner abuse in the Thames Valley – A two year follow up of crime harm and escalation’, (Unpublished M.St. Thesis in Applied Criminology and Police Management, University of Cambridge).
Barnham, L., Barnes, G. and Sherman, L. (2017) ‘Targeting Escalation of Intimate Partner Violence: Evidence from 52,000 Offenders’, Cambridge Journal of Evidence-Based Policing, 1(2–3): 116–142.
Bendlin, M. and Sheridan, L. (2019) ‘Risk Factors for Severe Violence in Intimate Partner Stalking Situations: An Analysis of Police Records’, Journal of Interpersonal Violence, 1–22.
Bertelsen, E., Sørensen, W.Ø. and Jensen, M.W. (2019) ‘Intimate Partner (Sexual) Violence: Danish Research and Policy’, Journal of Aggression, Maltreatment & Trauma, 28(1): 25–46.
Bland, M. (2014) ‘Targeting escalation in common domestic abuse: how much if any?’, (Unpublished M.St. Thesis in Applied Criminology and Police Management, University of Cambridge).
Bland, M. and Ariel, B. (2015) ‘Targeting Escalation in Reported Domestic Abuse: Evidence From 36,000 Callouts’, International Criminal Justice Review, 25(1): 30–53.
Bland, M. and Ariel, B. (2020) Targeting Domestic Abuse with Police Data, Cham: Springer.
Blumstein, A., Cohen, J. and Farrington, D.P. (1988) ‘Criminal career research: Its value for criminology’, Criminology, 26(1): 1–36.
Boivin, R. (2014). Prince George is not (and never was) Canada’s most dangerous city: Using police-recorded data for comparison of volume and seriousness of crimes. Social indicators research, 116(3), 899–907.
Bonomi, A., Holt, V.L., Martin, D.P. and Thompson, R.S. (2006) ‘Severity of Intimate Partner Violence and Occurrence and Frequency of Police Calls’, Journal of Interpersonal Violence, 21(10): 1354–1364.
Brown, J., Cohen, P., Johnson, J. G., & Salzinger, S. (1998). A longitudinal analysis of risk factors for child maltreatment: Findings of a 17-year prospective study of officially recorded and self-reported child abuse and neglect. Child abuse & neglect, 22(11), 1065–1078.
Cooper, K., & Obolenskaya, P. (2021). Hidden victims: the gendered data gap of violent crime. The British Journal of Criminology, 61(4), 905–925.
Costa, D., Soares, J., Lindert, J., Hatzidimitriadou, E., Sundin, O., Toth, O., Ioannidi-Kapolo, E., & Barros, H. (2015). Intimate partner violence: a study in men and women from six European countries. International Journal of Public Health volume 60, 467–478.
Danish National Police (2017) Vurdering af risiko for samlivsrelateret vold - SARA:SV – Brugervejledning, Ejby: Nationalt Forebyggelsescenter.
Danish National Police (2020b) Retningslinjer for politiets håndtering af sager om chikane, forfølgelse og stalking, Købehavn: Rigspolitiet.
Danish National Police (2021) Dimensioner. Retrieved 18th June 2021 from http://cas01qvp03/QlikView/.
Deen, L., Johansen, K.B.H., Møller, S.P. and Laursen, B. (2018) Vold og seksuelle krænkelser – En afdækning af omfang og udvikling af fysisk vold og seksuelle overgreb og omfang af seksuelle krænkelser samt en analyse af erfaringer med digitale seksuelle krænkelser, København: Statens Institut For Folkesundhed.
Dobash, R. E., Dobash, R. P., Cavanagh, K., & Medina-Ariza, J. (2007). Lethal and nonlethal violence against an intimate female partner: Comparing male murderers to nonlethal abusers. Violence against women, 13(4), 329–353.
Elisha, E., Idisis, Y., Timor, U., & Addad, M. (2010). Typology of intimate partner homicide: Personal, interpersonal, and environmental characteristics of men who murdered their female intimate partner. International journal of offender therapy and comparative criminology, 54(4), 494–516.
Fagan, J. (1989) ‘Cessation of Family Violence: Deterrence and Dissuasion’, Crime and Justice, 11: 377–425.
Goussinsky, R., & Yassour-Borochowitz, D. (2012). “I killed her, but I never laid a finger on her”—A phenomenological difference between wife-killing and wife-battering. Aggression and Violent Behavior, 17(6), 553–564.
Heinskou, M.B., Schierff, L.M., Friis, C.B. and Liebst, L. (2017) Seksuelle krænkelser i Danmark: Omfang og karakter, København: Det Kriminalpræventive Råd.
Helweg-Larsen, K. (2012) Vold i nære relationer – Omfanget, karakteren og udviklingen samt indsatsen mod partnervold blandt mænd og kvinder – 2010, København: Ministeriet for Ligestilling og Kirke.
Helweg-Larsen, K. and Frederiksen, M.L. (2007) Men’s violence against women – Extent, characteristics – and the measures against violence - 2007, Copenhagen: Minister for Gender Equality.
Helweg-Larsen, K. and Frederiksen, M.L. (2008) Vold mod mænd i Danmark – Omfang og karakter - 2008, København: Minister for Ligestilling.
Hines, D. A., Brown, J., & Dunning, E. (2007). Characteristics of callers to the domestic abuse helpline for men. Journal of Family Violence, 22(2), 63–72.
Jensen, V.L. and Nielsen, S.L. (2005) Når vold er hverdag – en undersøgelse af mænds vold mod kvinder i nære relationer, København: lokk – Landsorganisationen for Kvindekrisecentre.
Kärrholm, F., Neyroud, P., & Smaaland, J. (2020). Designing the Swedish Crime Harm Index: an evidence-based strategy. Cambridge Journal of Evidence-Based Policing, 4(1), 15–33.
Kerr, J., Whyte, C. and Strang, H. (2017) ‘Targeting Escalation and Harm in Intimate Partner Violence: Evidence From Northern Territory Police, Australia’, Cambridge Journal of Evidence-Based Policing, 1(2): 143–159.
Leth, P.M. (2009) ‘Intimate partner homicide’, Forensic Science, Medicine and Pathology, 5(3): 199–203.
Lévesque, S., & Rousseau, C. (2021). Young women’s acknowledgment of reproductive coercion: A qualitative analysis. Journal of interpersonal violence, 36(15–16), NP8200-NP8223.
Lloyd, S., Farrell, G. and Pease, K. (1994) Preventing Repeated Domestic Violence: A Demonstration Project on Merseyside (Police Research Group Crime Prevention Unit Series: Paper No. 49), London: Home Office Police Department.
Mele, M. (2009) ‘The Time Course of Repeat Intimate Partner Violence’, Journal of Family Violence, 24(8): 619–624.
Nicolaidis, C., & Paranjape, A. (2009). Defining intimate partner violence: Controversies and implications. Intimate partner violence: A health-based perspective, 19–30.
Ottosen, M.H. and Østergaard, S.V. (2018) Psykisk Partnervold – En Kvantitativ Kortlægning, København: vive – Viden til Velfærd.
Piquero, A. R., Brame, R., Fagan, J., & Moffitt, T. E. (2006). Assessing the offending activity of criminal domestic violence suspects: Offense specialization, escalation, and de-escalation evidence from the Spouse Assault Replication Program. Public health reports, 121(4), 409–418.
Plauborg, R., Johansen, K.B.H. and Helweg-Larsen, K. (2012) Kærestevold i Danmark – En undersøgelse af omfang, karakter og konsekvenser af volden blandt unge og udviklingen 2007–2011, København: Statens Institut For Folkesundhed.
Reichel, D. (2017). Determinants of Intimate Partner Violence in Europe: The Role of Socioeconomic Status, Inequality, and Partner Behavior. Journal of Interpersonal Violence, 32(12), 1853–1873.
Rye, S. (2018) ‘A Descriptive Analysis of Intimate Partner Homicide in Denmark 2007–2017’, (Unpublished M.St. Thesis in Applied Criminology and Police Management, University of Cambridge).
Sarnecki, J. (2021). Comment on Kärrholm et al. “Designing the Swedish Crime Harm Index: an Evidence-Based Strategy”. Cambridge Journal of Evidence-Based Policing, 5(1), 76–90.
Sherman, L., Strang, H., & O’Connor, D. (2017). Introduction—Key Facts About Domestic Abuse: Lessons from Eight Studies.
Sherman, L.W. (2007) ‘The power few: experimental criminology and the reduction of harm’, Journal of Experimental Criminology, 3(4): 299–321.
Statistics Denmark (2021) Area. Retrieved 18th June 2021 from https://www.dst.dk/en/Statistik/emner/geografi-miljoe-og-energi/areal/areal.
Straus, M.A., Hamby, S.L., Boney-McCoy, S. and Sugarman, D.B. (1996) ‘The Revised Conflict Tactics Scales (cts2) – Development and Preliminary Psychometric Data’, Journal of Family Issues, 17(3): 283–316.
Tjaden, P. G. (2000). Extent, nature, and consequences of intimate partner violence. US Department of Justice, Office of Justice Programs, National Institute of Justice.
Wolfgang, M. E. (1985). The national survey of crime severity. US Department of Justice, Bureau of Justice Statistics.