Marieke Liem is Professor of Violence and Interventions at Leiden University, where she and her team take part in the European Homicide Monitor, and are in charge of updating and maintaining the Dutch Homicide Monitor. A graduate of the University of Cambridge, she completed her PhD in Forensic Psychology from Utrecht University (cum laude). Before joining Leiden University, she was a Marie Curie fellow at Harvard University’s Kennedy School of Government. Her research interests involve interpersonal violence, with specific research projects on domestic homicide (including intimate partner homicide), homicide by the mentally ill, homicide followed by suicide, the effects of confinement on violent offenders, and international comparative research on lethal violence.

2.1 Homicide as a Yardstick

Homicide is generally considered the most serious of all crimes (Smit et al., 2012), and according to some, it constitutes the “tip of the iceberg” of underlying crime. In this view, homicide is the end result of lesser forms of crimes, such as robberies, rapes, and thefts (Ouimet & Montmagny-Grenier, 2014). The assumption is that different forms of crime are likely to share common causes, yet that police are much more likely to record homicides than other types of (non-lethal) crime (Lauritsen et al., 2016). From this line of reasoning, the homicide rate (reflecting the number of homicides per 100,000 population) is frequently used as an indicator of the level of violence in cross-national and historical studies (Nivette, 2011; Oberwittler, 2019; UNODC, 2019).

But there is another, more practical reason why homicide is frequently used as an indicator of the level of violence: It is seen as the most reliably measured of all crimes (Oberwittler, 2019; Pridemore, 2005). Homicides, unlike other (violent) crimes, leave a body behind, making this type of crime more visible and detectable by the authorities (Oberwittler, 2019; Ouimet & Montmagny-Grenier, 2014), regardless of reporting trends (Neapolitan, 1997). Other categories of crime data are thought to suffer from considerable validity problems (Neapolitan, 1997). More specifically, crimes of violence are not defined in the same way in different countries, and police also do not use the same thresholds of aggravation in the classification of violent offences in different countries. Additionally, police practices for recording crime are thought to be much more likely to affect nonlethal violent crimes than homicides (O'Brien, 1996). Against this backdrop, homicide data are believed to have a greater external validity than other types of crime (Andersson & Kazemian, 2018). Its lethal outcome and its universal condemnation make homicide particularly amenable to temporal (longitudinal) and cross-sectional (geographic) comparisons (UNODC, 2019).

Given its salience, it is perhaps surprising that for a long time, European comparative homicide research has remained a relatively marginal field. Compared to the United States and several commonwealth countries, Europe does not have a long tradition of studying homicide trends, patterns, and explanations (Liem, 2017). This may be due to the large differences that exist between European countries in legal definitions of and data sources on homicide. In addition, the overwhelming presence of the United States as reference point in studies on European homicide may have impeded comparative analyses within Europe (Granath et al., 2011; Liem et al. 2013). The past decade, however, saw a substantial body of new scholarly work on European homicide, combined with initiatives to systematically gather homicide data on a pan-European level (Liem, 2017). In this contribution, I will reflect on these initiatives; particularly, on the construction of the European Homicide Monitor (EHM) and how it relates to other initiatives, such as the Balkan Homicide Study (BHS) described in the book at hand. First, however, let me take the opportunity to provide a sketch of prior and current research on homicide in Europe, so that initiatives such as the EHM and BHS can be put into empirical perspective.

2.2 A Stocktaking of European Homicide Research

Research on homicide in Europe can roughly be divided into four clusters: sociological, historical, forensic, and descriptive studies (i.e., studying specific subtypes of homicide) (Granath et al., 2011). Next, I will provide a brief overview of each of these clusters.

Sociological Approaches to Homicide in Europe

One of the earliest accounts of sociological approaches to homicide in Europe can be traced back to the 1920s, when the Finnish scholar Verkko (1951) observed that the proportion of female homicide victims was higher when the overall homicide rate was low and vice versa. Homicides involving unrelated young males as offender and victim tended to be the most variable part. In other words, increases and decreases of homicide are typically explained by the prevalence of such male-to-male encounters. If male-to-male homicides increase, the proportions of other types of homicide (such as female homicide) tend to decrease. Similarly, if male-to-male homicides decrease, the relative proportion of other homicides increases (Kivivuori et al., 2012). Today, these laws are also known as “Verkko’s laws” and can still be applied to explain regional and historical variations in homicide (e.g., Gartner & Jung, 2014; Silverman & Kennedy, 1987; Trägardh, Nilsson, Granath, & Sturup, 2016).

Contemporary sociological approaches to homicide in Europe tend to focus on how the causes of homicide are located in the socio-demographic structure of society as well as in the recurring temporal and spatial dimensions and dynamics of everyday life (for an overview, see Granath et al., 2011). Much of this European research is inspired by US colleagues, as scholars have examined to what extent US findings can be found in Europe, too (Kivivuori et al., 2014). Central themes in these sociological approaches include the role of substance abuse in lethal violence: alcohol (Bye, 2008, 2012) and drugs (Schönberger et al., 2018), as well as the link between economic deprivation and homicide (McCall & Nieuwbeerta, 2007). Yet another strand of sociological perspectives in homicide research focuses on the relationship between firearms and homicide. The notion of guns facilitating violence is the key assumption behind the strict regulation of gun ownership in most European countries (Krüsselmann et al. 2021a, b). In a recent systematic review, we found some European studies showing a clear decline once availability of firearms is restricted, while other studies indicated a limited effect on only a very specific subgroup, such as female victims, or national guards with weapons at home. Due to methodological inconsistencies and regional differences, conclusive evidence on the relationship between the two is still lacking (Krüsselmann et al., 2021).

Historical Approaches to Homicide in Europe

Through historical analyses, scholars have been able to trace homicide figures in Europe back to the fifteenth century, when about 50 people per 100,000 were victimised in a homicide. Over the years, this figure decreased to about 1 per 100,000 – a downward trend that continued well into the twentieth century. Homicide rates in Western Europe have remained stable and low (below 2 per 100,000) until approximately the late 1960s. Starting in the early 1970s, homicide rates showed a slight increase throughout Europe, before decreasing again in the 1990s (Eisner & Nivette 2012). It has been argued that this increase can be attributed to an increase in homicides between young men in public places, who are often strangers to one another (Eisner, 2008). The overall European decrease in homicide rate in the early 1990s, in turn, could be explained by pan-Western cultural changes: around this time, pan-Western cultures were marked by an increased emphasis on self-control and more conservative cultural values. In their latest analysis of Western European homicide rates, Aebi and Linde (2014) hold that the increases and decreases of homicide can be seen as reflections from a change in lifestyle. They attribute the parallel trends in male and female victimisation since the 1960s to the integration of women into the labor market and the convergence of similar lifestyles by men and women. As a result, men and women are exposed to similar risks outside of their homes. From a lifestyle theory perspective, the decrease of homicide in the late 1990s could be attributed to the rapid development of computer technologies and the Internet, leading to an increase of time spent at home, especially for young people, and in turn, a lowered risk of homicide victimisation (Aebi & Linde, 2014; Aarten & Liem, 2021).

It is important to note that not all European countries followed this pattern: The homicide drop was particularly noticeable in Western European countries. Homicide levels in Eastern Europe remained relatively high and started to decline much later, while rates in southern European countries have converged to levels also found in Northern and Western Europe (Eisner, 2003). Recently, scholars from the Nordic countries have combined forces in generating a Historical Homicide Monitor that seeks to capture – much like the European Homicide Monitor (see later on) – individual-based and incident-based historical homicide data in a uniform way, allowing for international historical comparisons.

Forensic Approaches to Homicide

A third line of research on homicide in Europe involves forensic approaches to homicide, in which the study of the role of mental illness in homicide is most pronounced. Several population-based studies in England (Flynn et al., 2011; Nielssen & Large, 2010; Swinson et al., 2011), Denmark (Brennan et al., 2000), Sweden, and Finland (Eronen et al., 1996; Tiihonen et al., 1997) revealed a higher prevalence of mental illness among homicide offenders compared to the general population (for an overview, see Aarten & Liem, 2021). Similar findings have been reported on the relationship between mental illness and victims of homicide in studies in Sweden (Crump et al., 2013) and Denmark (Hiroeh et al., 2001). Within the forensic approach to homicidal behaviour, numerous European studies have focused on specific subtypes of mental illness. Here, the focus lies on the association between psychotic disorders, such as schizophrenia, and homicidal behaviour (Fazel et al., 2010; Sturup & Lindqvist, 2014; Vinkers & Liem, 2011). Each of these studies have taken a national perspective, describing the nature and incidence of the relationship between severe mental illness and homicide in separate countries. With the exception of several meta-analyses (Fazel, Gulati, Linsell, Geddes, & Grann, 2009; Nielssen & Large, 2010) that include various European countries other than Western countries, studies based on pan-European data are virtually absent.

Descriptive Approaches to Subtypes of Homicide

The fourth set of studies on homicide in Europe is also the most voluminous and the most rapidly growing (Kivivuori et al., 2014). These studies focus on specific subtypes of homicide, in which research on domestic homicide is well represented. This predominantly includes research on intimate partner homicide (for a detailed overview, see Corradi and Stöckl (2014)) and child homicide. Studies in the latter category mostly rely on forensic-psychiatric, rather than national, data (Vanamo, Kauppi, Karkola, Merikanto, & Räsänen, 2001; Liem & Koenraadt, 2008). Homicide followed by suicide constitutes another homicide subtype that has been studied in European countries separately (Flynn et al., 2009; Kivivuori & Lehti, 2003; Liem et al., 2009; Shiferaw et al., 2010) as well as several countries combined (Liem et al., 2011). Finally, due to their low prevalence in Europe, studies on other subtypes of homicide, such as sexual homicides, are rare (Greenall & Richardson, 2015; Häkkänen-Nyholm et al., 2009) or, such as in the case of serial homicides, virtually absent and limited to anecdotal accounts.

2.3 The European Homicide Monitor

The overview sketched above illustrates at least two main characteristics in European Homicide research, captured in what I would term the Balkanisation of European Homicide Research: first, the vast heterogeneity in types of studies and, closely related to that, the diversity in types of data that have been used in these studies. Due to the heterogeneity in sources, forensic mental health data cannot be one-on-one compared to data focusing on a specific type of homicide, which in turn cannot be one-on-one compared to historical data and so on. At the same time, existing international comparative studies on homicide conducted by large organisations, such as the UNODC or WHO, rely on aggregated national data. Such aggregated data, however, do not allow for detailed, individual-based, or case-based analyses. These aggregated data alone, in other words, do not tell us anything about potential international differences in motives, relationships between victim and perpetrator, and the context in which the homicide takes place.

To overcome these limitations, together with European colleagues from Finland and Sweden, we developed and launched the European Homicide Monitor (EHM) (Granath et al., 2011) about ten years ago. The EHM framework follows a uniform structure (same variables and values) that allows individual participating countries to code homicide data in a comparable format. Together, the EHM captures detailed incident, and victim and perpetrator characteristics. Since its inception, aside from the Netherlands, Finland, and Sweden, the EHM is now also applied in Estonia, Denmark, Paris, Scotland, Switzerland, and the Dutch Caribbean. Participating countries have previously transformed their primary data into the uniform EHM structure to allow for comparative analyses, showcasing the potential of this framework to be used in studying international trends (Suonpää et al., forthcoming), urban homicide (Krüsselmann et al., 2021), homicide clearance (Liem et al., 2018), the role of firearms in homicide (Krüsselmann et al., 2021), and specific types of homicide (Liem et al., 2017; Liem et al., 2013). In recent years, a slightly modified form of the European Homicide Monitor has also been applied to study homicide in the Nordic countries (Lehti et al., 2019).

Using the European Homicide Monitor as a standardised coding instrument is not, however, without shortcomings. The first, and perhaps the most important one, is that the types of data we rely on for homicide research are not initially collected for the purposes of research (Marshall & Block, 2004). Police files, for example, are drawn up for investigative purposes, typically in a diary type of way, where, during the investigation, “witnesses” may become “persons of interest,” who may ultimately become suspects. Filtering out information relevant for research purposes from such diary-oriented police systems constitutes one of the challenges. Similarly, other primary homicide data sources such as prosecution files, court transcripts, and autopsy reports and newspaper articles are not written with a research aim in mind. Not only do these sources differ in respect to their focus on the homicide incident (such as news reports), the victim (autopsy files), or the offender (criminal justice files), but consequently they also apply a different idiom to refer to these events: “death caused by exsanguination” in a coroner’s report may in other documentation be referred to as “died as a result of a gunshot wound,” which in a newspaper be reported as “victim died in a shooting” and in a court transcript reflected as “sentenced for second-degree murder.” Even though ideally we would apply and merge multiple data sources to verify the validity of the data at hand, this is oftentimes not possible. This leaves us with the challenge of finding a balance between coding cases from different sources according to a common denominator, without valuable details being lost.

The second key challenge concerns the coding of data. Coding , simply put, involves the transformation of narrative descriptions into an alphanumeric designation. The issue of coding becomes relevant when previously collected and previously coded homicide data are combined, such as in the European Homicide Monitor. Leaving aside the definitional issues that surround homicide – aspects that are almost universally coded are gender and age of the victim. Even though the EHM coding manual (see, for a detailed description: constitutes a comprehensive tool to code in a consistent and uniform manner, the recoding of other variables, which on the surface may appear to be straightforward and culturally homogeneous, becomes challenging when using data that were collected from a particular (non-research oriented) data source. A key example includes the variable motive: A newspaper report may reflect very different on the motive underlying the event when citing bereaved family members, compared to a police report, or to a forensic mental health evaluation of the suspect. Further, in its current form, the European Homicide Monitor coding scheme tends towards a lowest common denominator, the best example concerning “other” types of homicide: Next to pre-defined categories such as “homicides committed during robberies,” “homicides in the criminal milieu,” or “intimate partner homicides,” the EHM contains a category for “other” homicides. In one of our early studies using the EHM, this has resulted in 46 per cent of all Finnish homicides, 23 per cent of all Dutch homicides, and 20 per cent of all Swedish homicides in our combined dataset to be coded as “other” (Granath et al., 2011; Liem et al., 2011) – a result, one could argue, of recoding existing data into homogeneous categories tended towards a lowest common denominator. One example includes homicides occurring at night in Finland, for example, that are often preceded and precipitated by heavy drinking by both victim and offender, in a kitchen setting. Quantitative data alone do not allow for the reflection of such specific settings and contexts. Other examples of country-specific and culture-specific settings of homicide that should be maintained because of their cultural uniqueness include mafia-related homicides in Italy, homicides in groups of temporary workers in Western Europe, honour-related killings among immigrant groups, and so on. One of the lessons learned from working on the European Homicide Monitor is to allow for these unique settings: leaving room for a short descriptive (string) variable with room for a short narrative on the specific case. In this way, we will be better able to capture the cultural and contextual uniqueness of homicide cases in each country.

Another challenge we face when using and analysing data from the European Homicide Monitor concerns missing data. The EHM is not unique in this – missingness is a researcher’s curse encountered in many other large homicide datasets, including the FBI’s Supplementary Homicide Reports (SHR) (Fox, 2004), the National Violent Death Reporting System (NVDRS) (Logan et al., 2009), and the Chicago Homicide Dataset (Block & Block, 1998), to name a few. The most prominent type of missingness in data pertains to detailed offender and victim characteristics as well as to victim-offender relationship (Fox, 2004). Accurately documenting patterns and trends in homicide rates distinguished by the relationship between perpetrators and victims is an important issue for the epidemiology of crime in Europe. The extent of domestic and intimate partner homicides relative to acquaintance and stranger homicides tells us much about the nature of violent crime in Europe, how it differs across countries, and how it is changing over time. Yet, missing data compromise the ability to reach theoretically relevant conclusions about the context and meaning of homicide rates (Pampel & Williams, 2000). One reason for missingness in the European Homicide Monitor and other datasets alike concerns unsolved cases. An unknown offender implies an unknown motive, unknown circumstances, and an unknown victim-offender relationship (Liem et al., 2018). A persistent misconception in homicide research is that the “unknowns” in the victim/offender relationship variable are stranger homicides because this type of homicide is more difficult to clear by arrest than those in which victims knew their offenders. Decker (1993), however, showed that stranger homicides do not account for many homicides classified as unknown relationships; indeed, they may be distributed among uncleared cases in the same proportions as they are among cleared homicide cases.

Even though several statistical solutions have been applied previously in large-scale homicide databases – including imputation-based procedures, weighting procedures, and model-based procedures (Riedel & Regoeczi, 2004) – imputed data do not have the same standing as observed data. Statistical solutions for missing data are no substitute for data collection that results in no missing values (Riedel & Regoeczi, 2004). The solution, hence, lies in the minimisation of missingness by going back to the source. In further developing the European Homicide Monitor, this should be achieved by consulting additional data sources and by making efforts to follow up on homicides that are solved at a later stage and, therefore, are able to provide background information at a later stage. Another solution has been applied in the National Violent Death Reporting System (CDC, 2020) data coding process. Today, the NVDRS operates in all 50 US states. As states have joined in one by one, over the years, research staff provides training sessions and guidance to adequately code and enter data in the NVDRS software manual. Though costly, this elaborate process does not only decrease the occurrence of missing data from the bottom up but also strengthens the internal validity of the data. In further developing the European Homicide Monitor, we should learn from the wheel others invented before us when assembling large datasets, while at the same time reshaping this wheel according to the unique European – including Balkan – context.

Finally, in its present form, the European Homicide Monitor spans – in some countries – more than two decades, allowing for unique trend analyses. At the same time, we continue to encourage other countries to join this initiative and encourage other research fields to reap the benefits of this data coding and data collection endeavour. Examples include current projects on firearm homicide, in the context of illegal firearms trafficking (Project TARGET; and for an overview see, for example, Krüsselmann et al., 2021), and drug-related homicide, in the context of drug-related crime (Schönberger et al., 2018). To lower the threshold of applying the EHM structure in such affiliated projects, we now also offer a user-friendly, condensed nucleus set of 25 variables that capture the most important and readily retrievable victim, offender, and case characteristics (for an overview, see

2.4 Unique Challenges in a Balkanised Setting

The BHS team sought to overcome similar problems as we faced in developing and applying the European Homicide Monitor. This included, but was certainly not limited to, a lack of unified reporting system throughout the Balkans, a lack of a unified definition, and cultural-linguistic differences. However, there are some unique challenges involved in doing homicide research in the Balkans that deserve closer consideration. First, while criminology as a discipline has grown into a rich, versatile, and independent field of study in many Western and Northern European countries, this is not the case in the Balkans. This vacuum is reflected in the criminological research capacities, which are almost exclusively situated at universities. Second, criminological research in the area seems to focus on national rather than comparative issues. Further, as opposed to the Western and Northern European countries, most criminological studies are state-funded, with relatively few national or regional foundations that fund criminological research. Against this background, there is little experience in conducting homicide research specifically to such an extent that the BHS team struggled with a lack of experience and guidance on aspects such as sampling, recruiting, and training of field workers, as well as case analysis, quality control, and so on. In one of our many conversations on how to navigate these difficulties, Anna-Maria Getoš Kalac, the author of the book that lays before you, summarised the approach taken as “learning by doing” – yet learning the hard way.

These challenges, taken together, have made it very difficult to compare the findings from the BHS to studies conducted outside of the region. In moving forward, I can only encourage the BHS team to join the well-proven concept of the European Homicide Monitor (EHM). Despite its initial start-up challenges, the European Homicide Monitor has now expanded beyond the initial three pilot countries, and in regular feedback loops, we continue to improve it over time. These factors taken together, the EHM promises to be an even richer data source in the future to be used by researchers and policy makers. As I have discussed in this overview, despite the central need for sound knowledge on lethal violence, most EU countries, including the Balkan countries, lack well-developed data of the kind that is required for reliable assessments. Further developing the European Homicide Monitor can fill these lacunae. Such developments, I feel, go hand in hand with combining forces in sharing practices and lessons learned in successfully conducting comparative multi-country homicide studies. This does not only include sharing experiences in designing and conducting research based on a broad exchange of experiences but also involve opening the discussion on cultural aspects in homicide research. Expanding and combining our data collection efforts can provide unique opportunities to follow and make assessments of trends and factors that foster lethal violence, as well as preventive measures, sentencing policies and the treatment of perpetrators from a pan-European perspective. It is my hope that this would greatly improve the opportunities for EU-level initiatives to work in different ways to reduce the burden of lethal violence.