Faculty of Governance and Global Affairs
MSc Crisis and Security Management
INTIMATE PARTNER HOMICIDE IN ITALY (2002-2014)
A longitudinal and a cross-sectional analysis of rates and correlates
Eleonora Rossi - S1702947
Supervisor: Dr. M.C.A. Liem
Second Reader: Dr. G.M. van Buuren
2 Table of Contents
1 Introduction ... 5
1.1 Academic and social relevance ... 6
1.2 Italian trends in homicide and intimate partner homicide ... 7
1.3 Characteristics of intimate partner homicide in Italy ... 8
1.4 Research question ... 9
2 Theoretical review ... 10
2.1 Background ... 10
2.2 Verkko’s laws ... 12
2.3 Male sexual proprietariness theory... 12
2.4 Violence perspective and strain theory ... 13
2.5 Gender perspective ... 13
2.6 Exposure reduction and backlash/retaliation theories ... 14
2.7 Routine activity and lifestyle theories ... 15
2.8 Critical assessment of the theories ... 16
3 Empirical review ... 17
3.1 Background ... 17
3.2 ‘Conventionality’ of intimate partner homicide perpetrators ... 19
3.3 Intimate partner homicide and alcohol consumption ... 20
3.4 Lethal and non-lethal intimate partner violence ... 21
3.5 Male and female perpetrators of intimate partner homicide ... 23
3
3.7 Critical assessment of the literature ... 24
4 Methodology ... 26 4.1 Background ... 26 4.2 Data sources ... 28 4.3 Dependent variables ... 29 4.4 Independent variables ... 29 4.5 Operationalization ... 31 4.6 Data analysis... 34 5 Results... 34
5.1 Homicide and intimate partner homicide trends in Italy (2002-2014) ... 34
5.2 Longitudinal analysis ... 37
5.3 Homicide and intimate partner homicide rates by region (2013) ... 41
5.4 Regional distribution of potential predictors of intimate partner homicide ... 42
5.5 Cross-sectional analysis ... 46
6 Discussion ... 50
6.1 Bridging theoretical approaches and empirical findings ... 51
6.2 Limitations and future research ... 54
6.3 Recommendations ... 56
6.4 Conclusion ... 56
References ... 58
Appendix A ... 64
4 Appendix C ... 67
5 1 Introduction
One in seven homicides worldwide (14%) is committed by an intimate partner (Stockl et al., 2013). This proportion is highest in those regions – such as Europe – in which the homicide rate is lowest. In fact, although all types of homicide in Europe have been in a decreasing trend since the XIV century – except for the period 1960s-1990s – this decline overwhelmingly regarded male-to-male lethal violence. Consequently, the proportion of homicides committed by an intimate partner has steadily increased from the middle ages until modern times (Spierenburg, 2012). This trend is partly explained by Verkko’s laws (Verkko, 1951), which suggest that homicide rates are mainly driven by (young) males, as most homicides are male-to-male homicides and male-to-male homicide rates tend to vary more over time. Conversely, other types of homicides – such as intimate partner homicide – tend to be more stable over time (Corradi & Stockl, 2014).
Although the relative stability of intimate partner homicide rates over time seems to suggest that the issue has not been effectively dealt with, nor properly understood, and even though the prevalence of homicides committed by an intimate partner is particularly high in Europe, European research on intimate partner homicide is extremely scarce. The main reason is probably the lack of data availability and comparability, due to differences in legal definitions and data collection methods across countries. As far as international databases are concerned, only a few of them collect data on homicides by victim-offender relationship, and these cover only a few years. For instance, Eurostat (2016) database on intimate partner homicide not only is limited to the period 2008-2014 and includes very few countries, but it is also characterized by large amounts of missing data.
Because of the reasons discussed above, the few studies on intimate partner homicide conducted in Europe focus each on a single country. Similarly, the present thesis focuses on Italy and combines a longitudinal analysis of rates and correlates of intimate partner homicide from 2002 to 2014, with a cross-sectional analysis of rates and correlates of intimate partner homicide in 17 Italian regions in
6 2013. Differently from most studies on intimate partner homicide conducted in Europe, which are based on micro-level analysis (i.e. they describe characteristics of perpetrators, victims and the incidents) and do not say anything about macro-level predictors of intimate partner homicide, the present thesis aims to fill this knowledge gap and is based on a macro-level analysis.
Italy is a suitable ground for the study of rates and correlates of intimate partner homicide for three main reasons: first, the homicide rate in Italy varied widely over the past decades, thus it is the ideal ground for testing Verkko’s laws; second, the period 2002-2014 includes the financial crisis of 2008, which strongly affected Italy as a country, causing great variation in socio-economic indicators (e.g. unemployment rate) over time; third, Italy is characterized by great cross-regional differences, making a cross-sectional analysis particularly meaningful. Nevertheless, the aim of this thesis is also to serve as foundation for future comparative research.
In sum, the overall purpose of this thesis is to contribute to European homicide research by shedding light on an under-researched topic such as intimate partner homicide. As the share of homicides committed by an intimate partner is particularly high in Europe and has been increasing since the middle ages (Spierenburg, 2012), an improved understanding of the phenomenon is central to inform better evidence-based policies to tackle it. To help filling this knowledge gap, the present thesis tests correlations between the intimate partner homicide rate in Italy and potential predictors derived from European homicide research (Liem & Pridemore, 2012; Aebi & Linde, 2014), as well as from theoretical and empirical research on intimate partner homicide.
1.1 Academic and social relevance
As mentioned above, the enormous homicide drop over the centuries in Europe has not significantly affected rates of intimate partner homicide, which are decreasing at an alarmingly slow pace (Spierenburg, 2012; Corradi & Stockl, 2014). The proportion of intimate partner homicides is becoming increasingly important, especially as far as female victimization is concerned. In Italy, for
7 example, intimate partner homicides make the largest component of homicides with female victims (EURES, 2015). For this reason, intimate partner homicide is a relevant security issue that should be studied and tackled more effectively.
As far as the academic relevance is concerned, the present thesis adds to prior research on intimate partner homicide in Europe by providing a macro-level analysis of rates and correlates of intimate partner homicide in Italy. As most existing studies are based on micro-level analyses, the present thesis provides an original perspective on the issue, serving as foundation for future – possibly systematic and comparative – research.
Moreover, it contributes to important theoretical debates such as that on the relationship between homicide and intimate partner homicide rates illustrated by Verkko’s laws (1951), or that between the ‘violence perspective’ and the ‘gender perspective’. As far as the latter is concerned, some scholars argue that intimate partner homicide originates in socio-economic disadvantage and deviance, similar to other types of homicide, whereas others maintain that intimate partner homicide perpetrators are ‘conventional’ men driven by the gendered power structure of our patriarchal society. One of the aims of the present thesis is to shed some light on these issues.
The societal relevance derives from the fact that an improved understanding of correlates of intimate partner homicide and its risk factors can contribute to the development of more effective evidence-based prevention policies. And more informed prevention policies can finally lead to a decrease in this type of homicide, which has so far remained quite stable over time (EURES, 2013; 2015). 1.2 Italian trends in homicide and intimate partner homicide
For at least five centuries, the homicide rate in Italy has been much higher than that of other European countries (Barbagli, 2004; Eisner, 2003). As late as the 1930, Finland was the country with the highest rate, followed by Greece, Austria, Italy and Portugal. Although homicide rates in Europe started to converge after World War II, they started to increase again in most countries during the 1980s and
8 1990s. The increase was particularly intense in Italy. Throughout the 1990s homicide rates diminished in most countries, and again the change registered in Italy was much more dramatic than elsewhere (Barbagli, 2004). From 1990 to 2014, the number of homicides has decreased by 70.9% (from 1,633 to 475 homicides) and is now among the lowest in Europe, only 0.8 per 100,000 inhabitants (Preti & Macciò, 2012; EURES, 2015).
As homicides related to the domestic sphere have tended to remain quite stable over time, the origin of Italy’s peculiar homicide trend is to be found in male-to-male homicide (Barbagli, 2004). This fact, together with the fact that this change was strongest in Southern regions, suggests that the anomaly of the Italian homicide trend is to be sought in the activities of the organized crime (Barbagli, 2004). On the other hand, today more than two thirds of all homicides (39.5%) happen within a domestic context, thus involving relatives or – in half of the cases – intimate partners (EURES, 2015).
In contrast to other types of homicide, characterized by a large majority of male victims, most victims of intimate partner homicide worldwide (79%) are female (UNODC, 2013) and the share of female homicide victims killed by an intimate partner (39%) is six times higher than the proportion of male homicide victims killed by an intimate partner (6%) (Stockl et al., 2013). Similarly, intimate partner homicide in Italy disproportionately affects women to the extent that, in 2014, more than half of female homicide victims (53.3%) were killed by an intimate partner (EURES, 2015).
1.3 Characteristics of intimate partner homicide in Italy
The Italian Economic and Social Research Centre (EURES) published a Report on intentional homicide in Italy in 2013, which integrates official data on homicide – based on prosecution and court statistics – with information on police investigations – drawn from the press releases contained in the database of the national agency press (ANSA), as well as from the archive on criminal incidents of the Department of Criminal Police Service and Analysis of the Ministry of the Interior (Preti &
9 Macciò, 2012). The report contains a chapter dedicated to intimate partner homicide, and an update regarding female intimate partner homicide victimization has been published in 2015.
The Report on intentional homicide (EURES, 2013) shows that in 2012 most cases of intimate partner homicide in Italy involved cohabiting couples, especially married ones (58.1%). The second category was that of ex-partners (16.3%), while divorced and separated couples were rarer (8.2%). The main age groups for the victims were 35-44 (23.3%) and over 64 (20.9%), while the main age groups for the perpetrators were 35-44 (27.2), 55-64 (19.6%) and over 64 (18.5%). Moreover, more than half of the couples (54%) had children.
In the period 2010-2014, only 4.1% female victims of intimate partner homicide were unemployed, though 29.7% were either retired, housewives or students. Most of those who were employed were office workers (19.6%) or housemaids (10.9%). The main motives for killing were jealousy and possessiveness (44.3%) or frequent conflicts in the couple (24.7%). 1 in 4 (22.4%) female victims of intimate partner homicide had previously experienced some form of intimate partner violence (EURES, 2015).
1.4 Research question
The findings outlined above show that, despite the decline in the overall homicide rate, the intimate partner homicide rate has remained quite stable over time. The fact that intimate partner homicide is one of the main homicide categories in Europe today, and that women run the highest risk to be killed by an intimate partner, suggests that policies aimed at tackling this issue have not been effective so far, and there is a need for systematic research in this field. Despite these findings, research on intimate partner homicide and its predictors is extremely limited in Europe. The present thesis aims at filling this gap by providing new insights on intimate partner homicide in the European context. The focus is on Italy, but the aim is for this study to serve as foundation for future comparative research.
10 The research question this thesis aims to answer is to what extent are demographic and socio-economic indicators typically associated with homicide in European homicide research correlated with intimate partner homicide rates in Italy over the period 2002-2014 and in 17 Italian regions in 2013?
The following sub-questions are useful to guide the research:
a) What trends in homicide and intimate partner homicide can be observed in Italy over the period 2002-2014, and do they provide support for Verkko’s laws (Verkko, 1951)?
b) Are the demographic and socio-economic indicators typically associated with homicide in European homicide research (Liem & Pridemore, 2012; Aebi & Linde, 2014) correlated with the intimate partner homicide rate in Italy over the period 2002-2014?
c) How are homicide and intimate partner homicide rates for the year 2013 distributed in the 17 Italian regions included in this study and how do they differ?
d) Are the demographic and socio-economic indicators typically associated with homicide in European homicide research (Liem & Pridemore, 2012; Aebi & Linde, 2014) correlated with the intimate partner homicide rates in 17 Italian regions in 2013?
e) Are the findings of the present thesis consistent with prior European (theoretical and empirical) homicide research?
2 Theoretical review 2.1 Background
Criminological theory only plays a limited role in European homicide research and it is more typical to find descriptive or epidemiological studies adhering to a multiple-factor approach and describing the socio-demographic distribution as well as spatial and temporal dynamics of lethal violence (Kivivuori et al., 2012). However, European homicide scholars have drawn widely from
11 criminological theory when orienting their research and interpreting the findings. The main theories-in-use in homicide research include ‘strain, social control, self-control, learning, social capital, evolutionary and feminist perspectives’ (pp.105-106). Among those theories, the present review focuses on those that have been applied to intimate partner homicide.
First, Verkko’s laws, as well as the male sexual proprietariness theory – borrowed from evolutionary psychology – can be considered classic theories of intimate partner homicide research (Verkko, 1951; Wilson & Daly, 1996; 1998). The former describes the relationship between the homicide rate and the intimate partner homicide rate, whereas the latter attempts to explain the behaviour of male intimate partner homicide offenders (Ibid.). As far as intimate partner homicide perpetrators are concerned, two main approaches can be distinguished in contemporary intimate partner homicide research in Europe: the ‘gender perspective’ and the ‘violence perspective’ (Kivivuori & Lethi, 2012). The ‘gender perspective’ argues that intimate partner violence is a product of a patriarchal society, and derives from the gendered power structure rather than from socio-economic disadvantage and deviance (Ibid.). In line with this perspective, the present review also introduces the self-defence theory – which points out that female partner homicide offenders usually act in self-defence after a long-standing abuse by their partner – and the exposure reduction and the backlash/retaliation theories – which point to the role of gender (in)equality (Serran & Firestone, 2004; Dugan et al., 1999; 2003).
By contrast, the ‘violence perspective’ suggests that intimate partner homicide finds its origin in conditions of socio-economic disadvantage and deviance, similar to other types of homicide (Kivivuori & Lethi, 2012).In this context, this theoretical review also presents (socio-economic) strain theory. In addition, routine activities and lifestyle theories have been included as well to account for the role of alcohol consumption and the increased use of internet and computers (Kivivuori et al., 2012; Aebi & Linde, 2010; 2014). Although these theories and explanations can be partially overlapping, their core elements capture distinct traditions and areas of focus (Kivivuori et al., 2012). This theoretical review analyses these theories more in depth and assesses them critically, whereas
12 the final chapters consider their usefulness in explaining the results of the data analysis of the present thesis.
2.2 Verkko’s laws
‘Verkko’s laws’ – set out by the Finnish scholar Veli Verkko (1951) – are considered classic theories in European homicide research. According to these ‘laws’, the proportion of homicide victims that are female tends to be higher when the overall homicide rate is low, and vice versa (Kivivuori et al., 2012). Similarly, the proportion of homicides that are committed by an intimate partner tends to increase when the overall homicide rate decreases. In other words, when the homicide rate is disaggregated in its constituent parts, rates of male-to-male homicide tend to more variable than rates of other types of homicide. Consequently, variations in overall homicide rates tend to be driven by male-to-male homicide. In support of ‘Verkko’s laws’, some studies (Marshall & Summers, 2010; Stockl et al., 2013) found intimate partner homicide rates more likely to stay relatively stable, or only change very slowly, over time.
2.3 Male sexual proprietariness theory
Borrowing from evolutionary psychology, the ‘male sexual proprietariness’ theory argues that sexual proprietariness is aroused when men feel their sexual monopoly over their partner is threatened, due to her fertility and attractiveness and to pressure from potential rivals (Wilson & Daly, 1996; 1998). According to Wilson and Daly (1996), non-lethal violence and threats work to deter sexual rivals and to limit female autonomy, however their utility depends on the credibility of such threats. If the threatening party is perceived as unwilling to follow through when the threat is ignored or defied, then the threat loses its effectiveness. In such circumstances, if the female partner decides to desert the relationship, the male partner might overstep the boundary of utility and even commit murder (Wilson & Daly, 1996; 1998).
13 2.4 Violence perspective and strain theory
Several studies have found homicide concentrated on the lowest social strata and related to economic deprivation, providing some support for strain theory (Kivivuori & Lethi, 2006; Savolainen, 2010). The American sociologist Robert K. Merton conceptualized economic strain as ‘a discrepancy between culturally induced goals and structurally limited opportunities to achieve them’, from which crime tend to stem (Kivivuori et al., 2012, 97; Merton,1938). In a study conducted using Dutch communities as the unit of analysis, Nieuwbeerta et al. (2008) found that the effect of economic disadvantage on homicide risk was the same for all types of homicide, providing some support for the above mentioned ‘violence perspective’ (Kivivuori et al., 2012).
Torrubiano-Domínguez et al. (2015, 1) conducted an ecological longitudinal analysis to determine whether ‘the increase in unemployment among both sexes in different regions in Spain’ was related to an increase in the rates of male-perpetrated intimate partner homicide during the recent financial crisis period. They found that, instead, there has been a decrease in male-perpetrated partner homicide rates, which does not seem to have been influenced by unemployment. The observed tendency in intimate partner homicide may be attributed to ‘other factors such as the decrease in divorce or the protective effects of the gender violence law in effect prior to the beginning of the crisis’ (p.6). However, they note, the decrease in male-perpetrated partner homicides could be related to an increase in invisible (non-lethal) intimate partner violence cases.
2.5 Gender perspective
In a general sense, feminist theories can be considered a variant of learning theory, in that they understand male violence as a reflection of socialization processes in which males are taught to ‘assume hegemony and proprietary rights over women’ (Kivivuori et al., 2012, 103). Some feminist theories reject the notion that intimate partner homicide – and more in general, violence against women – is caused by individual deviance or social disadvantage, as offenders should not be dissociated from the ‘average man’ (Ibid.). This approach is called ‘gender perspective’ and argues
14 that the root causes of intimate partner homicide, as well as violence against women, are to be found in the power structure that supports gender inequality (Kivivuori & Lethi, 2012, 73).
Serran and Firestone (2004) investigated the inherent differences in nature and extent of male and female intimate partner homicide and argued that women usually kill their partners in self-defence when the victims initiate the violence and they feel that their life (and/or their children’s) is threatened.
2.6 Exposure reduction and backlash/retaliation theories
In the spirit of ‘Verkko’s laws’, Messner and Savolainen (2001) studied gender differences in homicide in Finland and the United States and suggested that national differences in the status of women – i.e. economic independence and domesticity – may have consequences for the sex ratio of intimate partner homicide (Kivivuori et al., 2012). They argue that Finnish women are less constrained to tolerate abusive domestic unions and, consequently, less likely to kill their male partners in an act of self-defence. Consistent with this assumption, the Finnish male intimate partner homicide victimization rate was found to be significantly smaller in Finland (0.18) than in the United States (0.51). Conversely, among females the equivalent risk was higher in Finland (1.18) than in the United States (1.09), even though the overall rate of female homicide in Finland was lower (p.104). Consistent with these findings, Dugan et al. (1999, 188) found that the decrease in intimate partner homicide in the United States resulted primarily from reduced exposure to abusive or violent relationships. Exposure reduction can be considered a consequence of the changing living arrangements of men and women (declining domesticity, as measured by marriage and divorce rates), improvements in the economic status of women, and increases in the availability of domestic violence services.
The authors tested these conjectures in a sample of 29 large US cities over the period 1976 to 1992 and found that the benefits of exposure reduction consist primarily in averting male victimization and
15 sanctioning the female perpetrator; instead, except for domesticity, they found little evidence of exposure reduction affecting the female victimization rate (p.209). In a successive study, Dugan et al. (2003) found that, although several types of prevention resources are linked to lower levels of intimate partner homicide, other resources are related to higher levels of homicide. This finding seems to suggest ‘a retaliation effect when interventions stimulate increased aggression without adequately reducing exposure’ (p.169). In other words, ‘too little exposure reduction […] may be worse than none at all’ (Ibid.).
2.7 Routine activity and lifestyle theories
Criminal routine activity theory, a sub-field of opportunity theory, sees criminal behaviour as an outcome of ‘opportunities arising from the structure and dynamics of ordinary routines’ (Kivivuori et al., 2012, 104). Although this theory has not been widely used in European homicide studies so far, two central topics of contemporary homicide research – namely alcohol consumption and gun availability – are examples of legal ‘routine activities’ that tend to increase the lethality of violent interactions (Ibid.).
To explain the homicide trends observed in Western Europe in the post-war period, Aebi and Linde (2014) adopted the theory of personal crime, best known as lifestyle theory, which has often been combined with the routine activities approach to explain the risk of victimization. The lifestyle approach, in fact, helps explain variations in homicide rates after the 1960s and after the 1990s, as both have been central turning points in the way of life of Western European populations. However, Aebi and Linde (2010; 2014) updated both theories to reflect the new range of opportunities and lifestyle changes introduced by computer technologies and the development of the internet. These developments, in fact, led to an increase in the time spent at home, and as going out often usually increases the risk of victimization, this may partly explain the decrease in homicide rates after the 1990s (Ibid.). On the other hand, intimate partner homicides tend to be committed at home. Consequently, we may hypothesize that the intimate partner homicide rate is related to the percentage
16 of the population who use the computer and internet often and the percentage of households that are connected to the internet over a broadband connection.
2.8 Critical assessment of the theories
Verkko’s laws (1951) and the male sexual proprietariness theory (Wilson & Daly, 1996; 1998) can be considered classic theories of intimate partner homicide research. Verkko’s laws found some support in homicide research (Marshall & Summers, 2010; Stockl et al., 2013) and will be further tested in the present study. The male sexual proprietariness theory borrows from evolutionary theory and argues that lethal and non-lethal acts of partner violence are adaptive responses by males confronted with perceived threats to their relationship, such as sexual competition and infidelity (Wilson & Daly, 1996; 1998). According to the ‘gender perspective’, acts of partner violence are instead a product of the patriarchal society we live in, with its power structure and culture, and are not therefore simply an ‘adaptive response’ (Kivivuori & Lethi, 2012). In line with this perspective, the self-defence theory underlines the differences between male- and female-perpetrated partner homicide, with the latter often caused by prior long-standing domestic violence (Dugan et al., 1999; 2003). Moreover, exposure reduction and backlash/retaliation theories point to the role of gender (in)equality (Serran and Firestone, 2004).
Whereas the ‘gender perspective’ argues that the origin of intimate partner violence is to be found in the gendered power structure of our society and therefore regards all social classes, the ‘violence perspective’ maintains that intimate partner homicide perpetrators tend to belong to the most disadvantaged strata of the society, like other types of homicide perpetrators (Kivivuori & Lethi, 2012). (Socio-economic) strain theory is in line with this perspective. Since the results of prior research in this regard are at times contradictory, socio-economic indicators are tested in the present study to verify the validity of this theory. Finally, routine activity theory has been included to account for the correlation between alcohol consumption and homicide, whereas lifestyle theory may relate
17 to intimate partner homicide as an increase in computer use corresponded to more time spent at home, which is where most intimate partner homicides occur.
3 Empirical review 3.1 Background
As mentioned in the introduction, one in seven homicides worldwide (13.5%) is committed by an intimate partner and women run the highest risk of being killed by an intimate partner (Stockl et al., 2013, 863). Although homicide rates in Europe have been decreasing since the XIV century – except for the period 1960s-1990s – this decline overwhelmingly regarded male-to-male lethal violence (Spierenburg, 2012). Consequently, the proportion of intimate partner homicides has been increasing from the middle ages until modern times (Ibid.).
Similarly, the drop in homicide rates after the 1990s in Italy – mainly driven by changes in the nature of organized crime – has corresponded to an increase in the proportion of domestic homicides (Barbagli, 2004). Today, the main category of homicides in Italy – which used to be that of homicides related to the organized crime – is that of domestic homicides: more than two thirds of all homicides (39.5%) happen within a domestic context, half of which are committed by intimate partners (Barbagli, 2004; EURES, 2015). Consistent with Stockl et al.’s (2013) findings, intimate partner homicide in Italy disproportionately affects women to the extent that, in 2014, more than half of female homicide victims (53.3%) were killed by an intimate partner (EURES, 2015). However, although these findings underline the importance of tackling intimate partner homicide, research on its occurrence in Europe – and in Italy – is limited.
‘Relative to the United States and several Commonwealth countries, Europe does not have a long tradition of studying the trends, patterns and explanations of homicide’ as ‘differences in legal definitions and data sources have so far hampered cross-national comparisons’ (Liem & Pridemore,
18 2012, 3). Trent and Pridemore (2012) carried out a thorough review of cross-national homicide research in Europe and concluded that the literature as a whole appeared directionless, research lacked repetition and it was rare for the same studies to use the same measures. Save for the association between economic inequality and violent crime, the literature failed to converge.
Similarly, Aebi and Linde (2014) assessed the state of research on correlates of homicide across nations and found that there is no agreement on which are the best predictors of homicide. Moreover, they tested a series of variables identified as potential predictors by prior research and found no correlation with homicide rates in Western Europe. The variables they tested are the following: percentage of population aged 0-14, percentage of that age group that are males, percentage of urban population, average population density per km2, unemployment rate, GDP per capita in US dollars, percentage of the population with post-secondary education aged 25 or over, HDI, infant deaths per 1000 births, alcohol consumption in litres per capita for the population aged 15 or over, and the Gini index (Aebi & Linde, 2014).
European research on intimate partner homicide is even rarer and more fragmented due to problems of data availability and comparability: databases on the international (e.g. Centro Reina Sofia) and regional (e.g. Eurostat) level have started collecting data on the victim-offender relationship only in recent years and data are scarce and discontinuous. For instance, Eurostat only covers the period 2008-2014 – and for many countries data is not available for each year – while the Centro Reina Sofia collected data on intimate partner female homicide victimization on the international level only for three years: 2000, 2003, 2006. As far as national sources are concerned, they lack comparability due to differences in legal definitions and methodologies used.
Nevertheless, the present chapter provides an overview of the few studies on intimate partner homicide in Europe published after 2000 in peer-reviewed academic journals. The articles discussed here have been primarily collected via the search engine Google Scholar and the online Catalogue of
19 Leiden University. Additional studies have been found by scanning the reference lists of such articles. Due to the obstacles to cross-national comparisons mentioned above, each of the studies presented focuses on a single European country. Four of the studies focus on the United Kingdom (Dobash et al., 2004; 2007; 2009; Dobash & Dobash, 2011), two regard Sweden (Caman et al., 2016a; 2016b), while the others focus on Denmark (Leth, 2009), the Netherlands (Liem & Roberts, 2009), Italy (Russo et al., 2009), Finland (Kivivuori & Lethi, 2012), Spain (Torrubiano-Domínguez et al., 2015) and Portugal (Cunha & Abrunhos Gonçalves, 2016).
3.2 ‘Conventionality’ of intimate partner homicide perpetrators
Many studies included in the present review (Dobash & Dobash, 2011; Dobash et al., 2004; Kivivuori and Lethi, 2012; Caman et al., 2016a; Leth, 2009) compare intimate partner homicide perpetrators to other types of homicide perpetrator, to investigate the supposed relative ‘conventionality’ of the former when compared to the latter. To further question the relative ‘conventionality’ of intimate partner homicide perpetrators, Dobash et al. (2007) and Cunha and Abrunhosa Gonçalves (2016) compare them to non-lethal intimate partner violence offenders, whereas Dobash et al. (2009) investigate the differences between intimate partner offenders with and without prior convictions. These studies can be situated within the debate on the ‘gender’ and the ‘violence’ perspectives: what they aim to discover is whether intimate partner homicide perpetrators belong to the disadvantaged strata of the society or whether they are more ‘conventional’ than other types of homicide perpetrators.
What emerges from the first group of studies is that, though ‘deviant’ from the average population in terms of ‘social disadvantage and personal maladjustment’, partner homicide offenders tend to be more ‘conventional’ than men who kill other men (Kivivuori & Lethi, 2012, 73-75). Dobash et al. (2004) compared a sample of 106 men who killed an intimate partner with 424 men who killed other men and found that, although both groups had more problems in the family of origin and in their childhood than might be expected in the general population, the first group tended to be more
20 ‘conventional’ than the second. Similarly, as adults, the intimate partner group ‘was again more conventional in terms of the level of education, employment, persistent criminal behaviour, and general use of physical violence’ (pp.599-600).
Kivivuori and Lethi (2012), who compared a sample of men who killed an intimate partner to other types of homicide perpetrators in Finland, found significant ‘differentials in how strongly various types of homicide were related to social disadvantage and individual traits’ (p.74). If male offenders killing unrelated men were the most socially marginalized and filicide offenders were the least ‘deviant’ group, male and female partner homicide offenders – as well as males killing male relatives – were consistently between these polar opposites (Ibid.). Consistent with these findings, in the Swedish context, Caman et al. (2016a, 7) found that intimate partner homicide perpetrators were ‘less disadvantaged with regard to employment, accommodation, and criminal history’ than other types of offenders.
3.3 Intimate partner homicide and alcohol consumption
Dobash et al. (2004) found that men who kill other men were ‘more likely to have been drunk or to have been using drugs and to use instruments or weapons’ than intimate partner homicide offenders (p.600). Although ‘chronic alcohol abuse was a problem in both groups’, it was more so in the male-to-male homicide group (Ibid.). Kivivuori & Lethi (2012), however, found that in Finland men who killed an intimate partner tend to resemble other offenders in terms of alcohol and drug abuse and ‘they too were likely to be associated with situational dynamics involving offender and victim alcohol intoxication’ (p.73). Interestingly, Leth (2009, 201) found that in Southern Denmark ‘significantly more victims of intimate partner homicide had a history of chronic substance abuse, primarily of alcohol, than victims of other types of homicide’.
21 3.4 Lethal and non-lethal intimate partner violence
The picture of conventionality of intimate partner homicide offenders further alters when the stability of intimate relationships and violence to previous women partners are taken into account. In fact, men who kill their intimate partner appear to experience more of these problems and tend to ‘specialize’ in violence against women, particularly intimate partners (Dobash et al., 2004). Most perpetrators of intimate partner homicide tend to have ‘problems in intimate relationships and a history of serious, repeat abuse of the woman they kill’ and to manifest ‘controlling behaviour as well as jealousy and possessiveness’ (Dobash & Dobash, 2011, 130). Similarly, Leth (2009) found that many intimate partner homicides in Southern Denmark were cases of fatal wife battering or caused by the woman’s wish to leave the relationship.
Dobash et al. (2004, 601) suggested that there may be two types of men who kill an intimate partner, one more closely resembling the stereotype of the murderer or batterer and one more closely resembling the ‘conventional man’ or the ‘ordinary guy’. Kivivuori and Lethi (2012, 74), as well, conceded that although intimate partner homicide is associated with social disadvantage, psychological stress, substance abuse and sustained criminality like other subcategories of homicide, this does not contradict the existence of a minority of partner homicide offenders who may be described as ‘nonpathological ‘family only’ types’.
Dobash et al. (2009) speculated such ‘conventional’, ‘family only’ types may be identified by the absence of a previous conviction as an adult for any type of offence. To investigate this hypothesis, they compared perpetrators with (at least) a previous conviction – i.e. ‘criminogenic’ – with those without a previous conviction – i.e. ‘conventional’. ‘Conventional’ perpetrators did not experience various childhood problems considered as precursors or correlates of serious crime/violence or were less likely to do so. As adults, they were ‘significantly more likely to be regularly employed, to obtain an educational qualification, to be in a marital relationship, and less likely to have problems such as
22 substance abuse’ (p.215). However, ‘nearly three quarter of these men had ongoing disputes within the relationship prior to the murder event and there was little difference between the two groups in terms of the ‘risky’ issues of jealousy/possessiveness and/or separation’ (pp.215-216). Even more interestingly, nearly half of the ‘conventional’ group had ‘previously used violence toward the victim’, though ‘this violence had gone unreported or undetected’ and they ‘initially appeared to have no history of this offense’ (p.216). Therefore, rather than ‘conventional’, this group would be more accurately described as ‘abusers turned murderers’ (Ibid.).
In another study, Dobash et al. (2007) compared men convicted of nonlethal violence against an intimate partner with those convicted of murdering an intimate partner. Although the backgrounds of both groups were more problematic than would be expected in the general population, the circumstances of those who killed were relatively more ‘conventional’ compared to those who committed non-lethal violence. The backgrounds of abusers in fact more closely resembled those of offenders, characterized by unemployment (or unskilled jobs), at least one previous conviction, a conviction for violence, and abuse of alcohol. Moreover, they were ‘likely to come from homes where their father had an alcohol problem and physically abused their mother’, while men who murdered a partner were ‘likely to have grown up in households were their mother was a homemaker and their farther had a skilled or white-collar job’ (pp.345-346). Similarly, Cunha and Abrunhosa Gonçalves (2016, 16) found that in Portugal intimate partner homicide perpetrators were ‘less likely to experience and/or witness violence in childhood, to abuse substances, to present a history of violating conditional release or restriction orders, and to minimize and deny spousal assault history’ than intimate partner (non-lethal) violence perpetrators. However, ‘men who murdered a partner were much more likely to have used violence against a previous intimate partner’ and, although 41% of those men had no history of violence against the woman they killed, this could likely be due to underreporting (Dobash et al., 2007; 2009).
23 Regarding the circumstances at the time of the violent event, those who killed were ‘less likely than abusers to be drunk’, thought ‘the conflict was more likely to involve possessiveness’ and the couples were ‘more likely to be separated or separating’ (Dobash et al., 2007, 346). The dynamic of this process may be one in which the man comes to believe that ‘the relationship is truly over and, therefore, becomes estranged from the woman’ (p.348). ‘At this point, he may ‘change the project’ from attempting to persuade, coerce, control, and punish her to ‘eliminating’ her’, as it is captured by the common phrase ‘If I can’t have you, no one can’ (Ibid.). Lethal violence was also ‘more likely to include a sex assault, strangling, and the use of an instrument or knife’, further representing the change in the offender’s orientation to the victim, toward an ‘objectification of the woman and/or alienation from her and the relationship’ (p.346). Although many of these findings confirm conventional knowledge and previous research, they also challenge the notion of a single progression from nonlethal to lethal violence.
3.5 Male and female perpetrators of intimate partner homicide
It has previously been noted that male and female perpetrators of intimate partner homicide seem to belong to distinct groups (Caman et al., 2016b, 30; Serran & Firestone, 2004). In a nationwide study of all female- (n = 9) and male-perpetrated (n = 36) intimate partner homicides committed in Sweden between 2007 and 2009, Caman et al. (2016b, 30) found a significant difference with regard to occupation: the majority of female perpetrators were in fact unemployed, compared to less than one-third of male perpetrators. It has been argued that economic disadvantage is essential in female offending, as ‘due to economic dependence on the partner and lack of resources’ these women might be ‘unable to leave destructive relationships’ (Ibid.). Caman et al. (2016b, 30) also observed that, although neither female nor male perpetrators of partner homicide seem to be particularly characterized by mental health issues, ‘more than half of females had previously been diagnosed with a substance abuse disorder’ and ‘the majority of female perpetrators were intoxicated’ during the offense. Moreover, they found that ‘more than half of the female perpetrators had been threatened
24 and physically abused by the male victims, as opposed to a minority of the male perpetrators’, which is consistent with self-defence explanations (p.31). This finding is also supported by Kivivuori and Lethi (2012). Consistent with prior research, both Caman et al. (2016b) and Russo et al. (2009) found that male perpetrators commit the crime within the context of separation, while females commit the offense in intact relationships. Surprisingly, however, Caman et al. (2016b) ‘found that most male victims had been threatened, and half of them physically abused, by the female perpetrators’ (Ibid.).
3.6 Intimate partner homicide-suicide
Liem and Roberts (2009), Leth (2009) and Caman et al. (2016a) found that homicide-suicide was more common among intimate partner homicides than in other types of homicide. Caman et al. (2016a) observed that in Sweden male-perpetrated intimate partner homicide constituted more than half of all homicide-suicide cases and Russo et al. (2009) found that in Italy 41.8% of cases of intimate partner homicide were followed by suicide or suicide attempt. Caman et al. (2016b, 31) also found that ‘20% of the male-perpetrated cases constituted homicide-suicides, while none of the female perpetrators committed suicide in connection to the homicide’. Russo et al. (2009) as well found that the perpetrators of homicide-suicide were predominantly male. Moreover, Caman et al. (2016a) suggested ‘potential associations between the use of firearms and homicide-suicide’ in intimate partner homicide cases. Liem and Roberts (2009) found a ‘prevalence of unemployment, depressive disorder, and previous suicide threats and suicide plans’ among perpetrators of intimate partner homicide committing a self-destructive act, suggesting that ‘these individuals have more in common with suicide victims than with homicide perpetrators’ (pp.348-349).
3.7 Critical assessment of the literature
In sum, most of the studies discussed in this chapter focused on comparing intimate partner homicide with other types of homicides (Caman et al., 2016a; Dobash et al., 2004; Kivivuori & Lethi, 2012; Leth, 2009). Many also compared lethal and non-lethal intimate partner violence (Cunha & Abrunhos
25 Gonçalves, 2016; Dobash et al., 2007) or male- and female-perpetrated intimate partner homicide (Caman et al., 2016b; Russo et al., 2009). One study in the Netherlands (Liem & Roberts, 2009) compared intimate partner homicide with and without self-destructive act, while Dobash et al. (2009) compared intimate partner homicide offenders with and without prior convictions. Dobash & Dobash (2011) investigated the cognitions of male perpetrators of intimate partner homicide in the UK, whereas one study in Spain (Torrubiano-Domínguez et al., 2015) determined whether the increase in unemployment due to the financial crisis affected intimate partner homicide rates.
What emerges from this review is that intimate partner homicide perpetrators, though ‘deviant’ from the average population in terms of ‘social disadvantage and personal maladjustment’, are in general less ‘disadvantaged’ and more ‘conventional’ than men who kill other men (Kivivuori & Lethi, 2012, 73-75). However, this picture of conventionality tends to alter when the stability of intimate relationships and violence to previous women partners are taken into account (Dobash et al., 2004). Although there seem to be two types of men who kill an intimate partner – one more closely resembling the stereotype of the murderer or batterer and one more closely resembling the ‘conventional man’ – Dobash et al. (2009, 216) found that nearly half of the offenders considered ‘conventional’ had ‘previously used violence toward the victim’, though ‘this violence had gone unreported or undetected’. Nevertheless, differences between perpetrators of lethal and non-lethal partner violence challenge the notion of a single progression from nonlethal to lethal violence.
These results seem to provide some support for the ‘violence perspective’, as all types of homicide perpetrators tend to be more ‘disadvantaged’ than the average population. However, the fact that intimate partner homicide perpetrators tend to be less ‘disadvantaged’ than men who kill other men seems to lend some support to the ‘gender perspective’ as well, underlining specificities of this type of homicide relative to others. Further support for this perspective is provided by the link between prior domestic violence and intimate partner homicide, as well as by the difference between male and female perpetrators.
26 As mentioned above, research on intimate partner homicide in Europe is extremely rare and each study is focused on a single country, due to differences in definitions and statistics. Moreover, these studies analyse different aspects of the phenomenon, from different points of view, and using different measures and indicators. Most studies – all except the one conducted in Spain by Torrubiano-Domínguez et al. (2015) – are based on a micro-level analysis (the perpetrators, the victims and/or the incidents). The present thesis aims to expand the knowledge on intimate partner homicide in Europe by using national and regional rates of intimate partner homicide as the level of analysis (macro-level) and by testing a wide range of potential predictors derived from European homicide studies (Liem & Pridemore, 2012; Aebi & Linde, 2014) and from the studies presented in this review. For the reasons discussed above, the focus is still on a single country – Italy – though the aim is for this study to serve as foundation for future comparative research.
4 Methodology 4.1 Background
As outlined in the previous sections, the aim of the present thesis is to test the main demographic and socio-economic predictors of homicide identified by European homicide research (Liem & Pridemore, 2012; Aebi & Linde, 2014) and some additional indicators derived from empirical and theoretical research on intimate partner homicide (see chapters 2 and 3) to verify whether they are correlated with the intimate partner homicide rate in Italy. The analysis is based on quantitative methods and combines a longitudinal and a cross-sectional approach. The former is employed to study correlates of intimate partner homicide rates in Italy between 2002 and 2014, whereas the latter is adopted to analyse correlates of intimate partner homicide rates in 17 Italian regions in 2013. Three regions – namely Valle d’Aosta, Molise and Basilicata – have been excluded because they have a very small population (less than 600,000) and because figures of intimate partner homicide are either not available or unreliable for these regions.
27 The decision to focus on a single country is motivated by the fact that data on homicide by victim-offender relationship is very scarce in international databases and comparing national data directly is problematic due to differences in definitions and methodologies. Although Eurostat has recently started to collect this type of information, there is still a large amount of missing data and the period covered is only from 2008 to 2014 (Eurostat, 2016). On the other hand, by focusing on a single a country it is possible to use national sources (e.g. Istat and EURES) and thus cover a longer period of time (2002-2014). Following a convenience approach, Italy is a natural choice as it is my country of origin and Italian is my mother tongue. Being able to understand the language of the country under study allows the consultation of a larger number of sources and improves the quality of the research. Moreover, although the focus is on a single country, the aim is for this study to serve as foundation for future comparative research on intimate partner homicide in Europe.
As ‘most research on the correlates of homicide is cross-sectional’ (Aebi & Linde, 2014), this thesis adopts, in the first part of the analysis, a longitudinal approach which can offer a different perspective on the problem of intimate partner homicide. Among the studies discussed in the literature review, the only one which adopts a longitudinal approach and is based on a macro-level analysis is the one conducted by Torrubiano-Domínguez et al. (2015). Differently from Torrubiano-Domínguez et al.’s (2015) study, which analyses the correlation between unemployment rates and percentage of foreign population in 17 Spanish regions and the rate of female intimate partner homicide victimization, the present study includes a wider range of demographic and socio-economic indicators (see sections 4.4 and 4.5).
The second section of the analysis adopts a cross-sectional approach to examine rates and correlates of intimate partner homicide in the 17 Italian regions under study for the year 2013. The study is limited to a single year because data on the regional distribution of intimate partner homicide is only available for that year on the Istat database. Nevertheless, a cross-sectional analysis of rates and correlates of intimate partner homicide in different regions is of interest because, depending on the
28 indicator, inter-regional differences in Italy can be great. Moreover, by testing some of the predictors with two different approaches, more interesting results can be obtained.
4.2 Data sources
The present research is based on administrative data collected from three main sources: Istat, Eurostat and the WHO European Health for All database (WHO/Europe, 2016).12 Istat, the Italian national institute of statistics, collects data on homicide from the Ministry of Interior. In the Italian penal code, intentional homicide does not include either infanticide or assault leading to death (i.e. omicidio preterintenzionale), therefore Istat presents these data separately.3 Moreover, the data presented refers to the incidents reported by the police to the judiciary and therefore excludes all cases reported directly to the judiciary or those discovered by the judicial authorities themselves.4 The term intimate partner homicide used by Istat is inclusive and indicates the killing of a former or current spouse, cohabiting partner, or non-cohabiting partner (i.e. boyfriend or girlfriend).5 This source has been selected because of its reliability – it is the national institute of statistics and it processes data obtained from the Ministry of Interior – and because the only two other sources (EURES, 2013; Eurostat, 2016) which offer data on intimate partner homicide in Italy cover much shorter periods of time. This source is also used for data concerning some independent variables, especially those relative to the regional level.
Eurostat defines intentional homicide as ‘unlawful death inflicted upon a person with the intent to cause death or serious injury’ and includes murder, voluntary manslaughter, serious assault leading
1 Istat: http://www.istat.it/en/ 2 Eurostat: http://ec.europa.eu/eurostat 3 Istat: http://noi-italia2015.istat.it/index.php?id=7&L=0&user_100ind_pi1%5Bid_pagina%5D=60&cHash=7e0eecde7d3a42352e67dc89 b91c015c 4 Istat: http://seriestoriche.istat.it/index.php?id=6&user_100ind_pi1%5Buid_categoria%5D=6&cHash=ba577ac006086d6e2a 2b3ace2c5808b5 5 Istat: http://noi-italia2015.istat.it/index.php?id=7&user_100ind_pi1%5Bid_pagina%5D=120&cHash=d46664a80cf0447a044e2f8e2d61 4392
29 to death and infanticide (UNODC, 2015). The definition of intimate partner homicide is consistent with that used by Istat (Ibid.). Data are provided to Eurostat by a network of national contact persons designated by each country. The contact persons are responsible for collating the data from various national sources, such as the Police, the Ministry of the Interior, the Ministry of Justice and the National Prison Administration6. Eurostat has been used as a source for those data not available on Istat databases, or when Istat only covered a short period of time for an indicator. When complete data were not available either on Istat or Eurostat databases, the WHO European Health for All Database (WHO/Europe, 2016) has also been employed.
4.3 Dependent variables
For the longitudinal analysis, the dependent variables used are intimate partner homicide victimization rates from 2002 and 2014, disaggregated by gender. Data are retrieved from Istat (2015b) from 2002 and 2013, while data for 2014 are retrieved from Eurostat (2016). For the cross-sectional analysis, the dependent variables are intimate partner homicide rates disaggregated by gender and by region, provided by Istat (2005a; 2005c).
4.4 Independent variables
The selection of independent variables is based on European research on homicide and intimate partner homicide, both empirical and theoretical. Among the indicators described in Chapter 3, only those consistent with theoretical and empirical research on intimate partner homicide and for which data was available for Italy and for the entire period covered in this study have been selected. In addition, a few more indicators have been derived from intimate partner homicide research and included in this study.
To test the reliability of Verkko’s laws (1951), the present study compares the Italian homicide rate from 2002 to 2014 with the proportion of homicides committed by an intimate partner, as well as
30 with the proportion of homicide victims that are female. As far as the longitudinal analysis is concerned, the following indicators are used to measure economic strain: unemployment rate, GDP per capita in €, Gini coefficient, infant mortality rate and household relative poverty incidence. Data for all indicators are retrieved from Istat, except for the infant mortality rate, which has Eurostat as data source. The percentage of population aged 25-64 with at most lower secondary education attainment is used as measure of social disadvantage and is retrieved from Istat as well.
Population density per km2, population growth rate, crude divorce rate and natural increase rate are used as measures of social disorganization. Moreover, as some studies have found a relation between ethnic heterogeneity and homicide rates (Kivivuori et al., 2012) and others have found a large proportion of crimes in Italy to be committed by foreigners (Barbagli, 2004, 153), the indicator percentage of foreign residents has been included as well. The data source for all the predictors mentioned above is Istat, except for the crude divorce rate, which is retrieved from Eurostat. Furtermore, to test the exposure reduction theory, the present study uses crude marriage and divorce rates as indicators of levels of domesticity, and the percentage of women aged 30-34 with tertiary education attainment as indicator of improvement in women’s status. The latter has Istat as data source, whereas the crude marriage rate is retrieved from Eurostat.
The indicator alcohol consumption in litres per capita per population aged 15 or over, related to routine activity theory, has the WHO European Health for All Database as source, while the percentage of individuals who use a computer and the percentage of those who use internet, relative to lifestyle theory, are retrieved from Istat. Finally, drawing from male sexual proprietariness, the present study tests the correlation between the intimate partner homicide rate and the percentage of births outside of marriage, which has Istat as a source as well.
As far as the cross-sectional analysis is concerned, all data is collected from Istat, since it is the only source which contains data for Italy by region. Most indicators are the same than those used for the
31 longitudinal analysis: unemployment rate, GDP per capita in €, household relative poverty incidence, population density per km2, percentage of the population (and of women) aged 25-64 with at most lower secondary education attainment, percentage of the population (and of women) aged 30-34 with tertiary education attainment, the natural increase rate and the percentage of residents that are foreigners. Moreover, the percentage of the population aged 11 or over who consume alcohol between meals is used as a measure of alcohol consumption, the international migration rate as additional measure of social disorganization, and the percentages of individuals aged 6 or over who accessed the internet at least once and at least every day in the past 12 months, together with the percentage of households that are connected to the internet over a broadband connection, are used to test lifestyle theory (i.e. following the development of internet people spend more time at home, which is also the place where most intimate partner homicides occur). Finally, since many studies have found prior non-lethal intimate partner violence to be a major predictor of partner homicide, the percentages of women aged 16-70 who experienced (sexual or physical) intimate partner violence during their lives, over the past 5 years or over the past year (data collected in 2014) are also included as potential predictors of intimate partner homicide.
4.5 Operationalization
Theory Concepts Indicators Data sources
Verkko’s laws (Verkko, 1951) The proportion of female homicide victims is higher when the overall homicide rate is low, and vice versa (Kivivuori et al., 2012, 96).
Homicide rate; Proportion of female
homicide victims.
Istat
The proportion of intimate partner homicides is higher when the overall homicide rate is low and vice versa (Kivivuori et al., 2012, 96).
Homicide rate; Proportion of intimate
partner homicides.
Istat
Economic strain theory + ‘Violence perspective’
Intimate partner homicide is concentrated in low social strata and related to economic deprivation like other types of homicide (Kivivuori & Lethi, 2012, 73).
Unemployment rate; Gini coefficient; GDP per capita; Infant mortality rate; Household relative
poverty incidence; Percentage of population
aged 25-64 with at most
Istat Eurostat
32
lower secondary education.
Social disorganization theory ‘In criminology, the concept of social disorganization refers to deficits in the normative regulation of human behaviour’, as measured by family instability, population change, population density etc. (Kivivuori et al., 2012, 98). Increases in social disorganization can result in elevated levels of criminal offending (Ibid.).
Population density per km2;
Population growth; Natural increase rate; International migration
rate;
Crude divorce rate.
Istat Eurostat
Percentage residents that are foreigners
Istat
Male sexual proprietariness theory
Sexual proprietariness is aroused when men feel their sexual monopoly over their partner is threatened, due to her fertility and attractiveness and due to pressure from potential rivals. Non-lethal violence and threats serve to deter sexual rivals and to limit female autonomy, but when such measures do not work, men can even decide to commit murder (Wilson and Daly, 1996; 1998).
Share of births outside marriage;
Percentage of women who have been victim of intimate partner violence during their lives, over the past 5 years, or over the past year.
Istat
Self-defence theory On the other hand, when women kill their partners, it is almost always as self-defence. The male partner initiates the violence and the female partner feels that her life – or her children’s – is threatened and reacts (Serran & Firestone, 2004).
Exposure reduction theory ‘The decrease in intimate partner homicide results primarily from reduced exposure to abusive or violent relationships’, which in turn is a consequence of the
Marriage rates; Divorce rates; Percentage of women
aged 25-64 with post-secondary education attainment.
Istat Eurostat
33
changing living arrangements of men and women (declining domesticity, as measured by marriage and divorce rates), improvements in the economic status of women, and increases in the availability of domestic violence services (Dugan et al., 1999, 188).
Backlash/retaliation theory Some resources ‘are related to higher levels of homicide, suggesting a retaliation effect when interventions stimulate increased aggression without adequately reducing exposure’ (Dugan et al., 2003, 169).
Criminal routine activity theory Criminal behaviour is ‘an
outcome of opportunities arising from the structure and dynamics of ordinary routines’ (Kivivuori et al., 2012, 104). Alcohol consumption is an example of a legal ‘routine activity’ that can increase the lethality of violent interactions (Ibid.).
Consumption of alcohol (in litres per capita, population aged over 15); Percentage of population
over 11 per alcohol consumption between meals.
Istat; WHO/
Europe
Lifestyle theory The development of computer technologies and the internet led to an increase in the time spent at home (Aebi & Linde 2010; 2014), which is the place where most intimate partner homicides occur.
Percentage of individuals aged 3 or over who use a computer;
Percentage of individuals aged 6 or over who use internet;
Percentage of individuals aged 6 or over who accessed the internet at least once in the past 12 months;
Percentage of individuals aged 6 or over who accessed the internet at least every day in the past 12 months;
Percentage of households that are connected to the internet over a
broadband connection.
34 4.6 Data analysis
All data is organized with Excel and when values are missing for one or two years, these are provided either by extrapolation or by linear interpolation. These are standard procedures for the replacement ofmissing data which are used, for example, by the World Health Organization (Aebi & Linde, 2010, 253). Moreover, as far as the longitudinal analysis is concerned, the data collected are first visualized in graphs, then bivariate correlations (Spearman’s rho) are conducted with the programme R to study eventual correlations between the intimate partner homicide rate and the predictors. Similarly, with regard to the cross-sectional analysis, bivariate correlations (Spearman’s rho) are conducted with the programme R. A time lag is also taken into account with regard to the variables unemployment rate and GDP per capita.
5 Results
5.1 Homicide and intimate partner homicide trends in Italy (2002-2014)
In line with the trends discussed in the introduction (see section 1.2), over the period 2002-2014 the homicide rate in Italy dropped from 1.13 to 0.80 per 100,000 population. Conversely, the intimate partner homicide rate remained quite stable, around a mean value of 0.15 per 100,000 population (see Fig.1).
35 Consistent with this finding, and in support of Verkko’s laws (1951), the proportion of intimate partner homicides to total homicides increased over time, from 12.4% in 2002 to 18.8% in 2014. Conversely, the proportion of female homicide victims to total homicides shows a slightly different trend, decreasing from 2002 to 2005, with a sudden peak in 2006, then starting to increase more steadily from 2007 (see Fig.2).
36 Figure 3 compares the trends of the intimate partner homicide victimization rate disaggregated by gender to the overall intimate partner homicide rate. What appears quite evident is that the female victimization rate and the overall intimate partner homicide rate have a similar trend, which is not surprising if we consider that most victims of intimate partner homicide in Italy are female. On the other hand, the differences in trend between the overall intimate partner homicide rate and the male intimate partner homicide victimization rate may be due to the extreme rarity of latter. In fact, values that are very small – i.e. the male intimate partner homicide victimization rate ranges from 0.03 to 0.09 per 100,000 male population – can show a greater variability from one year to the other.
In further support of Verkko’s laws (1951), the results of the bivariate correlations (Spearman’s rho) between intimate partner homicide victimization rate by gender, homicide rate and proportions of female homicide victims and intimate partner homicides (see Tab. 1) are consistent with what can be observed in Figures 1-3. In fact, there is a strong positive correlation between the intimate partner homicide female and overall victimization rates (rho = .79; p ≤ .001) and a strong negative correlation between the homicide rate and the proportion of intimate partner homicides (rho = -.75; p ≤ .01). Moreover, there is a strong positive correlation between the male intimate partner homicide victimization rate and the homicide rate (rho = .72; p ≤ .01). However, as already noted, the fact that
