Predictors of Intimate Partner Homicide in a Sample of Portuguese Male Domestic Offenders

Introduction

In recent years, increasing attention has been given to intimate partner violence (IPV) and intimate partner homicide (IPH). Estimates indicate that 15% to 71% of women report experiencing physical and/or sexual violence by an intimate partner at some point in their lives (WHO, 2005). A recent systematic review revealed that 30% of all women in the world and 25.3% of European women experience intimate physical and/or sexual violence during their lifetime (WHO, 2013). Moreover, the 2009/10 British Crime Survey revealed that 29% of women had experienced any type of domestic abuse (emotional, financial, physical, sexual assault, or stalking) after the age of 16 (Hall, 2011). In Portugal, where this study was conducted, approximately 23,000 cases of IPV were reported to the police in 2014 (Portugal has approximately 10.5 million inhabitants; Sistema de Segurança Interna, 2015). A Portuguese inquiry conducted in 2007 revealed that 6.4% of women reported experiencing at least one act of physical, emotional, and/or sexual violence in the previous year by an intimate partner or ex-partner (Lisboa, 2009).

Despite the high prevalence of IPV, the number of women killed by an intimate partner is comparatively small, and IPH is quite infrequent. However, the consequences of such experiences are tragic. Females are more likely than males to be the victims of intimate killings (e.g., Cooper & Smith, 2011; McFarlane, Campbell, & Watson, 2002). In the United States, from 1976 to 2005, approximately 11% of all female murder victims were killed by an intimate partner (Nichols-Hadeed, Cerulli, Kaukeinen, Rhodes, & Campbell, 2012). In the 2009/10 British Crime Survey, among the 198 female victims aged 16 or above, more than half of them (54%; 94 offenses) were killed by their partner, ex-partner, or lover.

There are few studies on IPH in Portugal (e.g., Almeida, 2012; Pais, 2010). Although this phenomenon is rare in Portugal, its numbers raise serious concerns. The Portuguese non-governmental organization União de Mulheres Alternativa e Resposta (UMAR; Women Union Alternative and Response, 2014) reported 35 cases of IPH in 2014 (seven cases more than in 2013 and six cases more than in 2012). Among the 35 cases registered in 2014, 65.7% of the victims were currently in an intimate relationship with the perpetrator, and 34.3% were already separated or divorced from the offender. Overall, 336 women were killed in Portugal between 2004 and 2014 by a current or a former intimate partner (União de Mulheres Alternativa e Resposta, 2014).

If IPH is the last link in the chain of violent behaviors, it is important to determine the predictors of such behavior. Predicting the risk of IPV and IPH facilitates awareness of the problem and a search for best solutions (Echeburúa, Fernández-Montalvo, Corral, & López-Goñi, 2009; Snider, Webster, O’Sullivan, & Campbell, 2009). With this intent, several studies have analyzed the factors that are significantly associated with IPV and IPH.

Evidence from a wide variety of research states that although IPV can occur across all sectors of society, it is more likely to be associated with economic and social disadvantages. IPV perpetration tends to be associated with younger, undereducated, unemployed, and low socioeconomic status (SES) men (e.g., Ali, Asad, Mogren, & Krantz, 2011; Caetano, Vaeth, & Ramisetty-Milker, 2008; Campbell, Glass, Sharps, Laughon, & Bloom, 2007; Campbell et al., 2003; Capaldi, Knoble, Shortt, & Kim, 2012; Stith, Smith, Penn, Ward, & Tritt, 2004; Walton-Moss, Manganello, Frye, & Campbell, 2005). In contrast, a study conducted by Breitman, Shackelford, and Block (2004) revealed that perpetrators of IPH tend to be older. Findings concerning the influence of the type of union on IPH are mixed (e.g., Sebire, 2017). Various studies suggest that cohabiting (i.e., common-law or de facto marriage) partners have higher rates of IPV (e.g., Abramsky et al., 2011; Gass, Stein, Williams, & Seedat, 2011) and IPH (e.g., Aldridge & Browne, 2003; Mize, Shackelford, & Shackelford, 2009). Wilson, Johnson, and Daly (1995) concluded that wives incurred much greater risk of both lethal and nonlethal violence in common-law unions than in registered unions. Campbell et al. (2003) also found that couples who had never lived together were at a lower risk of IPH. Nonetheless, a study conducted by James and Daly (2012) revealed an equalization of the risk rate between union types in both the United States and Canada. Another important risk factor especially for IPH is separation from the intimate partner (e.g., Campbell et al., 2003, 2007; Campbell, Webster, & Glass, 2009; McFarlane et al., 2002; Wilson & Daly, 1993). In fact, Dobash, Dobash, Cavanagh, and Medina-Ariza (2007) found that men who killed their intimate partners are more likely to be separated at the time of the incident. More recently, a study conducted in London (Sebire, 2017) revealed that couples were more likely to be together than separated at the time of the incident. Ethnicity has also been seen as an important risk factor, with non-Caucasian men more likely to commit both lethal and nonlethal violence than Caucasian men (Wareham, Boots, & Chavez, 2009). Nonetheless, because the majority of research concerning ethnic origin has been conducted in the United States, it is difficult to analyze the accurate influence of this variable. In addition, as Sebire (2017) stated, the over-representation of some groups (e.g., African Americans) must be considered within the context of the social, cultural, economic, and political position of these groups within American history.

Substance abuse, particularly alcohol abuse, also appears consistently in the literature as a risk factor for domestic violence perpetration (e.g., Abramsky et al., 2011; Cummings, Gonzalez-Guarda, & Sandoval, 2013; Stith et al., 2004; Walton-Moss et al., 2005) and for intimate partner murder (e.g., Belfrage & Rying, 2004; Campbell et al., 2003, 2007; Echeburúa & Corral, 2009). In addition, a number of studies have described male abusers (Stith et al., 2004; Walton-Moss et al., 2005) and male murderers (Belfrage & Rying, 2004; Campbell et al., 2007; Echeburúa & Corral, 2009; Walton-Moss et al., 2005) as suffering from mental health problems. For example, Moffitt and Caspi (1999) found that 88% of men who perpetrated severe physical violence fulfilled the criteria for psychological disorders and presented a 13-time greater probability of mental disorder than non-offenders. Men who commit violence against their partners tend to show higher levels of anxiety, hostility, and impulsivity (e.g., Grann & Wedin, 2002; O’Leary, Slep, & O’Leary, 2007; Stith et al., 2004), depression (e.g., Corvo & Johnson, 2013; Dutton & Karakanta, 2013; Shorey, Febres, Brasfield, & Stuart, 2012; Stith et al., 2004), suicidal ideation/intent (e.g., Kropp, 2009), and pathological jealousy (e.g., Caetano et al., 2008). Other studies have noted the existence of personality disorders, such as borderline, narcissistic, and antisocial disorders, among batterers (e.g., Huss & Langhinrichsen-Rohling, 2000; Kim & Capaldi, 2004). Psychopathy also appears to be associated with violence in intimate relationships (e.g., Grann & Wedin, 2002; Harris, Hilton, & Rice, 2011). A history of criminal behavior is associated with both IPV (Abramsky et al., 2011; Huesmann, Dubow, & Boxer, 2009; Rodríguez et al., 2011) and IPH (e.g., Dobash, Dobash, Cavanagh, & Lewis, 2004; Grann & Wedin, 2002), and a previous arrest for IPV is a risk factor for intimate partner murder (Campbell et al., 2003).

The literature also suggests that attitudes supporting domestic violence (e.g., Abramsky et al., 2011; Campbell et al., 2003; Riggs, Caulfield, & Street, 2000; Stith et al., 2004) are important predictors of IPV. In fact, a study by Archer (2006) showed that sexist attitudes and the legitimation of violence against women are related to variations in victimization rates.

In addition, nonlethal violence against an intimate partner has consistently been linked to adverse and untoward experiences in childhood. For example, Lee, Walters, Hall, and Basile (2013) found that batterers with a history of violence in the family consistently showed negative attitudes toward women and feminine attributes. Other studies have also revealed that men who witnessed domestic violence and/or experienced physical violence in their childhood are significantly more likely to use violence against a woman partner in adulthood (e.g., Abramsky et al., 2011; Coleman & Widom, 2004; Cummings et al., 2013; White & Widom, 2003). Wareham et al. (2009) found that experiences of physical maltreatment in childhood were associated with severe violence.

Research on contextual and situational factors reveals both similarities and differences among lethal and nonlethal forms of violence. For example, previous physical violence is identified as one of the most important predictors of future violence in intimate relationships (e.g., Campbell et al., 2007; Cattaneo & Goodman, 2005; Dutton & Kropp, 2000; Kropp, 2009; Riggs et al., 2000; Stith et al., 2004). Factors such as gun possession and/or gun use, violent threats, threats with a weapon or use of a weapon, homicidal ideation or intent, sexual violence and forced sex, and threatening behaviors (e.g., Campbell et al., 2003, 2007, 2009; Echeburúa et al., 2009; McFarlane et al., 2002; Nicolaidis et al., 2003; Stith et al., 2004; Walton-Moss et al., 2005) appeared to be related to partner abuse. Stalking has also been related to a significant increase in spousal abuse (Norris, Huss, & Palarea, 2011; Walton-Moss et al., 2005) and is a strong risk factor for lethal violence (Campbell et al., 2007; McFarlane et al., 2002). In addition, estrangement, controlling behavior, violence during pregnancy, attempted strangulation, and a stepchild in the home were identified in a review conducted by Campbell et al. (2007) as risk factors that significantly increase the risk of IPH.

Despite the vast literature on risk factors for IPV and IPH, a central question is whether there are differences in risk factors relevant to IPV and IPH. In fact, the literature provides support for two main perspectives on IPH. One perspective points to intimate murder as a sudden or unexpected event without a previous history of violent behavior. Research on this perspective suggests that murderers may have mental health problems (e.g., Bartok & Bartok, 2005) or that contextual and/or situational factors (such as crisis, stress, or opportunity) can lead to the crime (Wiesburd & Waring, 2001). The other view states that murder is the culmination of a history of violence and is not itself a discrete phenomenon (Stark & Flitcraft, 1996). In fact, Wilson et al. (1995), comparing data concerning IPV and IPH, found that some demographic risk patterns (i.e., marital status and age disparity of marital partners) were similar for lethal and nonlethal incidents. Moreover, a recent qualitative study (Goussinsky & Yassour-Borochowitz, 2012) found that in terms of motive and emotional dynamics, female partner homicide is not different from other manifestations of violence against a female partner. However, the authors also found that the circumstances surrounding murder are distinct from those of IPV, and that in the majority of cases, lethal violence is not spontaneous but is a premeditated act. The literature also suggests that perpetrators of IPV are a heterogeneous group (e.g., Cunha & Gonçalves, 2013; Holtzworth-Munroe & Stuart, 1994). A report by Roehl, O’Sullivan, Webster, and Campbell (2005) found that some factors seem to generally be present among men who batter; however, there seem to be differentiating characteristics among those who abuse and those who commit homicide. For example, prior domestic violence and handguns are the most consistent and strongest risk factors for homicide. Although violence outside the home, alcohol abuse, and minority ethnicity are also implicated in IPH, these factors seem to have a weaker impact than for other types of homicide. The authors also concluded that although some risk factors for IPV and IPH are the same, there seems to be a difference of degree and some differential patterns. For example, substance abuse is more of a risk factor in IPV, and perpetrator suicidality is more of a risk factor in IPH. Dobash et al. (2007) compared men’s lethal and nonlethal violence against an intimate woman partner. They found that murderers are more conventional in terms of childhood backgrounds, education, employment, and criminal careers, and are more likely to be possessive and jealous and to be separated from their intimate partner at the time of the incident. IPV perpetrators are more likely to have used violence against a previous partner, to have sexually assaulted and strangled the victim, and to have used a weapon or instrument. On the contrary, murderers were less likely to have been drunk at the time of the event and to have previously used violence against the victim. Therefore, authors concluded that their findings do not support the notion of a simple progression from nonlethal to lethal violence.

Based on these contradictory findings, it is necessary to understand and clarify whether there are differences between the risk factors for IPV and the risk factors for IPH. Thus, a quantitative study using a specific instrument to analyze risk factors (i.e., Spousal Assault Risk Assessment Guide [SARA]) was conducted among Portuguese men who murdered their female partners or ex-partners and men who battered who were recruited from prison and the community. The objectives of this study were to identify the risk factors for IPV and IPH, to examine whether there are differences between intimate partner murder and violence against female partners in terms of risk factors and to identify the variables that best predict IPH.

Method

Results

Demographic and Penal Characteristics

The participants’ average age was 43.83 (SD = 11.1). At the time of the incident, most batterers were married or in cohabitation with the victim (n = 90; 52.3%). More than a half of the batterers had a second-grade education or less (n = 131; 76.2%), and belonged to a low socioeconomic level (n = 111; 64.5%). The most part of the participants were Caucasian (n = 162; 94.2%), had children (n = 156; 90.7%), and were employed during the crime perpetration (n = 103; 59.9%). Almost half of them had no previous convictions for spouse abuse (n = 84; 48.8%) and more than a half had no other convictions (n = 96; 55.8%).

T tests and chi-square tests were used to examine the association between IPH and demographic and penal variables. The results are displayed in Table 1. Analysis only revealed significant differences between IPV and IPH perpetrators’ age, t(170) = −2.640, p = .009, with a medium effect size, d = .47, and in marital status at the time of the offense, χ²(2) = 8.972, p = .011, with a small effect size, V = .23. So, IPH perpetrators were far more likely to be elders and singles compared with IPV perpetrators. No differences were found for length of relationship duration, education, employment, SES, children in home, ethnicity, other crimes, and previous convictions.

OPEN IN VIEWER

Table 1.

Comparisons of Demographic and Penal Characteristics of Intimate Partner Violence (IPV) and Intimate Partner Homicide (IPH) Perpetrators.

	IPV (n = 137)	IPH (n = 35)	t
	M (SD)	M (SD)
Age	42.72 (10.32)	48.17 (12.96)	−2.640**
Length of relationship duration	13.33 (10.56)	14.03 (11.27)	−0.343
	n (%)	n (%)	χ2
Marital status (at the time of the incident)
Married/cohabitating	70 (51.1)	20 (57.1)	8.972*
Single	14 (10.2)	9 (25.7)
Divorced	53 (38.7)	6 (17.1)
Education
Primary education	105 (76.6)	26 (74.3)	5.649
Sixth grade	15 (10.9)	2 (5.7)
Ninth/12th grade or higher	15 (10.9)	4 (11.4)
Incapable of reading and writing	2 (1.5)	3 (8.6)
Employment
Yes	80 (58.4)	23 (65.7)	0.622
No	57 (41.6)	12 (34.3)
SES
Low	92 (67.2)	19 (54.3)	4.332
Medium	28 (20.4)	13 (37.1)
High	17 (12.4)	3 (8.6)
Children
Yes	125 (91.2)	31 (88.6)	0.388
No	12 (8.8)	4 (11.4)
Ethnicity
Caucasian	131 (95.6)	31 (88.6)	4.410
Afro Portuguese	1 (0.7)	2 (5.7)
Gipsy	5 (3.6)	2 (5.7)
Other crimes
Yes	62 (45.3)	14 (40.0)	0.312
No	75 (54.7)	21 (60.0)
Previous convictions
Yes	71 (51.8)	17 (48.6)	0.118
No	66 (48.2)	18 (51.4)

Note. SES = socioeconomic status.

*

p < .05. **p < .01.

Risk Factors

The results regarding the SARA risk factors are displayed in Table 2. In the analysis of the associations between IPV and IPH, six factors were statistically significant: past violation of conditional release or community supervision (19.9% for IPV vs. 2.9% for IPH, p = .009), victim of and/or witness to family violence as a child or adolescent (61.8% for IPV vs. 40% for IPH, p = .017), recent substance abuse/dependence (55.1% for IPV vs. 37.1% for IPH, p = .043), recent suicidal or homicidal ideation/intent (22.8% for IPV vs. 82.9% for IPH, p = .000), past use of weapons and/or credible threats of death (50% for IPV vs. 97.1% for IPH, p = .000), and extreme minimization or denial of spousal assault (79.4% for IPV vs. 48.6% for IPH, p = .000). More specifically, IPH perpetrators were 16 times more likely to have suicidal and/or homicidal ideation and 35 times more likely to use weapons or credible threats of death than IPV perpetrators. Inversely, IPV perpetrators were 8 times more likely to have previous violations of conditional release or community supervision, 2 times more likely to be victims or witnesses of violence in childhood, 2 times more likely to abuse alcohol and other drugs, and 4 times more likely to minimize or deny a history of spousal assault.

OPEN IN VIEWER

Table 2.

Comparisons of Risk Factors of Intimate Partner Violence (IPV) and Intimate Partner Homicide (IPH) Perpetrators.

Variables	IPV (n = 137)	IPH (n = 35)	χ2(2)	OR
	n (%)	n (%)
Assault of family members
Yes	50 (36.8)	8 (22.9)	2.402	—
No	86 (63.2)	27 (77.1)
Assault of strangers
Yes	48 (35.8)	9 (25.7)	1.268	—
No	86 (64.2)	26 (74.3)
Conditional release violation
Yes	27 (19.9)	1 (2.9)	5.872**	8.33
No	109 (80.1)	34 (97.1)
Relationship problems
Yes	128 (94.1)	32 (91.4)	0.334	—
No	8 (5.9)	3 (8.6)
Employment problems
Yes	70 (51.5)	13 (37.1)	2.288	—
No	66 (48.5)	22 (62.9)
Victim and/or witness of family violence
Yes	84 (61.8)	14 (40.0)	5.390*	2.42
No	52 (38.2)	21 (60)
Substance abuse
Yes	75 (55.1)	13 (37.1)	3.612*	2.08
No	61 (44.9)	22 (62.5)
Suicidal or homicidal ideation
Yes	31 (22.8)	29 (82.9)	44.090***	16.38
No	105 (77.2)	6 (17.1)
Psychotic or manic symptoms
Yes	35 (25.7)	11 (31.4)	0.459	—
No	101 (74.3)	24 (68.6)
Personality disorder
Yes	100 (73.5)	23 (65.7)	0.842	—
No	36 (26.5)	12 (34.3)
Past physical assault
Yes	130 (95.6)	34 (97.1)	0.171	—
No	6 (4.4)	1 (2.9)
Sexual assault/jealousy
Yes	67 (49.6)	16 (45.7)	0.171	—
No	68 (50.4)	19 (54.3)
Weapons or credible threats
Yes	68 (50.0)	34 (97.1)	25.703***	35
No	68 (50.0)	1 (2.9)
Escalation of assault
Yes	117 (87.3)	33 (94.3)	1.352	—
No	17 (12.7)	2 (5.7)
Violations of contact orders
Yes	11 (9.2)	1 (2.9)	1.510	—
No	109 (90.8)	34 (97.1)
Minimization/denial
Yes	108 (79.4)	17 (48.6)	13.464***	4.11
No	28 (20.6)	18 (51.4)
Attitudes toward violence
Yes	105 (77.2)	22 (62.9)	2.999	—
No	31 (22.8)	13 (37.1)

Note. OR = odds ratio.

*

p < .05. **p < .01. ***p < .001.

Predictors of Intimate Partner Homicide

A multivariate logistic regression was conducted to identify the predictors of IPH. The logistic regression estimated the odds of IPH occurring versus IPV. The variables that were statistically associated with IPH were included in the analysis. The predictors/variables were divided into two groups: sociodemographic variables (i.e., age and marital status) and SARA items (i.e., violation of conditional release or community supervision, victim and/or witness of family violence, substance abuse/dependence, suicidal and/or homicidal ideation/intent, use of weapons, and/or credible threats of death and minimization or denial of spousal assault history). For this purpose, the sociodemographic variables were entered first, followed by the SARA items, defined for this study as risk indicators. The results are summarized in Table 3.

OPEN IN VIEWER

Table 3.

Logistic Regression Model Presenting the Predictors of Intimate Partner Homicide (IPH).

	Predictor	χ2 (df)	B	SE	Exp(B)	p Value
Step 1	Age	22.704(3)	0.073	0.020	1.076	.000
	Married vs. non married		−0.853	0.519	0.426	.100
	Single vs. non single		−2.722	0.707	0.066	.000
Step 2	Age	69.514(6)	0.064	0.029	1.067	.026
	Married vs. non married		−0.915	0.695	0.401	.188
	Single vs. non single		−3.063	0.967	0.047	.002
	Conditional release violation		−3.248	1.440	0.039	.024
	Victim/witness of family violence		−1.476	0.627	0.229	.019
	Substance abuse		−0.687	0.585	0.503	.241
	Suicidal/homicidal ideation		2.574	0.658	13.118	.000
	Use of weapons or credible threats		2.889	1.137	17.976	.011
	Minimization or denial		−0.745	0.618	0.475	.228
Total		92.218(9)

Note. Dependent variables: IPV (0) vs. IPH (1); Independent variables: Conditional release violation (0 = no; 1 = yes); victim/witness of family violence (0 = no; 1 = yes); substance abuse/dependence (0 = no; 1 = yes); suicidal or homicidal ideation (0 = no; 1 = yes); use of weapons or credible threats (0 = no; 1 = yes); minimization or denial (0 = no; 1 = yes).

*

p < .05. **p < .01. ***p < .001.

The logistic regression model presenting the sociodemographic variables (Step 1) was statistically significant, χ²(3) = 22.704, p < .001. According to the pseudo r-square, between .124 (Cox & Snell) and .195 (Nagelkerke) of the variability was explained by this set of variables. This model classified 83% of all cases. In the logistic regression model, the variables of age and marital status emerged as strong predictors for the perpetration of intimate homicide, with older males presenting a higher probability of perpetrating homicide against an intimate partner and males who were not single presenting a lower likelihood of perpetrating IPH.

When we added the SARA items to this analysis (Step 2), the role of these variables produced a pseudo r-square between .417 (Cox & Snell) and .654 (Nagelkerke), which means that the total variance was a result of both sociodemographic characteristics and SARA items and accurately classified 90.6% of the cases. Based on these results, this contribution was revealed to be statistically relevant, χ²(6) = 69.514, p < .001, as was the final model, χ²(9) = 92.218, p < .001. A separate analysis of the variables used in the prediction of IPH (global model) made it possible to verify that six of them contributed significantly to the model: age (odds ratio [OR] = 1.067; 95 confidence interval [CI] = [1.008, 1.129]), marital status (OR = .047; 95 CI = [.007, .311]), violation of conditional release or community supervision (OR = .039; 95 CI = [.002, .653]), victim and/or witness of family violence (OR = .229; 95 CI = [.067, .782]), suicidal or homicidal ideation/intent (OR = 13.118; 95 CI = [3.610, 47.670]), and use of weapons or credible threats of death (OR = 17.976; 95 CI = [1.935, 166.999]). Therefore, IPH perpetrators are 1 time more likely to be older, 13 times more likely to have suicidal and/or homicidal ideation/intent in the year previous to the crime, and almost 18 times more likely to use weapons or credible threats of death. Inversely, the likelihood of perpetrating IPH decreases by 95% if the man is not single decreases by 96% if the man has a history of violating conditional release or community supervision and decreases by 77% if he has a history of victimization and/or exposure to family violence in childhood.

To evaluate the trade-off between sensitivity and specificity for all possible cutoff probabilities of the model, a maximum likelihood estimate of the ROC was obtained. The ROC curve analysis revealed that the AUC was .95 (p < .000; 95%; CI = [.92, .98]), which was significantly greater than .50, meaning that the generated model significantly predicted IPH.

Discussion

The aim of this study was to explore the differences between IPV and IPH and to identify the factors that best predict IPH based on offenders’ own reports. Because evidence on IPV has generally been drawn from victims and evidence about IPH has been drawn from national databases or police files, this study’s use of offenders’ own reports provides an important contribution to the understanding of these phenomena. Overall, the results challenge the assumption of murder as an inexplicable and sudden event and reveal that these offenses are explicable (Dobash, Dobash, & Cavanagh, 2009). The results of this study reveal that despite the fact that perpetrators of IPV and perpetrators of IPH share some characteristics, other factors and variables make it possible to distinguish between men who perpetrate IPV and those who perpetrate IPH.

IPH perpetrators are older than IPV perpetrators, and a higher proportion of IPH perpetrators are single compared with IPV perpetrators. In fact, these two factors appear to be predictors of IPH. These results are in accordance with the previous literature (e.g., Breiding, Black, & Ryan, 2008; Breitman et al., 2004; Grann & Wedin, 2002; Stith et al., 2004). Recent investigations have revealed that older men have a greater likelihood of inflicting daily violence (e.g., Awang & Hariharan, 2011). Johnson and Hotton (2003) argued that men were more likely to kill an intimate partner if they were not married to the partner (i.e., if they were not in a formal relationship), most likely because these relationships are more prone to dissolution and to infiltration by competitive males.

Although family, individual, and cultural factors have been identified as precursors or correlates of serious intimate partner abuse, IPH 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 a spousal assault history than IPV perpetrators. In fact, the literature has consistently associated such factors with partner abuse (e.g., Abramsky et al., 2011; Caetano et al., 2008; Cummings et al., 2013; Dutton & Kropp, 2000; Grann & Wedin, 2002; Kropp, 2009; Stith et al., 2004). A study conducted by Dobash et al. (2007) revealed that men who committed murder were significantly less likely than abusers to have a history of alcohol abuse and to experience physical abuse as children. Roehl et al. (2005) also noted that substance abuse is more of a risk factor in IPV than in IPH and that child abuse victimization and witnessing domestic violence in childhood are important risk factors for IPV. Nonetheless, in this study, only a history of victimization and/or exposure to family violence and conditional release violation appeared as predictors, and both substantially decreased (by more than 90%) the likelihood of a man committing IPH. These results may suggest that being victimized or witnessing violence in childhood can make individuals more aware of the consequences of lethal forms of IPV and that the implementation of civil or criminal protective orders (e.g., bail, probation, parole, restraining orders, peace bonds) is likely to prevent homicide (Kropp, 2009). At the same time, these results, especially those related to the violation of conditional release or restriction orders, suggest that perpetrators of severe offenses against intimate partners seem to have proper social behavior and may be less antisocial than perpetrators of less severe violence. A recent study found that previous property offenses decreased the likelihood of IPV and that a criminal history and prior violent criminality were not significant factors (Weizmann-Henelius et al., 2012). Dobash et al. (2004) compared male IPHs with other murders and concluded that IPHs resemble those in the general population. Nonetheless, these findings remain controversial; researchers have found that the most severe and persistent domestic offenders are those who commit other antisocial acts (e.g., Cunha & Gonçalves, 2013; Holtzworth-Munroe & Stuart, 1994; Walsh et al., 2010).

Comparisons between the groups also showed that perpetrators of IPH had a stronger history of suicidal/homicidal ideation or intent and a stronger history of the use of weapons and credible threats of death against the victim than IPV perpetrators did. These factors appear in the present study as important predictors of IPH, markedly increasing (more than 10 times) the likelihood of a male committing a murder against an intimate partner. The literature also identifies the use of weapons and threats of death (e.g., Campbell et al., 2003, 2007, 2009; Echeburúa et al., 2009; McFarlane et al., 2002; Snider et al., 2009) and suicidal and/or homicidal ideation/intent (Belfrage & Rying, 2004; Grann & Wedin, 2002) as predictors of lethal violence. Dobash et al. (2007) stated that men who killed their intimate partners were more likely to have used violence against a previous partner and to have used a weapon or instrument. Likewise, Echeburúa et al. (2009) found that the use of threats with dangerous objects presented a high discriminative capacity to predict severe violence. In sum, the results revealed that perpetrators of severe violence had a history of serious injuries and threats against their partners, which has been identified by previous studies as a major risk of lethal violence (Campbell et al., 2009; Cattaneo & Goodman, 2005). This finding supports previous findings about the predictive power of prior domestic violence in later IPV (e.g., Cattaneo & Goodman, 2005; Crowne, Juon, Ensminger, Bair-Merritt, & Duggan, 2012; Stith et al., 2004) and IPH (Campbell, 2004; Campbell et al., 2003, 2009; Dobash et al., 2007; Echeburúa et al., 2009). These results also reinforce the view that IPH is not a sudden or unexpected event or an unexpected phenomenon perpetrated by a relatively conventional man. As Dobash et al. (2009) stated, men who committed a murder against an intimate partner “might more accurately be described as ‘abuser turned murder’” (p. 216). However, the lack of differences between IPV and IPH perpetrators with regard to psychotic or manic symptoms and personality disorders contradicts the view that intimate partner murder is an unexpected phenomena and that murder is committed by mentally disordered individuals.

The existence of homicidal or suicidal threats and the use of weapons and/or credible threats of death during the year preceding the crime can suggest some degree of planning on the part of the perpetrator, contradicting the popular view of IPH as a “spontaneous crime of passion” or a random or unpredictable event (Dawson, 2005; Dobash et al., 2009; Goussinsky & Yassour-Borochowitz, 2012; Juodis, Starzomski, Porter, & Woodworth, 2014). This means that in some of these cases, prior formal risk assessments and interventions might have been underestimated or neglected and that at least some homicides or attempted homicides could be potentially preventable crimes. Individuals who present suicidal and/or homicidal ideation/intent need to be taken seriously (Stanko, 1997). More accurate risk assessment instruments must be developed to better identify cases of greatest risk, and multi-agency high-risk case management must be implemented. Some of these efforts could include judicious adjudications of cases, monitoring of correctional outcomes via regular court reviews or specialized probation/parole programs, continuing safety planning for victims, and risk management plans for perpetrators.

Despite their differences, perpetrators of IPV and IPH also share some similarities. A similar and considerable proportion of men in both groups had a history of assault of family members and assault of strangers, an ongoing history of problems in intimate relationships, a history of physical assault, sexual assault or jealousy against women, ongoing escalation of assault in the year previous to the incident, violations of contact orders, and legitimating attitudes toward violence. These similarities call attention to the importance of men’s cognitions and orientations with regard to intimate relationships (Dobash et al., 2009). As Dobash et al. (2009) stated, the similarities between IPV and IPH perpetrators with regard to their orientation toward women, intimate relationships, and the use of violence against women point to an emphasis on a set of cultural beliefs, individual behavior, and institutional responses. To address these phenomena, it is crucial to develop coordinated responses to IPV to create an environment of non-tolerance to IPV (e.g., Almeida, 2012; Jewkes, 2002).

Mental health services could also be a valuable target for risk management in IPV, and batterers’ treatment could be an important measure to reduce the risk of IPV and IPH (Johnson & Hotton, 2003), especially programs that combine an educational and a therapeutic approach. Programs that focus on the development of problem-solving strategies and coping mechanisms may help to reduce the likelihood of homicide and attempted homicide (Serran & Firestone, 2004). Programs that simultaneously promote perpetrators’ skills and focus on the underlying mechanisms of IPV are effective for preventing recidivism (Novo, Fariña, Seijo, & Arce, 2012). This intervention at the individual level must be combined with intervention at the societal level to change attitudes about the acceptance of aggression in males and to challenge our views about the private nature of family life (Serran & Firestone, 2004). In this sense, it could be important to implement domestic abuse prevention programs in educational settings and to develop programs for young people to reduce both boys’ and girls’ acceptance of domestic violence; to create new laws against domestic violence (e.g., more restrictive laws); and to change the ways courts handle domestic violence cases (Gadd, Fox, & Hale, 2014). In terms of policies, the results from this study call attention to the need to develop educational public campaigns, to promote professional training in the law enforcement system (e.g., police officers, magistrates, prosecutors) and for health care professionals, to develop and implement risk assessment instruments for both IPV and IPH, to develop community governmental and institutional policies to promote equal gender relationships, and to develop more effective legislative and judiciary measures to eliminate violence against women in Portugal.

Despite our findings, this survey has some limitations. First, a retrospective approach was used that relied on the batterers’ discourse. Batterers’ and murderers’ reports can be affected by desirability and a tendency to deny or minimize the history of violence. On the contrary, it is possible that abusers and/or killers may show a tendency to exaggeratedly portray themselves as victims, for example, of early abuse. Second, the present survey only included factors related to the offender and the offenses and omitted variables related to the victim and the victim’s report of the facts. Recently, Kuijpers, van der Knaap, and Winkel (2012) found evidence for the importance of the victim’s characteristics in IPV re-victimization. Other surveys (e.g., Cattaneo, Bell, Goodman, & Dutton, 2007) have concluded that victims’ reports are quite accurate and should be included in the evaluation of risk prediction. Therefore, future studies should take such elements into account. Third, our sample was not representative, which may have limited the generalizability of the results. Fourth, we have used only static linear predictive relationships and did not include interaction effects in the prediction of violence severity. Fifth, other relevant factors that have been found to be important with regard to IPH, such as estrangement, violence during pregnancy, and stepchildren in the home, were not analyzed despite being referenced in the literature (see Campbell et al., 2007). This limitation is, in part, related to the instrument used to identify the risk factors (i.e., SARA), which also uses broader definitions of the evaluation criteria (e.g., relationship problems include separation from the partner or moderate to extreme conflicts in the relationship), thus affecting the degree of precision of risk assessment. Further research should include these variables to determine whether these issues apply to the Portuguese context. Moreover, contextual factors such as patriarchy, sexism, and gender roles should also be included in further analyses because some researchers (e.g., Knickmeyer, Levitt, Horne, & Bayer, 2004) note that IPV is a consequence of patriarchy and is part of a systematic attempt to maintain male dominance. It is also important to note that this study was conducted with a sample of offenders identified by the judiciary system. Given that a substantial number of men have escaped justice with regard to domestic violence, different data could emerge from this additional knowledge. Nevertheless, the current research provides relevant and interesting data on a phenomenon that, in addition to being global, can exist at specific cultural and social levels.