The Influence of Criminal History on the Likelihood of Committing Lethal Versus Nonlethal Violence

Introduction

Serious violence is an alarming and frequent phenomenon with severe consequences in the short term as well as in the long term. Across the globe, it causes serious bodily harm, material damage, and a plethora of health problems. Indirectly, it affects community feelings of safety in public and private space (Crespi & Rigazio-DiGili, 1996; Krug, Dahlberg, Mercy, Zwi, & Lozano, 2002). In the Netherlands, more than 120,000 violent offenses are registered every year (Statistics Netherlands [CBS], 2010). These violent offenses vary from minor assault to more serious violence such as serious assault, threats of violence, unlawful deprivation of liberty, sexual offenses, and homicide. In its most extreme form, serious violence becomes lethal, resulting in a homicide. In the Netherlands, on average 223 persons each year die in a homicide (Ganpat & Liem, 2012).

In the literature, there are several explanations for why serious violence ends lethally or nonlethally. Individual characteristics of offenders and situational factors are considered important determinants in explaining offending behavior. Carrying a weapon and having used alcohol are examples of such situational factors (e.g., Phillips, Matusko, & Tomasovic, 2007). Some emphasize differences in personal characteristics of individuals that influence the likelihood of committing lethal violence (e.g., Gottfredson & Hirshi, 1990). Here, the general assumption is that individuals differ in the tendency to commit crime: some individuals have personal characteristics making them more prone to commit lethal violence compared with individuals without these characteristics. Overall, these personal characteristics comprise biological factors (e.g., gender, testosterone, and intelligence) and/or psychological factors (e.g., personality traits such as antisocial characteristics and impulsivity). In particular, criminal propensity, impulsivity, and self-control are found to be important (e.g., Gottfredson & Hirschi, 1990; Moffit, 1993).

In the literature, a person’s criminal history is generally seen as an important manifestation of one’s personal characteristics, and especially of criminal propensity, impulsivity, and self-control. Moreover, it is well-established that an individual’s criminal history is a crucial determinant of future offending. Numerous empirical studies have shown that criminal history is one of the strongest predictors for future offending behavior (e.g., Blokland, 2005; Bonta, Law, & Hanson, 1998). Among others, the following aspects of one’s criminal history are especially relevant. First, individuals who have committed a high number of prior crimes have a higher likelihood of future general offending (e.g., Wartna, Tollenaar, & Blom, 2005). Second, those with a violent criminal history have an even higher likelihood of future general offending (e.g., Wartna et al., 2005). Third, individuals with a violent criminal history have a higher likelihood of future violent offending (Farrington, 1989). And fourth, individuals with an early onset of their criminal activities have a higher likelihood to be involved in more severe crimes (e.g., Berk, Sherman, Barnes, Kurtz, & Ahlman, 2009; Moffit, 1993). Thus, studying one’s criminal history seems to be crucial due to its probable predictive power with regard to future offending behavior.

The aim of this study is to examine the influence of criminal history on the likelihood of committing lethal violence versus nonlethal violence. By comparing two groups of offenders we intend to add to a better understanding of why certain serious violent offenses may or may not end lethally. In doing so, we focus on the criminal history of offenders who committed serious violence. The following two research questions will be addressed: (a) To what extent does the criminal history of offenders convicted of lethal violence differ from offenders convicted of serious nonlethal violence?; (b) To what extent does the criminal history of offenders convicted of serious violence influence the likelihood that violence ends lethally?

Empirical studies have indicated that criminal history and lethal versus nonlethal outcomes are linked. First, studies have empirically shown that a majority of homicide offenders had a criminal history for all kinds of offenses prior to homicide (e.g., DeLisi & Scherer, 2006; Dobash, Dobash, Cavanagh, Smith, & Medina-Ariza, 2007; Soothill, Francis, Ackerley, & Fligelstone, 2002). Second, studies have shown that among homicide offenders with a criminal history, a significant proportion had a violent criminal history (e.g., Dobash, Dobash, Cavanagh, Smith, et al., 2007; Farrington, Loeber, & Berg, 2012; Loeber, Lacourse, & Homish, 2005; Loeber, Pardini, et al., 2005; Soothill et al., 2002) and sometimes had a prior conviction for serious violence (Dobash, Dobash, Cavanagh, Smith, et al., 2007). Third, certain specific types of serious crime in a person’s criminal history—such as blackmailing, kidnapping, or threats to kill—increase the likelihood to commit homicide (Farrington, et al., 2012; Loeber, Pardini, et al., 2005; Soothill et al., 2002). Fourth, research that compared the criminal records of lethally versus nonlethally violent offenders found similarities as well as differences between the two groups.

In this study we are able to link a person’s criminal history to a lethal versus nonlethal outcome of violent incidents. To do so, we use two data sets. First, we use data on all incidents with a lethal outcome committed in the Netherlands in the period 1993-2009. This data set comprises incidents involving all offenders who have been convicted for manslaughter or murder. In addition, we use a selected population of incidents that ended in a serious nonlethal outcome registered in the period 2005-2009. This data set consists of data on offenders who have been convicted for “attempted manslaughter” or “attempted murder.” In doing so, we used an unconventional comparison group: in the literature, there seems to be a tradition to consider attempted and completed homicide as one (and the same) group, often due to the assumption that the distinction between attempted and completed homicide is based on chance. In fact, according to DiCataldo and Everett (2008,p. 171) some researchers claim that “often the only difference between a completed homicide and an uncompleted one is due to such random occurrences as the lethality of a gunshot or the proximity and quality of a medical care available for the victim.” Based on this assumption one would expect that there is no difference in the criminal history of attempted and completed homicide offenders. Using these unique data, we are able to examine the influence of criminal history on the likelihood that serious violence ends lethally or not. We acknowledge that situational factors may also be important, however, this is not the focus of this study. To our knowledge, this is the first empirical study based on such large-scale and nationwide data to explicitly compare the criminal history of offenders using these two samples of serious violent offenders.

Theoretical Background

The focus of this study is to examine the influence of criminal propensity on the likelihood of committing lethal versus nonlethal violence by using a person’s criminal history as an indicator for criminal propensity. There are several theories focusing on one’s criminal propensity in explaining offending behavior, of which the General Theory of Crime (Gottfredson & Hirschi, 1990) is considered one of the most important. The founders of this theory postulate that self-control theory can explain several types of criminal behavior, including homicide. According to the central notion of the self-control theory, individuals differ in their propensity to commit crime. According to this theory, the propensity to commit crimes can be explained by a common personal characteristic—low self-control—which is mainly caused by inadequate child-rearing practices. Individuals with low self-control possess individual traits making them more prone to commit criminal behavior compared with people with higher self-control. Impulsivity, the inability to consider the consequences of one’s acts, low frustration-tolerance, a physical rather than verbal orientation, and risk-seeking behavior are considered to be important facets of low self-control (Gottfredson & Hischi, 1990, p. 90). Due to these traits, people with low self-control tend to have a “here and now orientation” making them less able to consider the consequences of their acts, resist crime and other acts with short-term gratification. Once self-control is established early in life, it remains relatively stable during the life course making individuals with lower self-control more prone to be involved in risky behavior—including criminal behavior. In this study, we will test the General Theory of Crime in predicting lethal versus nonlethal outcomes by considering criminal history as an indicator for one’s criminal propensity.

Overall, from self-control theory it can be derived that people with a higher criminal propensity are not only (a) more likely to commit crimes, but also to (b) have an early onset of their criminal career, (c) commit a higher number of crimes, (d) commit a wide variety of crimes, and (e) have a lengthy criminal career (Gottfredson & Hirschi, 1990). Numerous empirical studies have tested self-control theory and confirmed the predictive power of low self-control on criminal behavior. Some researchers have argued that those who start offending at an early age could be considered as those who are “most enduring crime prone” (Nagin & Farrington, 1992). Also, in line with self-control theory, some researchers have argued that individuals with low self-control are not only more prone to be involved in risky conflict situations, but also to violent reactions in such conflict situations. In fact, empirical studies have confirmed that low self-control has a strong and direct influence on violent offending behavior (e.g., Baron, Forde, & Kay, 2007; Piquero, MacDonald, Dobrin, Daigle, & Cullen, 2005). Offenders with low self-control also have been found to have more arrests for violent offenses than offenders with higher self-control (DeLisi, 2001). Thus, we might hypothesize that people with low self-control tend to use violence in conflict situations and therefore are not only more likely to have a violent criminal history than people with high self-control, but also have a higher number of violent offenses in their criminal history. Above all, Gottfredson and Hirschi (1990) postulated that low self-control also explains homicide offending: people with lower self-control have a higher likelihood to commit homicide than people with higher self-control. However, when it comes to the severity of crimes, Gottfredson and Hirschi do not provide an elaborate explanation for why some offenders are more likely to commit more severe violent crimes (i.e., lethal versus nonlethal violence) than other offenders.

Previous Studies

The literature shows that certain aspects of one’s criminal history are especially related to (serious) violent offending. Overall, in the literature early onset is considered one of the most important predictors for (a) a long-term criminal career, (b) involvement in more severe crimes, (c) showing a wide variety of crimes and for (d) committing a higher number of crimes compared with offenders who initiate crime at a later age. However, even though numerous studies have shown that individuals with an early onset have a higher likelihood to be involved in more severe crimes, other research has also provided some evidence that homicide offenders often have their first criminal record in adulthood (Soothill et al., 2002). Empirical studies did not only confirm that a significant proportion of homicide offenders had a violent criminal history (e.g., Dobash, Dobash, Cavanagh, Smith, et al., 2007; Farrington et al., 2012; Loeber, Lacourse, et al., 2005; Loeber, Pardini, et al., 2005; Soothill et al., 2002), but also showed that homicide offenders tend to have more arrests for violence than chronic offenders (DeLisi, 2001). Moreover, research has shown that violent behavior at an early age was associated with a higher number of violent offenses and with more serious violent behavior (e.g., Tolan & Thomas, 1995). Other studies have shown that the higher the number of crimes committed, the higher the likelihood to commit violent crimes (e.g., Capaldi & Patterson, 1996).

More specifically, earlier studies that compared the criminal history of those who committed lethal versus nonlethal violence found that most offenders of both groups had a prior conviction record (Dobash, Dobash, Cavanagh, & Medina-Ariza, 2007; Soothill et al., 2002). However, nonlethally violent offenders were more likely to have a prior conviction than lethally violent offenders (Dobash, Dobash, Cavanagh, & Medina-Ariza, 2007). Also, offenders of (attempted) homicide have committed a higher number of crimes than offenders convicted for (attempted) aggravated assault (Smit, Bijleveld, Brouwers, Loeber, & Nieuwbeerta, 2003). When it comes to the types of crimes committed in the criminal history of lethally versus nonlethally violent offenders, the literature shows that a criminal history including burglary or drug offenses was more prevalent among lethally violent offenders (Soothill et al., 2002), whereas public order offenses were less prevalent (Smit et al., 2003). When it comes to a history of violence, the literature shows mixed results: some studies found that a conviction history for violence was more prevalent among lethally violent offenders (Soothill et al., 2002), whereas others found that it was more prevalent among nonlethally violent offenders (Dobash, Dobash, Cavanagh, & Medina-Ariza, 2007; Smit et al., 2003). When the relative risks for committing lethal violence were compared among violent offenders, certain types of crimes in the criminal history—including violent crimes such as wounding (endangering life), robbery, kidnapping, and arson—significantly increased the likelihood of committing lethal violence (Soothill et al., 2002).

Also, comparing the age at the time of the index offense, it appeared that lethally violent offenders were on average older compared with nonlethally violent offenders (Dobash, Dobash, Cavanagh, & Medina-Ariza, 2007; Smit et al., 2003; Soothill et al., 2002). However, the literature provides mixed results when it comes to the age of first offense: some studies found that lethally violent offenders were on average younger when they were convicted for the first time compared with nonlethally violent offenders (Soothill et al., 2002); in contrast, others research found that lethally violent offenders were older compared with nonlethally violent offenders (Smit et al., 2003). In sum, previous studies provided some mixed results when it comes to comparing the criminal history of lethally versus nonlethally violent offender groups.

A possible explanation for differences in these results may be related to the fact that these studies used different methodologies and samples. For instance, Dobash, Dobash, Cavanagh, & Medina-Ariza (2007) focused only on intimate partner violence; thereby they compared a sample of intimate partner murderers with a sample of offenders of nonlethal intimate partner violence. Soothill et al. (2002) focused on murderers of all subtypes—thus without restricting themselves to intimate partner violence—and made comparisons with other violent offenders. Smit et al. (2003) compared offenders of murder and manslaughter with three groups of offenders: (a) attempted homicide offenders, (b) offenders of attempted aggravated assault, and (c) offenders of aggravated assault. Also, Soothill et al. (2002) used a matched case-control design (e.g., matching by age and gender) whereas both Dobash, Dobash, Cavanagh, & Medina-Ariza (2007) and Smit et al. (2003) made comparisons without matching. These aspects might have resulted in different findings.

Based on the above-mentioned theoretical considerations and earlier empirical studies, we have derived the following hypotheses, including conflicting hypotheses on the influence of early onset (Hypotheses 1a and 1b).

Method

Results

To answer our first research question—to what extent the criminal history of lethally versus nonlethally violent offenders differs—we conducted descriptive analyses. Table 1 displays the results regarding background characteristics and criminal history characteristics of both groups of offenders. When comparing background characteristics, the results show that nonlethally violent offenders are more likely to be born in the Netherlands, and were found to be younger at the time of the index offense.

OPEN IN VIEWER

Table 1.

Background Characteristics and Criminal History Characteristics for Lethally versus Nonlethally violent offenders.

	Lethally violent offenders (N = 2,049)	Nonlethally violent offenders (N = 3,387)	p	Statistical test
Male (%)	92	92 a	ns	Pearson Chi-Square
Born in the Netherlands (%)	54 b	65	**	Pearson Chi-Square
Birth country c (%)				—
The Netherlands	54	65
Morocco	5	4
Turkey	8	4
Dutch Antillean	8	7
Surinam	8	5
Other Western countries	8	6
Other non-Western countries	9	9
Unknown	0	0
Mean age at index offense	31.0 (SD = 10.2)	29.9 (SD = 11.5) d	**	Mann-Whitney
Age category at index offense e (%)				—
< 17	5	11
18-25	31	33
26-40	47	36
41-65	17	19
>66	0	0
Unknown	0	1
Existence of prior criminal record (%)	68	76	**	Pearson Chi-Square
Average number of prior criminal records	6.7 (SD = 12.7)	7.1 (SD = 13.1)	**	Mann-Whitney
Mean age onset	23.2 (SD = 9.5) f	21.1 (SD = 9.2) g	**	Mann-Whitney
Age category onset (%)				—
<14	6	9
14-17	29	37
18-25	34	33
26-40	24	15
41-65	6	5
>66	0	0
Unknown	1	1
Average length of criminal history (in years)	7.7 (SD = 8.3)	8.8 (SD = 9.2)	**	Mann-Whitney
Existence prior violent record (%)	38	48	**	Pearson Chi-Square
Average number of prior violent records	0.9 (SD = 1.8)	1.3 (SD = 2.3)	**	Mann-Whitney
Mean age onset violent offenses	27.6 (SD = 10.2) h	25.6 (SD = 10.9) i	**	Mann-Whitney
Existence property offenses (%)	45	46	ns	Pearson Chi-Square
Average number of prior cases of which property offenses	2.8 (SD = 7.8)	2.7 (SD = 7.8)	ns	Mann-Whitney
Existence of drug offenses (%)	10	13	**	Pearson Chi-Square
Average number of prior cases of which drug offenses	0.2 (SD = 0.8)	0.3 (SD = 0.9)	**	Mann-Whitney
Existence other offenses (%)	56	62	**	Pearson Chi-Square
Average number of prior cases of which other offenses	2.8 (SD = 5.0)	2.9 (SD = 5.0)	**	Mann-Whitney
Prior prison sentence (%)	45	49	**	Pearson Chi-Square
Average number of prior prison sentences	1.8 (SD = 3.5)	2.0 (SD = 4.2)	**	Mann-Whitney

Note: ^aMissing =1. ^bMissing = 5. ^cMissing = 5. ^dMissing = 26. ^eMissing = 26. ^fMissing = 28. ^gMissing = 29. ^hMissing = 28. ⁱMissing = 29.

*

p < .05.**p < .01. ns = not significant.

When we compare the criminal history between the two groups, the results show that—even though the majority of both groups had a criminal history—nonlethally violent offenders are more likely to have a criminal history prior to the index offense compared with lethally violent offenders. The results further indicate that nonlethally violent offenders are more likely to have committed a higher number of crimes compared with their counterparts. On average, nonlethally violent offenders are more likely to begin offending at an earlier age than lethally violent offenders. ⁷ In comparing the average length of the criminal history between the two groups, the results indicate that nonlethally violent offenders have a longer criminal history compared with lethally violent offenders. When we compare the types of crimes in the criminal history between the two groups, the results show that nonlethally violent offenders are not only more likely to have a violent history, but are also more likely to have committed more violent crimes and to begin committing violence at an earlier age compared with their counterparts. ⁸ Also, nonlethally violent offenders are not only more likely to have committed at least one drug offense and at least one other offense but are also more likely to have committed a higher number of these offenses compared with lethally violent offenders. Finally, nonlethally violent offenders are not only more likely to have a prior conviction involving a prison sentence, but on average they are also more likely to have a higher number of prior prison sentences compared with lethally violent offenders. Thus, overall, nonlethally violent offenders tend to have a more severe criminal history compared with lethally violent offenders in many respects.

Next, to answer our second research question—to what extent are criminal history variables influential in predicting lethal versus nonlethal outcomes—we used binary logistic regression analyses. Table 2 reflects which characteristics of criminal history are influential in predicting lethal versus nonlethal outcomes. ⁹ We present two models that contain the same variables, but with one exception: Model I shows the effect of the dichotomous variable “existence of a prior criminal record” and “existence of a prior violent record,” whereas Model II shows the effect of the continuous variables “Total number of prior criminal records” and “Total number of prior violent records.” Model I is included to test hypotheses 2a and 3a whereas Model II is included to test the other hypotheses. In all models we also included our control variables. First, we examined whether our control variables (i.e., gender, age at index offense, and birth country) influence the likelihood of lethal versus nonlethal outcomes. Model II shows that age at index offense and country of birth have a significant negative effect on lethal versus nonlethal outcomes: the younger a person’s age at committing the index offense the higher the likelihood to commit lethal violence. Also, those who are born outside the Netherlands had a significantly higher likelihood of committing lethal violence compared with those born in the Netherlands. These findings should be interpreted with care, as this may be related to homicide within the criminal milieu (e.g., drug related or contract killings), as the majority of offenders of this subtype are born outside the Netherlands.

OPEN IN VIEWER

Table 2.

Logistic Regression Model for criminal history: Lethal (1) vs. Nonlethal (0).

	Model I		Model II
	Exp(b)	Exp(S.E.)	Exp(b)	Exp(S.E.)
Age at index offense	0.999	1,004	0.990*	1,004
Male	0.996	1,112	0.966	1,112
Born in the Netherlands	0.662**	1,062	0.650**	1,062
Age of onset	1.017**	1,005	1.028**	1,005
Existence of prior criminal record	0.889	1,094	—	—
Total number of prior criminal records	—	—	1.023**	1,003
Existence of prior violent record	0.687**	1,078	—	—
Total number of prior violent records	—	—	0.835**	1,022
Prior prison sentence	1.262**	1,084	1.132	1,077
Constant	0.637**	1,169	0.585**	1,161
Nagelkerke R square	0.037		0.049
N	5,371		5,371

*

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

In testing our first hypothesis (conflicting hypotheses 1a and 1b) that the earlier the age of onset of offending, the higher/lower the likelihood of committing lethal violence compared with nonlethal violence, Table 2 shows—in contrast to hypothesis 1a, but in line with hypothesis 1b—that age of onset of offending had a significant positive effect on the likelihood of committing lethal violence compared with nonlethal violence (Model I and II). In other words, the later the age of onset, the higher the likelihood of a lethal outcome. For every one year increase in age of onset, the odds of a lethal versus nonlethal outcome increased by a factor of 1.028 (Model II).

Next, we tested hypotheses 2a and 2b: that a criminal history increases the likelihood of committing lethal violence and the higher the number of crimes in a person’s criminal history the higher the likelihood of committing lethal violence compared with nonlethal violence. The results indicate—in contrast to hypothesis 2a—that having a prior criminal record did not have a significant influence (Model I),—but in line with hypothesis 2b,—the total number of crimes in one’s criminal history had a positive significant influence on lethal versus nonlethal outcomes (Model II). For each one-unit increase in the total number of prior criminal records, the odds of a lethal versus nonlethal outcome increased by a factor of 1.023. Then, we tested our third hypothesis (hypothesis 3a) that a history of violence increases the likelihood of committing lethal violence compared with nonlethal violence. Model I indicates that—in contrast to hypothesis 3a—having a history of violence had a negative significant influence on lethal versus nonlethal outcomes. We also tested hypothesis 3b that the more violent crimes in one’s criminal history, the higher the likelihood of committing lethal violence compared with nonlethal violence. In contrast to hypothesis 3b, Model II shows a significant negative effect of the number of violent crimes in one’s criminal history on the likelihood of committing lethal violence compared with nonlethal violence. Finally, we tested the influence of having at least one prior conviction for a prison sentence on the outcome of violent conflicts. Although the results in Model I suggest that those who had at least one prior conviction involving a prison sentence had a significantly higher likelihood of committing lethal violence, its effect disappeared when controlling for the total number of prior crimes (Model II). Thus, those individuals who had committed a high number of prior crimes, started offending at a later age, and those who did not have any prior violent record or had committed a low number of prior violent crimes had a significantly higher likelihood of committing lethal violence.

Conclusion and Discussion

This study examined to what extent criminal propensity can explain some individuals committing a homicide in a given situation, while other individuals commit violent acts that do not end lethally. More specifically, the aim of this study was to compare the criminal histories of lethally versus nonlethally violent offenders and to examine the influence of criminal history on the likelihood of committing lethal violence, using the offender’s criminal history as an indicator for criminal propensity. Using a unique data set of criminal record data for two groups of offenders, our results showed that—in line with previous research—most lethally as well as nonlethally violent offenders had a criminal history.

In line with the results of Dobash, Dobash, Cavanagh, & Medina-Ariza (2007), nonlethally violent offenders were significantly more likely to have a criminal history compared with lethally violent offenders. Also, on average, we found that nonlethally violent offenders were significantly more likely to have committed more prior crimes compared with lethally violent offenders. In addition, similar to the results of Smit et al. (2003), we found that on average nonlethally violent offenders were significantly more likely to begin offending at an earlier age than lethally violent offenders. Furthermore, in comparing the average length of the criminal history between the two groups, nonlethally violent offenders had a significantly longer criminal history compared with lethally violent offenders. In line with previous studies, we found that a significant proportion of lethally violent offenders had a violent criminal history (Dobash, Dobash, Cavanagh, Smith, et al., 2007; Farrington et al., 2012; Loeber, Lacourse, et al., 2005; Loeber, Pardini, et al., 2005; Soothill et al., 2002), and that nonlethally violent offenders were more likely to have a violent history compared with lethally violent offenders (Dobash, Dobash, Cavanagh, & Medina-Ariza, 2007). In addition, in this study we also found that nonlethally violent offenders were more likely to have a higher number of prior violent records and to have a prior prison sentence compared with lethally violent offenders. Although previous research has shown that initiating violent behavior at an early age is associated with committing a higher number of violent offenses and also with committing more serious violent behavior as well (e.g., Tolan & Thomas, 1995), our results showed that, on average, most serious violent offenders in our study initiated violent behavior in adulthood. In sum, our results demonstrate that nonlethally violent offenders had a more severe criminal history compared with lethally violent offenders. Thus, when we use one’s criminal history as an indicator for one’s criminal propensity, our results indicated that—contrary to what we expected based on the central notions of the self-control theory—nonlethally violent offenders have a higher criminal propensity than lethally violent offenders. These findings are not in line with the expectations derived from the self-control theory. Therefore, we answer our first research question by concluding that one’s level of criminal propensity—measured by one’s criminal history—does not seem to be sufficient in explaining why certain individuals commit a homicide whereas others do not. We suggest that future studies should not solely restrict themselves to examining the role of criminal propensity, but should also consider the role of other factors in explaining lethal versus nonlethal outcomes, especially the role of certain specific situational factors.

In answer to our second research question, we found that a person’s criminal history can be influential in predicting lethal versus nonlethal outcomes, although the influence of certain aspects of one’s criminal history were not always in line with our expectations. More specifically, in contrast to what we expected, a prior criminal record did not significantly predict lethal versus nonlethal outcomes. However, in line with what we expected, we found that the total number of previous crimes had a significant positive effect on the likelihood of committing lethal violence, suggesting that the more crimes one has committed prior to the index offense, the higher the likelihood for a lethal outcome to the violent act. In contrast to hypothesis 1a—but in line with hypothesis 1b—we found that those who started offending at an earlier age did not have a higher likelihood of committing more severe crimes (i.e., lethal violence). In fact, we found that age of onset had a significant positive effect on the likelihood of committing lethal violence, suggesting the later one has started with the first crime, the higher the likelihood of a lethal outcome. These results contradict findings from numerous previous studies showing that individuals with an early onset have a higher likelihood of being involved in more severe crimes compared with those who initiate crime at a later age (e.g., Moffit, 1993). In fact, in our study, on average, most serious violent offenders (both lethally as well as nonlethally violent offenders) started committing their first registered crime in adulthood.

This result questions the general applicability of the assumption that those who start offending at an early age could be considered the “most enduring crime prone” (Nagin & Farrington, 1992). For serious violence, including homicide, we do not find support. In this respect, our findings are more in line with the work of Soothill et al. (2002), who also found that homicide offenders often have their first criminal record in adulthood. It may be the case that for other types of crimes the hypothesis still holds. One possible explanation for the late onset of serious violent offenders might be related to difficulties that emerge with the transition to adulthood. For instance, Dobash, Dobash, Cavanagh, Smith, et al. (2007) found that lethally violent offenders who started offending at a later age tend to experience a relatively unproblematic childhood, but—in contrast—tend to experience many problems in adulthood, such as substance abuse and relationship problems. Tentatively, this may apply to lethally as well as nonlethally violent offenders.

Next, in contrast to what we expected, having a history of violence had a negative effect on the likelihood of committing lethal violence. The results also indicated that the number of violent records matters: in contrast to hypothesis 3b we found that committing a higher number of violent crimes has a negative effect on the likelihood of committing lethal violence, indicating that the higher the number of prior violent crimes, the lower the likelihood to commit lethal violence. This result is not in line with our expectations derived from the self-control theory. Alternatively, this finding may be explained as follows: those individuals who commit a higher number of violent crimes learn and gain certain experience and skills in committing crimes, which can make them more able to restrain themselves and to prevent a violent conflict from ending lethally. After all, criminal behavior can be learned. As lethally violent offenders tend to have a lower number of violent records they may lack the skills and experience that other individuals do have to resolve a conflict in a nonlethal manner. Also, when interpreting this result, one has to take into consideration that—while strictly speaking in the Netherlands an intent to kill is required to convict a person for attempted homicide—evidence indicates that over the past two decades, the Dutch Public Prosecutor has decided increasingly to prosecute cases for attempted homicide rather than aggravated assault (see, for example, Smit & Nieuwbeerta, 2007). Possibly, the above-mentioned argument may gain in strength when this development is also considered. However, it must be stressed that the study at hand does not allow for firm conclusions here. Future research could test the validity of this explanation. Also, it has to be noted that our sample only consists of serious violent offenders, so we should not conclude that committing a high number of violent crimes decreases the likelihood of committing lethal violence for all offenders. This stresses the need for future research to examine in more detail the influence of criminal history by also including offenders who have committed less serious violent crimes.

This study aimed to contribute to the existing knowledge on violent crime. It is the first empirical study based on large-scale and nationwide data to explicitly compare the criminal history of offenders using these two samples of serious violent offenders. In contrast to much of the existing research in the field, this study also combined descriptive analyses and explanatory analyses when studying the criminal history of serious violent offenders.

Despite the insights delivered, some limitations should also be addressed. A first limitation is that in this study—like is done in many other studies—we used criminal history as an indicator for criminal propensity. Ideally, anticipating on one of the major criticisms of the self-control theory—namely that it is to some extent tautological (e.g., Akers, 1991)—future research should measure criminal propensity more directly, for instance by using low self-control scales. This, however, requires access to the population. A second limitation concerns the data we used. Criminal record data only provide information about individuals who were arrested or convicted for an offense from the age of 12 years onwards. Therefore, undetected offending behavior and behavior that was never registered by the police is not included, producing an underestimation of criminal history of some individuals. Also, as a result of our large sample size, relatively small differences between groups can be found significant. Therefore, our results should be interpreted with care. A third limitation concerns the group of serious violent offenders—namely offenders of attempted versus completed homicide—included in this study. As this may be a specific group of serious violent offenders our findings cannot be extended to offenders of less severe forms of violence. Therefore, we suggest that further research also includes a comparison group consisting of offenders who were convicted of different types of less severe violent crimes. The same can be said about different types of homicide, which have been grouped together in the present study. A fourth limitation is that in this study we only considered criminal propensity—more specifically criminal history—in explaining lethal versus nonlethal outcomes. Other factors not included in our model can potentially play a more contributory role in explaining lethal versus nonlethal outcomes. After all, like many other types of crimes, serious violent offenses do not occur in a vacuum. Although Gottfredson and Hirschi (1990) acknowledge that opportunities may also matter, the theory has been criticized for not systematically addressing the role of opportunities in explaining criminal behavior (e.g., Geis, 2000). Our results also imply that the self-control theory may be improved by addressing more systematically the role of opportunities. To conclude, although most previous studies have neglected the study of the criminal history of serious violent offenders, this has been the central focus of our study.

By doing so, we have provided several new insights to the existing literature. First, this study questions the general assumption that the distinction between an attempted and completed homicide is solely based on chance. In fact, this study suggests that nonlethally violent offenders (i.e., attempted homicide offenders) have a higher criminal propensity than lethally violent offenders, which was not in line with our expectations derived from the self-control theory. Second, when the influence of criminal history is studied among serious violent offenders, this study suggests that certain aspects of criminal history do not always have the effect among serious violent offenders as what is generally assumed in criminology, especially when it comes to age of onset and frequency of a violent history.

In sum, our study emphasizes the need for future research to examine the influence of criminal propensity on lethal versus nonlethal outcomes of violent incidents more systematically and in more detail, which may be especially relevant for treatment policy for repeat violent offenders. It also calls for additional research that combines both criminal propensity of the offender and situational factors, to provide a more complete picture of the complex realities of serious violent events and a more in-depth understanding of why certain conflict situations end lethally and others do not. This may also require the study of processes of interaction between all parties present, including victims and third parties. In an ideal situation these types of insights may even support preventative measures with respect to the most harmful and frightful types of crimes: lethal violence.