Homicide Sanctioning in The Netherlands

HOMICIDE SANCTIONING IN

THE NETHERLANDS

A quantitative study on the flow of homicide through the

system and the judicial sentencing decision

Valerie Dekker

S1334182

Master Thesis

Crisis and Security Management

January 2019

Supervisor: Dr. M. C. A. Liem

Second reader: Dr. J. Vüllers

Leiden University

Faculty of Governance and Global Affairs

The Netherlands

2

3

Table of contents

Table of figures ... 5

Table of tables ... 5

1. Introduction ... 6

1.3 Prior homicide research in the Netherlands ... 9

1.4 Method ... 9

1.5 Relevance ... 10

1.6 Research objective and reading guide ... 11

2. Literature Review ... 13

2.1 Homicide and the Dutch justice system ... 13

2.2 Sanctions in the Netherlands ... 14

2.3 The Netherlands in international perspective ... 16

2.4 Sentencing theories ... 17

2.5 Theoretical background of characteristics ... 17

2.5.1 Characteristics of the perpetrator ... 18

2.5.2 Characteristics of the victim ... 19

2.5.3 Characteristics of the offense ... 21

2.6 Empirical background of characteristics ... 22

2.6.1 Characteristics of the perpetrator ... 22

2.6.2 Characteristics of the victim ... 23

2.6.3 Characteristics of the offense ... 24

2.7 Hypotheses ... 25

3. Methodology ... 27

3.1 Data ... 27

3.2 Strategy ... 28

3.2.1 Strategy in the explorative part ... 28

3.2.2 Strategy in the explanatory part ... 31

3.2.2.1 The dependent variable ... 32

3.2.2.2 The independent variables ... 33

3.3 Validity and reliability ... 37

4

4.1 Explorative results ... 39

4.1.1 The flow of homicide through the system ... 39

4.1.1.1 Sentences ... 39

4.1.1.2 Sanctions and imprisonment ... 42

4.1.1.3 The final flow ... 44

4.1.1.4 Homicide categories, imprisonment continued ... 45

4.1.2 Cross tabulations ... 47

4.2 Explanatory results ... 49

4.2.1 Independent variables of the regression analyses ... 49

4.2.1.1 Prison, perpetrator and victim characteristics ... 49

4.2.1.2 Prison, offense characteristics ... 50

4.2.2 Regression analysis for prison sanctions ... 51

4.2.3 Regression analysis for prison sanctions including outliers ... 52

4.2.4 Regression analysis for the sanctions prison and prison in combination with psychiatric care ... 53

5. Conclusion ... 55

Bibliography ... 58

Appendix A ... 65

Appendix B ... 70

Appendix C ... 71

5

Table of figures

Figure 1: Rough sketch on the flow of homicide through the Dutch system ... 6

Figure 2: The flow of homicide through the system; prosecution and sentencing... 42

Figure 3: The flow of homicide through the system; prosecution, sentencing and imprisonment ... 45

Figure 4: The final flow of homicide through the Dutch judicial system ... 55

Table of tables

Table 2: Sentenced for homicide ... 40

Table 3: Sentenced for other crime(s) ... 41

Table 4: Was the perpetrator sanctioned? ... 42

Table 5: Legal type of crime ... 45

Table 6: Cross tabulation Sentenced & Sanctioned ... 47

Table 7: Cross tabulation Legal type of crime & Sanctioned ... 48

Table 8: Perpetrator and victim characteristics ... 49

Table 9: Regression prison sanction and characteristics ... 51

Table 10: Regression prison sanction, including outliers, and characteristics ... 52

Table 11: Regression prison sanction, including prison combined with psychiatric care, and characteristics ... 53

6

1. Introduction

This thesis researches homicide in the Netherlands. In particular, the criminal justice response to homicide. Homicide is, arguably, the worst of crimes. Taking one’s life is a very serious form of interpersonal violence. In 2017, 158 individuals became victims of murder or manslaughter in the Netherlands (CBS, 2018). This asks for a fitting criminal justice response.

1.1 Background

Homicide investigations typically start with a dead body. If it cannot be ruled out that the individual died of non-natural causes, a body is classified as a suspicious death. See figure 1.

Figure 1: Rough sketch on the flow of homicide through the Dutch system

When this is the case, the criminal investigation team of the Dutch National Police carry out an investigation (Politie.nl, 2018). Some cases take a long time to solve and sometimes cases do not get solved at all. These cases stay ‘uncleared’ (Liem et al., 2018). When cases do get clearance and there is enough information present, the police can proceed to arrest suspects. After a suspect is arrested, the now called defendant enters the criminal justice system. In the Netherlands, the Public Prosecution Office is responsible for the detection of a perpetrator. When a case can be put together, the public prosecutor will prosecute the defendant. The

Suspicious death

7

defendant receives notice that he or she is brought to trial (Tak, 2008). In the courtroom, the judge is the one who decides whether the defendant is guilty or not. If he or she is, the judge also sentences the defendant, now called the perpetrator. A common sentence for homicide perpetrators is the sanction of imprisonment (Rechtspraak.nl, 2018a).

1.2 Problem outline

Research is needed into the flow of homicide through the system. As shown in figure 1, not all sentenced perpetrators in the Netherlands end up in prison. In fact, if zoomed out on the whole system, an interesting phenomenon is seen. From beginning until end, there is an outflow of homicide cases and/or defendants at every step. In the criminal justice process, not all prosecutions lead to sentencing and not all sentencings lead to imprisonment (Dutch Homicide Monitor, 2019). There is no accurate, recent information available on the exact number of cases and/or people at every step as depicted in figure 1. Moreover, it is not known exactly what has happened to all defendants that have been prosecuted but not sentenced and what has happened to all perpetrators that have been sentenced but not imprisoned. This thesis aims to shed a light on this.

The main focus of this thesis is the final step in the process described above. The step from sentencing to imprisonment. According to Wingerden & Nieuwbeerta (2010), it is often unknown what exactly happens at that last step when cases are at trial. Different homicide defendants receive different lengths of prison sentences. Some receive other sanctions and some do not receive a sanction at all (Dutch Homicide Monitor, 2019). This is interesting because the Dutch criminal law states consequences of committing homicide. Life imprisonment or a 30-year maximum sentence for murder (Article 289, WvSr) and a 15-30-year maximum sentence for manslaughter (Article 287, WvSr). Tak (1995) outlines that in the Dutch criminal justice system, the judge has a wide discretionary power. A judge decides what exact sentence is given. There are no minimum sentences or further rules for this. This causes variation in sentences. On the one hand the discretionary power can lead to personalized and appropriate sentences. On the other hand, it can result in unwanted inconsistencies and inequality in sentencing (Tak, 1995). It is believed that there are multiple factors that can have an influence on, or possibly explain, the sentencing decision by a judge. Or as stated by Black (1976); the quantity of law received by the perpetrator.

8

At trial, judges will aim to gain as much background information on the perpetrator as possible to be able to make an accurate decision. According to Steffensmeier, Ulmer & Kramer (1998), this is not always possible. The focal concerns theory claims that judges will fall back on stereotyping if some of the needed background information is missing. As a result, judges will incorporate certain characteristics of the perpetrator into their decision (Steffensmeier et al., 1998). Spohn & Holleran (2000), claim that the blameworthiness of a perpetrator contributes to the sentencing decision of a judge. The more blame appointed, the more dangerous the perpetrator is, the higher a sentence. According to Daly & Bordt (1995) for example, the gender of the perpetrator affects the height of a sentence given. Female perpetrators are seen as less dangerous and less blameworthy and are therefore more likely to be sentenced leniently.

Another way to appoint more or less blame to a perpetrator is to research the victim. Baumer, Messner & Felson (2000) for example, theorized that if the victim is a young child, the perpetrator can expect a harsher sentence because of the innocence of the victim. Finally, offense characteristics are believed to have an influence on the judges’ decision. These are factors that are associated with the offense or homicide itself. An example is the modus operandi, the method of killing. Glaeser & Sacerdote (2003), empirically found that death by burning or fire, led to the longest prison sentence. An explanation for this is that the social cost of arson is high.

Most research into the judicial sentencing decision and the perpetrator, victim and offense characteristics, has been carried out in the United States (Johnson, Wingerden & Nieuwbeerta, 2010). Consequently, most theoretical ideas that are statistically tested in this thesis, are derived from American research. However, it is important to note that this thesis is based on homicide offenders the Netherlands and that the Dutch and American legal systems are quite different. The Netherlands makes use of the inquisitorial system (Komter & Malsch, 2012). As explained by Kim (2013), the legal system in the United States is adversarial. This means that in the Netherlands, the judge makes a decision based on information acquired by him or herself. On the contrary, American judges make a decision based on information presented to them by the involved parties. The Dutch justice system has more unique factors. An example is the ability of judges to impose a life sentence in which life means life. More differences between systems in the Netherlands and the United States will be explained in chapter 2.

9

1.3 Prior homicide research in the Netherlands

Over the past years, empirical research into the sentencing decision in homicide cases in the Netherlands has carried out. Nieuwbeerta & Wingerden (2006) for example, researched the differences in the requested sentences by the public prosecutor and the sentence given by a judge. Over the years 1993-2004, 65% of the Dutch homicide perpetrators received a lower sentence than originally asked for by the public prosecutor. The database Moord en Doodslag (murder and manslaughter) from the Dutch study centre for crime and law enforcement was used. It has also been researched if the person of the judge influences the decision or if different areas where the courts are situated lead to different sentencing decisions (Berghuis, 1992; Schuyt, 2011). Assens, Dekkers, Keppenne & Pepels (2010) claim that unrelated factors can

influence a judge and therefore influence the decision. An example is the presence of an attorney, this significantly reduces the chance of a harsh sentence.

This thesis focusses on the relation between the length of a prison sentence and the perpetrator, victim and offense characteristics. Research into these characteristics and their influence on the sentencing decision has been carried out often as separate research, so for example the influence of perpetrator characteristics on the sentences (Wingerden & Wermink, 2015), or even more narrow, the influence of ethnicity of a perpetrator on the final sentence (Wermink, Wingerden, Wilsem & Nieuwbeerta, 2015). Only the research of Johnson et al. (2010) and Wingerden & Nieuwbeerta (2010), combine all characteristics and research the influence of these characteristics on sentences given, similar to this research. However, both use a different and older database compared to this thesis and do research over a different time period; 1993-2004. Homicide cases that happened in more recent years are not included, neither are the more recent findings in literature.

1.4 Method

In aiming to find an answer to the main research question, quantitative analysis is used. Statistical tests using data from the Dutch Homicide Monitor will be carried out in statistical program SPSS. The Dutch Homicide Monitor is a dynamic monitor containing information on Dutch homicides since 1992. Leiden University took the initiative to set up this database because there was no existing database where homicide data could be registered, found and

10

stored centrally, making empirical research hard. All data on the flow of homicide through the system, perpetrator, victim, offense characteristics and sentencing durations are derived from this database. Analysis of this data can help us gain valuable insights compared to research that uses other, older, databases. The database starts with homicide cases that took place in 1992. Unfortunately, the first years do not have Public Prosecution Office linked case numbers that make it possible to retrace case specific information necessary for this thesis. Without this information our research cannot be carried out which is why the first years are deleted from the database beforehand. Interest is given to the years 1997-2016. But in these years too, sentencing information and many characteristics are unknown. Desk research is conducted to find as much information as possible before carrying out statistical tests.

1.5 Relevance

Even though homicide is an important subject to study, it is an under researched topic. Prior to the existence of the Dutch Homicide Monitor, it was hard to gather information. Data on Dutch homicide cases was not kept track of centrally. Prior research, such as that of Johnson et al. (2010) and Wingerden & Nieuwbeerta (2010), uses older databases and a different, older, time period. An updated research is therefore in place. Characteristics of the victim and offense in particular, have not been empirically researched often. Testing theories in Dutch context can provide valuable information. Homicide is one of the most serious crimes that can be committed. It is the most lethal type of violence. These criminal acts do not only have a big impact on families and friends of the victim and perpetrator, but also on society as a whole (Liem et al., 2018). In addition, social and financial costs that come into existence after these crimes are high. Accordingly, acts of homicide ask for a fitting criminal justice response.

If compared with the sanctioning of minor offenses, homicide sanctioning is harsher. A better understanding on the way homicide is punished will be valuable to all parties involved in the criminal process. Society often evaluates and/or judges the system, by looking at sanctioning. Insight in this process will provide a better understanding. It will be tested if certain characteristics have an impact on the length of prison sentences. For some of the characteristics, there is an understandable theoretical thought behind them whereas for other factors it might seem unfair to have them playing a role in the sentencing decision. The origin of a perpetrator for example, is expected to play a role in the judicial decision, even though it should technically

11

not matter where the perpetrator comes from. Acknowledging inequality and addressing that problem is important to provide transparency and equality in the Dutch justice system, which is why these theories need to be tested. More insight on the decision-making process of sentencing can help society in general. This knowledge is an asset for all people who are involved in that decision-making process. Judges, lawyers and foremostly the suspects themselves. Especially sanctions for committing homicide can bring huge changes to someone’s life. Legal uncertainty is unwanted. Groot-van Leeuwen, Laemers & Sportel (2015) concluded their research with the claim that more insight in why and how a verdict is established leads to more acceptance of that verdict amongst all parties involved.

The Dutch Homicide Monitor is part of a European initiative to create a European Homicide Database. A joint database on lethal violence. Due to a coding and operationalization manual that countries use for storing data, comparison has been made possible opening many possibilities for future research as well.

1.6 Research objective and reading guide

As stated, the main aim of this is research is to find an explanation for the difference in sentence lengths that perpetrators of homicide receive by Dutch judges. It will be tested to what extent characteristics of the victim, perpetrator and offense, influence the sentencing decision and can explain the variation is prison sentence lengths. Therefore, this research is mostly explanatory. It brings us to the following main research question:

‘To what extent can perpetrator, victim and offense characteristics explain the difference in

prison sentence durations for homicide perpetrators in the Netherlands over the years 1997-2016?’

For a better understanding of sentencing and sanctioning in the Netherlands, a closer look will be taken into the two last steps in figure 1. From prosecution to sentencing and from sentencing to sanctioning (or imprisonment). There is an outflow of people at those two steps, not all perpetrators that are prosecuted are sentenced, and not all perpetrators that are sentenced are imprisoned. These two gaps of people will be explored. The aim is to find out what happened to all prosecuted perpetrators and to find accurate numbers how many people are sentenced and

12

sanctioned. Therefore, this research is partly explorative as well. In the following chapter a theoretical framework will be given. After that, the methodology is discussed. An analysis of the data is next, after which the conclusion will aim to answer the main research question.

13

2. Literature Review

In the theoretical framework, an introduction into homicide and the justice system in the Netherlands will be given. After that the different types of sanctions will be introduced. Thirdly, the characteristics that are expected to influence a judges’ sentencing decision will be explained. For the characteristics, a theoretical and empirical background will be given. The characteristics are divided into perpetrator, victim and offense characteristics.

2.1 Homicide and the Dutch justice system

Homicide consists of multiple categories. All are described in the Dutch criminal law. Homicide can be categorised into murder and manslaughter. Both legal types of crime need the intent of the perpetrator to take away someone’s most precious possession, one’s life. For a murder conviction, as opposed to a manslaughter conviction, premeditation by the perpetrator needs to be proven. This means the perpetrator has to have had a comprehensive plan to murder the victim. In addition, the perpetrator has to have taken measures to prepare the crime before it happened (Leistra & Nieuwbeerta, 2003). Both categories come with different sentence maximums. For murder, life imprisonment or a thirty-year maximum sentence can be given (Article 289, WvSr). For manslaughter, this limit is set at fifteen years (Article 287, WvSr). It is expected that the legal type of crime influences the sentence length. Murder is seen as a more serious crime than manslaughter. According to Johnson et al. (2010), premeditation increases the perception of dangerousness of the perpetrator. The maximum sentence that can be given is therefore higher. Apart from the main article for murder, infanticide can be classified as murder too (Article 291, WvSr). Apart from the main article of manslaughter, qualified manslaughter and infanticide classified as manslaughter (Article 290, WvSr) exist as well (Leistra & Nieuwbeerta, 2003). Qualified manslaughter is an aggravating circumstance. It means that another crime happened during or before the homicide making the crime easier or concealing it. Punishment maximums of qualified manslaughter are equal to murder (Article 288, WvSr). On the contrary, infanticide is a mitigating circumstance. For murder there is a nine-year maximum prison sentence and for manslaughter the maximum is six years. Children in infanticide cases are less than 1 year old (Liem & Haarhuis, 2015).

14

According to Johnson et al. (2010), the Netherlands has a system with nineteen courts that govern punishment. The Public Prosecution Office has a vital role in criminal proceedings. In homicide cases, the public prosecutor is directly involved and leads the investigation that is carried out by the police. The public prosecutor also decides whether or not to prosecute the defendant. If it is not in the public interest to prosecute, the public prosecutor can decide to drop a case (Blom & Smit, 2006). After the investigation, the public prosecutor puts together a report. In that report the public prosecutor makes a recommendation on charges. The recommendations are not legally binding (Johnson et al., 2010). The judge is the one who brings out a verdict in which the defendant is told his or her sanction. In the case of serious offenses such as homicide, the court makes use of the full bench. This means there is a panel of three judges deciding together (Blom & Smit, 2006).

To reach a verdict, the judge consults the law, the facts, circumstances and explanations by experts and witnesses. In addition, Dutch judges work with jurisprudence, verdicts in other cases. This is also known as case law. These verdicts can be used as a guideline, but judges are also allowed to deviate from them (Rechtspraak.nl, 2018). The time limit for judges to come to their conclusion is two weeks. Both the defence and the prosecution can start an appeal after the sentence is given (Johnson et al., 2010). According to Tak (2008), judges in the Netherlands are independent. The minister does not have any authority over a judge and the judges are appointed for life. The Public Prosecution Office is responsible for the execution of sanctions.

2.2 Sanctions in the Netherlands

The Dutch law, or more specifically the Dutch criminal law, prescribes what acts are forbidden, and if committed, what sanction could be imposed. Criminalised acts are divided into two categories. Less serious felonies and more serious offenses (Leij & Jongste, 2010). Due to the seriousness of committing homicide, this act is classified as an offense. All offenses can be found in chapter two of the criminal law. Homicide can be found in the category crimes against life. The sanctions of homicide perpetrators are categorised as principal penalties. They need to be proportional to the offense (Tonry & Bijleveld, 2007).

The principal penalties found in the Dutch criminal law from most to least severe are imprisonment, detention, task penalties and fines. The law prescribes maximums and not

15

minimums (Tak, 2008). In addition, entrustment orders can be given. This happens when the perpetrator is not fully responsible due to mental health issues. If treatment is not successful, perpetrators could spend their whole life hospitalized in confinement. This can be combined with a prison sentence (Tonry & Bijleveld, 2007). These sanctions are applicable to young offenders as well. The Netherlands has special procedures and authorities responsible for young perpetrators (Beijerse, 2017). Perpetrators under the age of eighteen are trialled in separate juvenile courts and sentenced to separate juvenile sanctions (Tak, 2008). Extrajudicial settlements are more common for young perpetrators. According to Beijerse (2017), The so-called bureau HALT, in cooperation with the Public Prosecution Office, aim to rehabilitate, recover and re-educate young perpetrators to reduce the risk of unwanted behaviour happening again.

When defendants are prosecuted for homicide, they are not always found guilty. According to OM.nl (2018), when a judge does not believe it is lawfully proven that the defendant committed the crime, a judge can exonerate the defendant. A perpetrator can also be dismissed of all legal procedures. This happens when the crime was committed in self-defence or when the defendant cannot be held accountable due to mental health issues and has received a full insanity judgement (Leij & Jongste, 2010). In the Netherlands, the defendant has certain rights. The Public Prosecution Office needs to handle all cases lawfully and respect the rights of the defendant. Whenever these rights are not respected, homicide cases can be dismissed as inadmissible. An example when this happens is if the defendant has to wait too long before the trial takes place or when the trial is not fair for the defendant due to mistakes the police, the Public Prosecution Office or the court made (Leij & Jongste, 2010). Another special measure is a mistrial. This takes place for multiple reasons. An example is when it is already noticeable that the defendant will not receive a sentence (Rechtspraak.nl, 2018). When a defendant suffers from such mental illness or underdevelopment and is not able to understand the prosecution in any way, prosecution can be suspended (Article 16, WvSr). The judge can also decide to not impose a penalty. This can be decided when the crime committed is not serious or when the offense of personality of the perpetrator ask for a measure like this (Article 9, WvSr). This can also be applicable when a perpetrator is already serving a sentence, the case is too old or due to mental health issues.

16

2.3 The Netherlands in international perspective

A large amount of published research on homicide sentencing has been carried out in the United States. According to Wingerden, Wilsem & Johnson (2014), theories developed in the United States are also useful in the Dutch context. Although certain American theories are appropriate to be transferred to Dutch cases, it is important to note that the Dutch legal system is very different compared to that of the United States. The Netherlands makes use of the inquisitorial system. This means that in the Netherlands, the judge makes a decision based on information acquired by him or herself (Komter & Malsch, 2012). The state is responsible for the investigation into criminal offenses, the judge has an active role in finding the truth (Brants, Field & Jörg, 1995). On the contrary, the legal system in the United States is adversarial. American judges make a sanction decision based on information presented to them by the involved parties (Kim, 2013). According to Hirsch Ballin (2012), the adversarial system makes use of a jury. This jury, ‘the people’, choose what version of the presented story is the truth.

As stated by Tak (2008), the Netherlands does not make use of a jury system. In contrast to the United States, there is no plea bargaining and there are no guilty plea procedures such as receiving a more lenient sentence when admitting guilt. There are no lay judges or mandatory minimum sanctions for serious offenses. This leaves Dutch judges with a broad discretionary power. The Netherlands does not make use of the death penalty. Within the prison sanction, life sentences are in use. In the Netherlands, life means life. A life sentence does not have a limit, it ends when the perpetrator deceases (Tak, 2008). The latter has been subject of discussion over the past years. The Netherlands has been condemned by the European Court for the protection of Human Rights. According to the court, the implementation of the life sentence is not in accordance with the European Convention for the protection of Human Rights. The life sentence is not humane (Hattum, 2016). After these judgement on the Netherlands, it became apparent that changes had to be made to the life sentence. In 2016 it was proposed to make requests for a pardon available for perpetrators after serving 27 years. Prior to this new rule, this could only be requested in case of terminal illness. The pardon can only be assigned by the King (Hattum, 2018).

17

2.4 Sentencing theories

According to Haak (2013), one of the most basic and traditional thoughts why we punish perpetrators is that perpetrators need to pay retribution for committed crimes. Another reason why we sentence perpetrators is deterrence. It is needed to try to prevent crimes, particularly homicides, from happening (Haak, 2013). According to Rausch (1983), the deterrence theory assumes that individuals who experienced a punishment will no longer participate in deviant behaviour. If the threat of punishment weighs more than the reward of committing a murder, they will not act. This can work for people who never committed homicide and for people who did commit one. The deterrence effect differs for each person. A sanction for a first-time homicide perpetrator may have more impact than the fifth sanction for a recidivist (Rausch, 1983). Punishing homicide perpetrators also aims to protect others by keeping the perpetrators of the streets. Rehabilitation and recovery are aspects in the sentences received by perpetrators. It is attempted to treat perpetrators and educate them towards more wanted behaviour (Haak, 2013).

Steffensmeier et al. (1998), talk about the focal concerns theory of sentencing. This theory says that judges have three main concerns when deciding on the sentence. The first one being the culpability; to what extent the perpetrator can be blamed for the offense. Secondly the dangerousness of the perpetrator, to what extent society needs to be protected from him or her. Lastly the practical effects. Wermink et al. (2016) summarises these in blameworthiness, community protection and individual or organizational constraints. The theory underlines that judges do not always have all the information about these focal concerns, they therefore fall back on stereotyping. They try to get as much information as possible from the files and trials but when aspects are unknown, the age, gender, origin, social class or other characteristics of a perpetrator will play a role in addition to seriousness of the crime and criminal past (Steffensmeier et al., 1998).

2.5 Theoretical background of characteristics

The characteristics that are expected to have an influence on the judicial sanctioning decision, are divided into perpetrator, victim and offense characteristics. A theoretical background will be provided. Ideas derived from American research will be used even though the topic of this

18

thesis are Dutch homicide perpetrators. It is important to note the differences in the legal systems as described earlier in this chapter.

2.5.1 Characteristics of the perpetrator

According to Glaeser & Sacerdote (2003), the sentence of a perpetrator will be higher if it is difficult for a judge to understand or relate to the perpetrator. Spohn & Holleran (2000) found that the blameworthiness of a perpetrator can play a role on imprisonment decisions. The more a perpetrator can be blamed, the more dangerous he or she is, the higher a sentence. The judge is the one who must estimate the dangerousness of a perpetrator. To estimate this, certain characteristics are expected to have an influence. Age, gender and origin are the characteristics that will be described in this chapter.

In prison sentencing, the oldest perpetrators are punished less severe. In stereotyping, they are seen as physically less threatening to society because of their age (Steffensmeier et al. 1995). According to Wingerden & Nieuwbeerta (2010), besides the oldest perpetrators, the youngest perpetrators are also likely to be treated more leniently than middle aged perpetrators. The group between the ages of 21 and 30 is punished most severe. Perpetrators above the age of 30 and under 50 are punished less severe and finally, perpetrators above 50 years old are punished least severe. Steffensmeier et al. (1995) also describe this curvilinear relationship between age and sentencing. They explain that perpetrators under the age of 20 are punished less severe because this group is less responsible due to their age. This group also deserves the best chance to improve themselves. From that age, sentencing harshness increases until they reach the age of 50.

Over the past years it has been theorised that female defendants can expect lower prison sentences. Franklin & Fern (2008) explain this. They state that chivalry and male paternalism are deeply rooted in society. These views contribute to a certain stereotype of strong men and females being the weaker sex. According to Moulds (1978), the term chivalry dates from the middle ages, it was believed that men are required to protect women. Women are more vulnerable. Wingerden & Nieuwbeerta (2010) explain this phenomenon by the patriarchal theory, a traditional thought where men are seen as the dominant sex. Female perpetrators are automatically seen as less blameworthy because of that (Franklin & Fern, 2008). In addition,

19

women are seen as less guilty due to the fact that their expected risk on recidivism is less. The threat they impose on society in general is less big than the threat that male perpetrators impose on society in physical aspect (Wingerden & Nieuwbeerta, 2010). Farrell & Swigert (1986) conclude that male perpetrators are seen as more blameworthy and therefore receive higher sentences.

Racial minorities seem to be overrepresented in crime, victims as well as perpetrators (Sampson, Morenoff, & Raudenbush, 2005). Much research on this topic has been done in the United States. This means many theoretical thoughts focus on the difference between races; whites and blacks. This is difference is the base of race stereotyping in the United States. In early research by Johnson (1941) it is said that blacks are more likely to be sentenced harshly than whites. He states this does not happen because of the seriousness of the crime but because of the threat that the ‘different’, black communities impose on the dominant group, the whites. White interests are protected in the racial caste system (Johnson, 1941).

In the Netherlands, there is no stereotyping based on blacks and whites. The Netherlands does have immigrants. These are often not from Western countries. According to Wermink et al. (2015), from major immigrant groups, immigrants from Surinam are seen most positive, after that Turks. The least popular immigrant groups are Antilleans and Moroccans due to their cultural and socioeconomic differences. In 2009, only fifty three percent of the prison population in the Netherlands consisted of Dutch people (Linckens & Looff, 2010). Immigrants in the Netherlands often experience difficulty in finding jobs, have lower education levels and poorer Dutch language skills (Ours & Veenman, 2003). Therefore, this thesis focusses on the division between Western and non-Western perpetrators. The perception of immigrants is more negative. Judges might see non-Westerners as more dangerous than Western defendants. In the Netherlands, Western people are the dominant group. Many immigrants are different from that main group, potentially posing a threat as Johnson described. According to Wingerden & Nieuwbeerta (2010), it can be expected that perpetrators born outside the Netherlands will be precepted as more dangerous and receive higher sentences.

2.5.2 Characteristics of the victim

Another way to appoint more or less blame to a perpetrator is to research his or her victim. The same characteristics as before will be discussed; age, gender and origin. Stevens (2000)

20

describes, that in the United States a phenomenon is seen that is called ‘good character’ evidence. Survivors of the crime or the state can try to prove the victim had a good character. If this is the case, it can influence the sentencing decision concerning the perpetrator.

Baumer et al. (2000) argue that killing young victims is seen as a worse homicide than killing middle aged victims. More years of life are taken away from young victims. Also, children are seen as innocent. If the perpetrator’s victim was innocent and defenceless, this leads to a higher culpability of the perpetrator (Wingerden & Nieuwbeerta, 2010). When perpetrators cause unrest, disbalance and damage on society through their choice of victims, sanctions will be harsher (Glaeser & Sacerdote, 2000). This will be applicable if young children fall victim of homicide. According to Johnson et al. (2010) people who get to the age of 18 turn this effect around. In stereotyping, young adults are connected to violence. They are not seen as innocent anymore. In addition, very old people are weaker victims as well (Johnson et al., 2010). As described previously, women are seen as more vulnerable and are often stereotyped as more innocent. The same applies if the woman is a victim (Curry, Lee & Rodriguez, 2004). For their perpetrators, the same effect is expected as for perpetrators with very young and very old victims. Baumer et al. (2000), state that violence against women is seen as a worse crime than violence against men.

In the United States, homicides committed against minorities such as blacks or Latinos are less likely to lead to a death penalty for the perpetrator (Baumer et al., 2000). This claim is derived from a theory by Hawkins (1987), that stated that the lives of whites are more valued by society than that of blacks. If whites become victim of homicide, more blame can be accounted to the perpetrator because the life of the victim is valued more. This is deeply rooted in society and can be seen as an unofficial racial caste system. When translated to Dutch context, it would mean that hypothetically, perpetrators with a non-Western victim, would be more likely to be sentenced lenient than when the perpetrator commits homicide involving a Western victim. This thesis will test if the perception of immigrants is more negative than that of Dutch people; if the life of non-Western people or immigrants is valued less, and the victim belongs to that category, this can influence the sentencing decision. It can be argued that the homicide would be seen as less severe, the perpetrator as less blameworthy and the prison sentence could therefore be less severe.

21

2.5.3 Characteristics of the offense

Characteristics that are associated with the homicide itself; offense characteristics, are also believed to be predictors of sentence length. The characteristics that will be discussed are the locus operandi, modus operandi, the type of homicide, the relation between the perpetrator and victim and the gender combination between the perpetrator and victim. As Auerhahn (2007a) mentioned, the offense characteristics and their influences on the judicial response, have been researched very sparsely.

The locus operandi withholds the place of the crime. The location is expected to be a contributing factor to the height of a sentence. If the crime takes place in a public place this is a worse disruption of public order compared to when a murder is committed at home. Society needs to be protected. It is therefore expected that homicides in a public place lead to higher sentences. (Wingerden & Nieuwbeerta, 2010) Johnson et al. (2010) expected that increased community risks lead to more blame and higher sentences for the perpetrator. The community risk is higher if the homicide is committed in a public place.

The modus operandi tells us how the homicide is committed. It can be expected that if more violence is used the punishment will be harsher. However, it is hard to decide what could be seen as a more violent crime. In the United States it is believed that death by burning is the most violent act (Wingerden & Nieuwbeerta, 2010). Higher social costs that are caused by a particular type of modus operandi, can lead to higher sentences (Glaeser & Sacerdote, 2000). In Dutch research, it has been argued that these social costs and increased community risks particularly apply to the use of lethal weapons; firearms (Johnson et al., 2010).

Wingerden & Nieuwbeerta (2010) state that the killing of a child by a parent is a homicide without instrumental motive. They therefore expect it to lead to less severe sentences. Killing within the family is seen as less blameworthy. Johnson et al. (2010) state that because of increased community risk, crimes outside the familial milieu should increase punishment. Perpetrators who target family members are punished less severe. Black (1976) states that the degree of intimacy between the victim and perpetrator can influence the amount of law received. The greater the distance between them, the more law received. Killing a stranger will lead to harsher sentences than killing an acquaintance. Johnson et al. (2010) too argue that

22

killing strangers is associated with increased community risks and that these homicides are therefore responded with harsher sentences.

The gender combination between the perpetrator and victim as an offense characteristic is associated to the gender characteristics in perpetrators and victims. Farrell & Swigert (1978) theorized that males killing females received most severe convictions. After that, females killing females, males killing males and the least severe convictions were given to females killing males. This is a continuation on the theories described above. Males are seen as more violent and women as more innocent. Glaeser & Sacerdote (2000) add, that perpetrators who kill male victims receive a lenient sentence because on contrast to females, male victims are more likely to have started the conflict.

2.6 Empirical background of characteristics

Most empirical research, again, has been carried out in the United States. This thesis tests if the same conclusions are reached in Dutch context.

2.6.1 Characteristics of the perpetrator

Research done by Steffensmeier et al. (1998), has found that there is a significant relation between age and the height of prison sentences given in Pennsylvania. Steffensmeier, Kramer & Ulmer (1995) concluded that older perpetrators, above fifty, are more likely to receive lenient sentences. In research by Champion (1987), it was statistically that there is a curvilinear relationship between age and severity of sentencing; young and old perpetrators receive more lenient sentences than the group in between. This research was carried out in the United States over the years 1970-1984. It included all types of offenses. An example of empirical research into the influence of gender on prison sentences, is the work of Daly & Bordt (1995). They analysed existing American statistical literature and concluded differences between gender and sentences are evident in data sets in the United States. Steffensmeier, Kramer & Streifel (1993) come to that same conclusion, women are less likely to be incarcerated in Pennsylvania over the years 1985–1987. Myers (1979) and Spohn (1995) on the other hand, could not find any significant effects between gender and sentence lengths.

23

Bernstein, Kelly & Doyle (1977) found that in the Unites States, black perpetrators receive less severe sentences. Dixon (1995) found mixed results. On the contrary, multiple researchers claim to have found that black perpetrators receive longer sentences than whites. An example is research of Steffensmeier et al. (1998), or that of Spohn (1990). Spohn & Holleran (2000) also saw an effect between ethnicity and the response of justice agencies, black and Hispanic perpetrators received more response of agencies than whites. More arrests were made for example. This is not the same as higher sentences, but it can be argued that more severe prison sentence lengths would be a logic continuation of that trend. Spohn & Holleran’s (2002) research was carried out in Chicago, Kansas City and Miami. It appeared to be the case that young, Afro-American, male perpetrators are seen as more blameworthy and face longer imprisonment charges compared to their opposites. In the translated Dutch context, Engbersen, Leun & Boom (2007), described that negative stereotyping is common among police officers in the Netherlands and that non-Western immigrants are more often stopped by the police than Dutch, Western people. Non-Western people receive more law, it can be expected that this happens in the sanctioning stage as well.

2.6.2 Characteristics of the victim

Franklin & Fern (2008) conclude that if a female is targeted, the prison sentence is generally higher. Their research was carried out in the Unites States. Curry, et al. (2004) reach the same conclusion in their empirical research on violent crimes in Texas and the effect on sanctions if victims are male or female. Curry et al. (2004) researched violent offenses, there was no focus on homicide. Baumer et al. (2000) and Farrell & Swigert (1986), found that perpetrators whose victims were women generally experienced longer sentences. Perpetrators who murder females are more likely to be convicted (Baumer et al., 2000; Beaulieu & Messner, 1999). Glaeser & Sacerdote (2000) found that in the United states, murderers who kill blacks in their sample, receive 26,8% shorter sentences. Baldus, Woodworth & Pulaski (1986) too state that perpetrators in American cases with white victims are more likely to be imposed to the death penalty than perpetrators in cases with other ethnicity victims.

24

2.6.3 Characteristics of the offense

Johnson et al. (2010) expected that increased community risks lead to more blame and higher sentences for the perpetrator. The community risk is higher if the homicide is committed in a public place. Their statistical analysis in the Netherlands showed that homicides in private places indeed led to less severe sentences. (Johnson et al., 2010) In the United States it is believed that death by burning is the most violent act. (Wingerden & Nieuwbeerta, 2010) This claim is supported by Glaeser & Sacerdote (2000), who found that the harshest penalties (fifty years) are given to perpetrators who commit arson as a circumstance of the murder. Johnson et al. (2010) expected that increased community risks lead to more blame and higher sentences for the perpetrator. This happens through the use of lethal weapons such as fire arms. Their statistical analysis in the Netherlands showed that firearms indeed led to higher sentences compared to other M.O.’s (Johnson et al., 2010).

Johnson et al. (2010) empirically found that robbery homicide and sexual homicide perpetrators received a significantly higher sentence and that partner, parent and child homicides, homicides in the familial milieu, received more lenient sentences. According to Glaeser & Sacerdote (2003), murders in the criminal milieu receive harsh sentenced and murders that are a result of argument receive lenient sentences. Johnson et al. (2010) conducted research in the Netherlands and found that homicides involving acquaintances instead of strangers, received more lenient sentences. Dawson (2006) researched victim-offender relationships in Canada. Her findings show that perpetrators who kill relatives or intimate partners receive lower prison sentences than when they kill strangers, friends or acquaintances. Glaeser & Sacerdote (2003) believe that murders that result from romantic arguments have a high arrest rate and result in a lower sentence. Auerhahn (2007b) too found that intimate partner homicides receive less severe sanctions.

Farrell & Swigert (1978) found that males killing females received most severe convictions. Franklin & Fearn (2008) also examined the role of gender in victim-perpetrator pairs. Their findings are in line with those of Farrell & Swigert (1978) and show that male perpetrators who target females receive the longest sentence. Glaeser & Sacerdote, who carried out research in 33 large urban counties in the United States, found that in a cross-tabulation analysis of means, male perpetrators with female victims turned out to be sentenced most harsh compared to any other gender pair. Curry et al. (2004), concluded the same.

25

2.7 Hypotheses

Derived from literature, the following hypotheses come into existence. The hypotheses will be tested, if outcomes are significant in the statistical tests, the 0 hypotheses will be rejected.

H0 – The age of the perpetrator does not affect the prison sentence length

H1 – The age of the perpetrator does affect the prison sentence length

H0 – The gender of the perpetrator does not affect the prison sentence length

H1 – The gender of the perpetrator does affect the prison sentence length

H0 – The origin of the perpetrator does not affect the prison sentence length

H1 – The origin of the perpetrator does affect the prison sentence length

H0 – The age of the victim does not affect the prison sentence length

H1 – The age of the victim does affect the prison sentence length

H0 – The gender of the victim does not affect the prison sentence length

H1 – The gender of the victim does affect the prison sentence length

H0 – The origin of the victim does not affect the prison sentence length

H1 – The origin of the victim does affect the prison sentence length

H0 – The locus operandi does not affect the prison sentence length

H1 – The locus operandi does affect the prison sentence length

H0 – The modus operandi does not affect the prison sentence length

H1 – The modus operandi does affect the prison sentence length

H0 – The type of homicide does not affect the prison sentence length

26

H0 – The relationship between the perpetrator and victim does not affect the prison sentence

length

H1 – The relationship between the perpetrator and victim does affect the prison sentence length

H0 – The gender combination of the perpetrator and victim does not affect the prison sentence

length

H1 – The gender combination of the perpetrator and victim does affect the prison sentence

27

3. Methodology

In this chapter, the methodology of this thesis will be described. The database, inclusion criteria strategy and operationalisation are subjects that will be addressed. For this research, a quantitative method is used. The aim of this thesis is twofold. The flow of homicide through the system is explored. This will provide valuable background information on the Dutch judicial system, from prosecution to the final stage of a sanction. The main aim however, is to test statistically if perpetrator, victim and offense characteristics can explain the variance in the length of prison sentences, given to homicide perpetrators by a judge. This thesis is therefore explorative as well as explanatory. To study the Dutch judicial response to homicide, the Dutch Homicide Monitor is used (Universiteit Leiden, 2018a).

3.1 Data

The Dutch Homicide Monitor is an active database set up by Leiden University to gain more insight in homicide in the Netherlands. It contains information on all Dutch homicide cases since 1992. The database was set up because previously, there was no place where data on Dutch homicide cases could be recorded and kept track of centrally. This made research very hard. The database focusses on homicide. This means that prosecutorial charges of abortion, euthanasia, involuntary manslaughter, assault, battery or other violent crimes that may or may not result in death, are not registered in the Dutch Homicide Monitor. Important to note, is that it is not expected that all perpetrators prosecuted for homicide, will be convicted of homicide as well. The information in the database comes from various sources. Data from the Public Prosecution Office, data from the National Police and newspapers are used to select cases that can be included in the dataset. This leads to accurate information but to enrich the data set even more, additional open sources are used. This provides detailed, case-specific and individual-level information such as the characteristics of perpetrators, victims and the offense needed for this thesis. The open sources that are used in the database are case files, information from the Legal Services Department, criminal records of the Research and Documentation Centre and finally, statistics from the Dutch Bureau of Statistics (Universiteit Leiden, 2018b).

The focus of this research is on homicide. Though homicide is an important subject to study, the lack of accessible information in the Netherlands, before the Dutch Homicide Monitor was

28

set up, caused it to be an under researched topic. Homicide is one of the most lethal types of violence. Societal impact of these crimes is big. Hence, punishment is harsher compared to minor offenses. A better understanding on the way homicide is punished, will be valuable to all parties involved in the criminal process. Society often evaluates and/or judges the system, by looking at sanctioning. Insight in this process will provide a better understanding. In addition, some characteristics may appear to play a role even though they should not, origin for example. Acknowledging inequality and addressing that problem is important to provide transparency and equality in the Dutch justice system

This thesis studies the Netherlands. This choice has been made because empirical research into the sentencing decision using all described characteristics and using the Dutch Homicide Monitor, has not been carried out yet. In addition, there is access to the database making this research feasible. The database contains information over the years 1992-2016. By way of contrast, this thesis has a focus on the years 1997-2016. The first years are taken out of the database. This choice has been made due to missing case numbers that can make the connection between perpetrators and the Public Prosecution Office files. If this number is missing, it is not possible to find detailed information such as the characteristics of perpetrators and victims. This information is necessary to carry out this research; this information is exactly what makes this database unique and different compared to other older databases used in other research projects.

3.2 Strategy

In working towards reaching the two aims of this thesis, a quantitative strategy is chosen. Statistics program SPSS is used to carry out tests with the dataset. Theories from literature will be tested statistically for Dutch homicide defendants, making this research deductive.

3.2.1 Strategy in the explorative part

In the explorative part, it is researched how homicide cases move through the Dutch system, described in figure 1. As stated before, not all arrests in homicide cases lead to prosecution, not all prosecutions lead to sentencing and not all sentencings lead to imprisonment. At this point, there are no accurate figures that can describe the dimension of these groups in the Netherlands. This thesis aims to find these numbers. It is known that there is an outflow of cases at every

29

step described above, it is therefore studied what happens to those outflowing perpetrators. As the topic of this thesis is the judicial response to homicide, the study focuses on accounting for perpetrators that are prosecuted but not sentenced and for perpetrators that are sentenced but not imprisoned. For information on the steps before prosecution, it is recommended to read other theses within this research project.

To be able to provide accurate numbers, desk research is necessary. The Dutch Homicide Monitor already contains two main variables necessary for this part of the analysis; variables Sentenced and Sanctioned. It has been chosen to renew variables in the same scheme instead of restructuring new ones, because these variables are in line with the coding manual of the European Homicide Monitor. This way, international comparison can be made possible. The reason desk research is necessary, is that these variables contain many missing values. To find missing information, the case number from the Public Prosecution Office data, name of the individual and date of the crime are used to retrace missing information online. As a starting point of these searches www.google.nl is used. Most information was retraced through verdicts found on www.rechtspraak.nl, www.jure.nl and www.navigator.nl. In some cases, online newspaper articles were used. Especially from local newspapers such as www.tubantia.nl. In addition, other existing variables from the database are used.

The variable Sentenced in the Dutch Homicide Monitor answers to the question whether or not the perpetrator has been sentenced. Possible categories within this variable are that the perpetrator was found not guilty, that the perpetrator was not held accountable due to mental illness, that the perpetrator was not sentenced for other reasons, that he or she has been sentenced for homicide or that he or she has been sentenced for another crime or multiple crimes. (Universiteit Leiden, 2018c). First, an entirely new variable is made; SentencedVal_Comment. This variable contains information extracted from different variables. The variable OMparket is researched. Here, it can be found if perpetrators committed suicide or if cases are unsolved. In CONV_OM the Public Prosecution Office data of the sentencing can be found. The variable OM_VONNIS_OMSCHRIJVING is used to find cases that are fully dismissed of all legal proceedings due to an insanity judgement. They will be added to the category perpetrators that are not held accountable due to mental illness. To supplement the variable SentencedVal, this firstly made variable SentencedVal_Comment is used and in addition, some information is copied from the variable Sentenced itself (Universiteit Leiden, 2018a).

30

Important to note is that all information in the variables is derived from the judges’ decision made in the first instance. Appeal decisions are not taken into account. If homicide and other crimes are committed at the same time, the perpetrator is categorised as ‘The perpetrator is sentenced for homicide’. In the category other crimes, the most serious crime is recorded even though some perpetrators are sentenced of multiple other crimes. In the Dutch Homicide Monitor, 999 is used if information within a variable is unknown (Universiteit Leiden, 2018a).

In the coding manual of the European Homicide Monitor, the variable Sentenced categorises the perpetrators in a broad way. All perpetrators sentenced for homicide are put into one category. However, homicide consists of multiple components (Universiteit Leiden, 2018c). To gain more accurate results, a new variable is added to the database that specifies what type of homicide sentenced perpetrators received. Examples are murder or manslaughter. Both crimes need the intent of the perpetrator to kill someone. For a murder sentence, an extra component is needed. The crime needs to be premeditated (Article 289, WvSr). The variable CategorieVal is made. In SentencedVal_Comment it is seen what the perpetrators are sentenced off, all murder related crimes are categorised as murder. All manslaughter related crimes are categorised as manslaughter and all other crimes are categorised as other. In the variable FINAL_uitspraak_VD the legal type of crime can be seen for the years 2006-2016. Marieke Schemmelkens also carried out research using the Dutch Homicide Monitor. She had already started this new variable for the years 2006-2016 (Universiteit Leiden, 2018a). Complicity in homicide and accessory to homicide are categorised in the variable Sentenced as a perpetrator sentenced for homicide. In the variable CategorieVal however, only complicity in homicide is marked as murder or manslaughter. Accessory and attempt are classified as other. Attempted homicide is classified as a perpetrator sentenced for other crimes.

Furthermore, it is aimed to find the type of sanction received by perpetrators that are sentenced for homicide or sentenced for another crime or multiple crimes. The variable Sanctioned categorises all sentenced perpetrators into different categories describing the sanction they received. This variable is supplemented as well. It is possible that perpetrators did not receive a sanction, that perpetrators were sanctioned to prison, that he or she received acute psychiatric care, long term psychiatric care, prison and psychiatric care, youth prison, youth institutional treatment or youth prison and youth institutional treatment. Lastly, the perpetrator can be sanctioned to something else, this category is called other (Universiteit Leiden, 2018c). For the variable Sanctioned_new, OM_comments, Straf and Sanctioned are used. With this desk

31

research and newly added information, it is aimed to find exact numbers on Dutch prosecutions and sentencings, and to explain the gap in the last two steps of figure 1, the gap of people that are prosecuted but not sentenced and the gap of people that are sentenced but not imprisoned. Descriptive results for the variables mentioned above are given such as frequency tables, descriptive statistics and crosstabs. For schematic overview on the coding and categorising of these variables, appendix A is added.

3.2.2 Strategy in the explanatory part

The explanatory part of the result section starts with descriptive results. Frequency tables, descriptive statistics and crosstabs are used to provide an overview of the variables used in the analyses. Individuals in the Dutch Homicide Monitor are divided into victims and perpetrators. The dependent variable prison sentence length in days is connected to solely perpetrators because they are the individuals receiving the sentence. However, in the regression analyses it will also be tested if the victim characteristics can be predictors of the prison sentence received by the perpetrator. This means information needs to be transferred so that the prison sentence length will be in the same row as the characteristics linked to the victims. To do this, the excel formula vertical lookup is used. The database case numbers were used to connect victims to perpetrators. In the case of multiple victims, the characteristics of the principal victims are used. In the regression analyses, it is tested if the null hypotheses mentioned in this chapter, can be rejected. Apart from the R2 and p values in the statistical output tables, all figures are rounded off to two decimals. This has been done to ensure a better presentability. In reporting the statistic outcomes, APA guidelines have been followed. This means commas used in numbers are replaced with decimal points in that particular section.

In the explanatory part of this thesis, multiple regression analysis is used to test if it is possible to predict the dependent variable; sentence length in days, by the independent variables; the characteristics. There are multiple independent variables which is why multiple regression analysis is used. Because the independent variable is measured on a ratio level, a scale variable, and the independent variables are categorical or scale, the choice has been made to use linear regression analysis. For the regression testing, a confidence interval of 95% has been chosen. This means the p value in the regression needs to be smaller than .05 in order to be able to reject the null hypothesis.

32

Before carrying out the actual regression analyses, the assumptions that need to be met are tested. In the SPSS chart builder, it is seen that the assumption of linearity is met for the variables. The assumption of homoscedasticity can be seen in plots. The assumption is not met but the fact that cases with larger disturbances have more weight than other observations is accepted. The assumption of normality is seen in the histogram in plots. This assumption is met. There are no repeated measurements, the independence of the error terms is secured. By performing a Durbin-Watson Test it is shown that de residues are uncorrelated, the values are between 1 and 3. The prevention of multicollinearity is secured by not entering all dummy variables within one variable.

3.2.2.1 The dependent variable

The dependent variable is, as stated, the length of a prison sentence given to a perpetrator. This is an open, numeric variable measured in the amount of days the perpetrator has to go to prison (Universiteit Leiden, 2018b). This variable is used as the dependent variable, the y, in the statistical regression analyses. The variance in this variable is what needs to be explained. The variable VRIJH_STR is the dependent variable in this thesis. It measures the prison sentence length in days. It is completed with help from FINALjaar_VD, a variable made by Marieke Schemmelkens, LENGTHSENTENCE and online verdicts (Universiteit Leiden, 2018a).

The dependent variable contained outliers. Namely, the life prison sentence in which life means life. The length of a life sentence differs for each perpetrator, not all perpetrators decease at the same age. To be able to add life sentences in the regression analyses, they have been operationalised as 50 years. This amount of years has been chosen to indicate the fact that the sentence is significantly longer than the maximum limited sentence of 30 years. When a perpetrator is around 20 or 30 years old when they are sanctioned a life prison sentence, they could reach this age. VRIJH_STR_REGRESSIE is a duplicate of the variable VRIJH_STR, but the life sentences are coded as 50 years. Length of the prison sentence is also known for the youth sanctions and the sanction prison combined with psychiatric care. The limit for youth prison, two years, is significantly lower than for the adult prison sentences. This is why they are not included in the regression analyses. The sanction prison combined with psychiatric care on the other hand, is more similar. This sanction can therefore be included in the regression. To ensure accurate results, it has been chosen to carry out three separate models. There are differences between the sanctions in the three models. In the first model, the dependent variable

33

only contains the prison sentence lengths of those who were sentenced to a ‘regular’ prison sentence. Secondly, an analysis is carried out where the outliers, life sentences, are included. In the third model, the prison sentence lengths are used of perpetrators that received a prison sentence or a prison sentence in combination with psychiatric care. By using three models, results will be more accurate and the differences between models can be studied.

3.2.2.2 The independent variables

The regression analyses test if the independent variables, the x’s or predictors, can predict the dependent variable. The independent variables are divided into perpetrator, victim and offense characteristics. For the perpetrator and offense characteristics these are age, gender and origin. For the offense characteristics these are modus operandi, locus operandi, type of homicide, relation between the victim and perpetrator and the gender combination between the victim and perpetrator. These variables have many categories. In regression analyses, only categorical variables with two categories can be included. The variables needed to be recoded into dichotomous or dummy variables. Dichotomous variables consist of two categories. Dummy variables too, in which the categories usually indicate whether or not something is applicable.

By adding too many predictors in a regression model, it is hard to draw conclusions. There is a risk factors will hardly say anything anymore. This is why the variables are recoded, and only 5 recoded variables are used in the regression analyses. For the same reason, the characteristics are run in blocks within the different models. The perpetrator, victim and offense characteristics are run separately. With the method ‘enter’, the selected independent variables are all adopted into the model at the same time. This has been done because there is a small set of predictors. It is unknown which of them will be significant, which is why they are not selected beforehand, but all entered at the same time. The predictors, or independent variables, are concepts derived from literature. The independent variables used in the regression models are expected to influence the dependent variable. The independent variables, the x’s, are divided into categories. Characteristics of the perpetrator, characteristics of the victim and characteristics of the offense.

Characteristics of the perpetrator and characteristics of the victim are divided into age, gender and origin. In the Dutch Homicide Monitor, the variable age is an open numeric variable, the age is measured in years. This variable remains a scale variable in the regression. The variable