Arsonists and Violent Offenders Compared: Two Peas in a Pod?

Participants

The research group of this retrospective cross-sectional study was composed of De Waag clients. As forensic outpatient treatment center in the Netherlands, De Waag offers several outpatient therapies to juveniles (12-18 years) and adults with criminal tendencies. Cognitive-behavioral treatment is offered in a group or individually, as well as pharmaceutical, relation, and systemic therapy. Therapy is mainly focused on learning to identify and avoid risk situations, and building a prosocial way of life. Acute psychosis and severe addiction are general exclusion criteria, which need to be addressed prior to admittance to De Waag. For this study, participants accepted for treatment at De Waag between 1992 and 2003 were selected from the client registration system. Inclusion criterion was the reason for admittance for treatment, with the type of offense being convicted arson or (domestic) violence. Due to a limited number of available cases, almost all of the obtainable arsonists were included (provided that file information was sufficient), except for women and participants with a violent offense in their criminal history. Because of the larger scope of the violent offender group, the included participants were randomly selected using Statistical Product and Service Solutions (SPSS; 2009).

Arsonist Sample

The arsonist sample consisted of 55 males. The main part of the group (89.1%) was born in the Netherlands, and the remaining six were born in Surinam, Morocco, Canada, or Indonesia. The mean age of the arsonists at the time of their admission was 29.3 (SD = 9.89; range = 18-58 years). The highest level of completed education was elementary school for 34.5% of the participants, middle school for 27.3%, high school for 30.9%, and college/university for 3.6%. Almost half of the arsonists were unemployed (45.3%). The mean treatment duration had been 352 days (SD = 267.92; range = 45-1,264 days).

Violent Offender Sample

The sample of violent offenders contained 41 males. Preceding, two violent offenders were excluded from the sample, because of arson antecedents in their criminal history. Similar to the arsonists, most violent offenders were born in the Netherlands (82.5%), and the remaining seven originated from Morocco, Turkey, Great Britain, and Spain. The mean age of the violent offenders was 29.3 years (SD = 8.85; range = 18-53 years). The highest level of education was elementary school for 45%, middle school for 40%, high school for 5%, and college/university for 7.5 %. For 2.5 % of the participants, the level of education was unknown. About a quarter of the sample was unemployed (25.6%). The mean treatment duration had been 423 days (SD = 403.47; range = 37-2,171 days).

Independent t tests revealed no significant differences between the two research samples regarding age at treatment admittance and treatment duration, t(94) = 0.009, p = .993 and t(94) = −1.028, p = .307, respectively. Differences between birth country, educational level, and employment were tested with chi-square analyses. Only educational levels showed significant differences between arsonists and violent offenders, χ²(1, n = 95) = 10.324, p = .035.

Measures

Following review studies such as Gannon and Pina’s (2010), a number of variables were selected considered relevant to the study of arsonists. These were collected by means of risk assessment. Operationalizations were adopted from items of several risk assessment instruments, such as the Historical Clinical Risk management-20 (HCR-20; Webster, Douglas, Eaves, & Hart, 1997), the Structured Assessment of Violence Risk in Youth (SAVRY; Borum, Bartel, & Forth, 2002), and the Level of Service Inventory–Revised (LSI-R; Andrews & Bonta, 2000). For the present study, these variables were uniformly coded 0 to 2, representing no, moderate, or high presence. Unemployment was marked “yes” if the participant was not employed for 6 months or longer. Alcohol and drug abuse/dependence and DSM Axis I disorder and DSM Axis II disorder were dichotomously (yes/no) scored according to Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; DSM-IV-TR; APA, 2000) criteria. In addition, the types of disorders were recorded. Cognitive impairment was marked “yes” in case of mental retardation; learning disabilities were not regarded as cognitive impairment. Participants were labeled socially isolated if they indicated an absence of social network (excluding family members or partners). Suicidal thoughts were considered present if the client had expressed any of such thoughts in present or past.

Recidivism was defined by the first new conviction for any offense (general recidivism). Following the Dutch Criminal Code, the specific offense types were categorized into arson, physical violence, domestic violence, verbal aggression, vandalism, property crime, and sexual offense. Previous convictions, that is, offenses committed prior to admittance to treatment, were categorized correspondingly. Participants were labeled a specialist if their prior and current offenses were of exclusively one type (arson or violence) and a generalist in case there were convictions concerning two or more offense type categories.

Procedure

Client files were studied to obtain information about the research variables. These files generally consist of an extensive intake, data from judicial and police reports, and reports of other (forensic) mental health care centers. During a 60-min intake session, a psychologist or psychiatrist conducts a structured interview about relevant issues to outpatient forensic care, such as family (situation), education, and work. Moreover, a clinically assessed diagnosis according to the standards of the DSM-IV-TR (APA, 2000) is derived in consultation with a multidisciplinary team (psychiatrist, psychotherapists, and psychologists).

Two psychologists with several years of experience in risk assessing carried out the risk assessments by means of the Risk Assessment Instrument for Forensic outpatients Mental Health (RAF MH; Horn, Wilpert, Eisenberg, Scholing, & Mulder, 2012). This instrument is composed of, inter alia, items of the above-mentioned risk assessment instruments. In earlier research (Horn, Wilpert, Bos, Eisenberg, & Mulder, 2009), the interrater reliability of the items of the RAF MH proved excellent (Intraclass Correlation Coefficient = .88).

To survey the criminal history and recidivism of the sample, information was requested from the judicial authorities (Judicial Documentation System [JDS]). The JDS is managed by the Central Judicial Documentation Service (CJD) and offers an overview of all (legal) persons registered for violations of the law (leading to settled as well as non-settled criminal cases). In this study, only convictions were included as offense. The follow-up time (end of treatment until reference date) varied between 34 and 221 months (M = 111.19 months; SD = 39 months).

Chi-square and Mann–Whitney tests were performed to compare the two study samples regarding the categorical and ordinal variables described under Measures, except for recidivism. Recidivism rates were calculated using Kaplan−Meier survival analysis, which takes individual differences in follow-up periods into account (Bijleveld, Mooijaart, Brouwers, & Loewenthal, 1998; Wartna, 2000). The survival analyses were performed with the dates of the first recidivism offense, and all subsequent reoffenses were excluded. Finally, the log-rank test was utilized to determine differences between the survival curves.

Characteristics of Arsonists and Violent Offenders

The first aim of this study was to investigate the similarity of arsonists and violent offenders regarding several personal characteristics. In Table 1, the numbers of the participants meeting the criteria for the presence of these variables are presented.

OPEN IN VIEWER

Table 1.

Characteristics of Arsonists (n = 55) and Violent Offenders (n = 41) in Numbers (n) and Valid Percentages (%).

	Arsonists		Violent offenders
	n	%	n	%
Young age at first manifestation of antisocial behavior
Younger than 11 years	5	9.1	24	58.5
11-18 years	10	18.2	13	31.7
18 years and older	40	72.7	4	9.8
Victim of child abuse
No presence	36	67.9	20	52.6
Moderate presence	6	11.3	4	10.5
High presence	11	20.8	14	36.8
Impulsivity
No presence	9	17.6	3	7
Moderate presence	21	41.2	17	39.5
High presence	21	41.2	23	53.5
Coping skills
Good	2	3.6	4	7.3
Moderate	14	25.5	16	39
Inadequate	39	70.9	22	53.7
Anger management problems
No presence	3	5.7	2	4.7
Moderate presence	35	66	16	37.2
High presence	15	28.3	25	58.1
Instability of partner relationships
No presence	19	35.8	13	33.3
Moderate presence	22	41.5	13	33.3
High presence	12	22.6	13	33.3
Problems in the relationship with parents
No presence	23	43.4	12	30.8
Moderate presence	9	17	15	38.5
High presence	21	39.6	12	30.8
Alcohol abuse/dependence	34	65.4	23	57.5
Drug abuse/dependence	13	24.5	20	48.8
DSM Axis I disorder	33	63.5	31	81.6
DSM Axis II disorder	31	60.8	13	33.3
Cognitive impairment	9	17.3	2	4.9
Socially isolated	14	26.9	3	8.1
Suicidal thoughts	18	35.3	8	21.6
Past mental health treatment	32	61.5	29	74.4

Note. DSM = Diagnostic and Statistical Manual of Mental Disorders.

Compared with the violent offenders, a significantly greater proportion of the arsonists displayed PDs, χ²(1, n = 90) = 6.665, p = .010), higher levels of social isolation (Fisher’s exact test, p = .030), and inadequate coping skills (U = 938.00, p = .040, r = −.21). The violent offenders, on the other hand, showed a higher incidence of drug abuse/dependence, χ²(1, n = 94) = 5.968, p = .015, as well as a younger age at their first manifestation of antisocial behavior (U = 332.50, p = .000, r = −.65).

Exploring the DSM Axis I and Axis II disorders further, results showed that the arsonists were mostly diagnosed with impulse control disorder (n = 18, 54.5%) and PD not otherwise specified (PD-NOS; n = 15, 68.2%), with traits of narcissism appearing most frequent (31.8%). Four participants (7.3%) were diagnosed with pyromania. The arsonists demonstrated a higher disposition to a combination of Axis I and II disorders compared with the violent offenders (26.9% vs. 19.4%, respectively), although this difference was non-significant, χ²(1, n = 88) = .655, p = .418. In the violent offender sample, the most frequently diagnosed disorders were also impulse control disorder (n = 19, 44.2%) and PD-NOS (n = 13, 76.5%), however, with antisocial traits being most prevalent (43.3%).

Criminal history and recidivism

To further compare arsonists and violent offenders, their criminal careers were studied. Table 2 shows an overview of the criminal history and recidivism by offense type for each study sample.

OPEN IN VIEWER

Table 2.

Criminal History and Recidivism Arsonists (n = 55) and Violent Offenders (n = 41) Based on Convictions for Several Offense Types in Numbers (n) and Percentages (S%).

Offense type	Criminal history		Recidivism
	n	S%	n	S%
Arsonists
General	29	52.7	26	47.3
Arson	14	25.5	5	9.1
Physical violence	8	14.5	13	23.6
Domestic violence	3	5.5	1	1.8
Verbal aggression	2	3.6	3	5.5
Vandalism	9	16.4	6	10.9
Property	15	27.3	7	12.7
Sexual	2	3.6	1	1.8
Violent offenders
General	35	86	25	58.1
Arson	0	0	1	2.3
Physical violence	29	67,4	13	30.2
Domestic violence	0	0	1	2.3
Verbal aggression	5	11.6	1	2.3
Vandalism	12	27.9	4	9.3
Property	24	55.8	5	11.6
Sexual	2	4.6	0	0

More than half of the arsonists (52.7%) had been previously convicted for one or more different types of offenses. The violent offenders surpassed this with 85.4%, resulting in a significant difference, χ²(1, n = 96) = 11.260, p = .001. The discrepancy in the number of generalists within the samples, 65.5% in the arsonists and 36.6% in the violent offenders, was also significant, χ²(1, n = 96) = 7.861, p = .005.

Figure 1 depicts the recidivism rates for general offenses during the first hundred months after treatment. Almost half of the arsonist sample (47.3%) was convicted for a new offense. Out of the five participants reoffending with arson, one was diagnosed with pyromania. The general recidivism rate for the violent offenders was 58.1%, which differed significantly from the arsonists’ reoffending: log rank, χ²(1, n = 96) = 3.864, p = .049.

OPEN IN VIEWER

Figure 1.

General recidivism percentages convictions arsonists (n = 55) and violent offenders (n = 41) following 100 months after treatment.

Limitations and Suggestions for Future Research

The generalizability of the research sample can be considered binary. As studies investigating residential populations predominate the literature, results are harder to compare with previous outcomes. However, the outpatient character of De Waag clients draws the study participants closer to the “normal” population, compared with patients in residential facilities. Outpatients are a group of great importance, because they live in society and have more opportunities to reoffend and claim victims. It seems useful to conduct more research with outpatients to gain more insight into this relatively underexposed group.

Another issue worth addressing is the perhaps seemingly arbitrary assigning of participants to the research samples in this study. The categorizing of offender types itself is a rather precarious operation. In the present study, the choice as to make reason for admittance decisive was born out of necessity. Upon registration, there is little to no information about the criminal history of the participants, almost automatically establishing an offender type based on the referral target, that is, the type of offense. For study purposes, classified information was formally requested from the CJD. Subsequently, looking at the criminal histories of the participants diffuses categorizing offenders (except of course when they are first offender). A possible solution to this could be investigating first offenders exclusively, which could also prevent the possible bias resulting from the inclusion of generalists. The disadvantage, however, would be the inability to survey the development of criminal behavior.

The retrospective nature of the research brings indisputable limitations, because the acquisition of information is restricted to the completeness and quality of client files. Future research could benefit from a prospective approach, in which client interviews and/or self-report questionnaires could be utilized to reduce missing values in relevant information and increase the reliability of research results. Consequently, creating the opportunity to gather a specific set of variables potentially being more relevant to the topic. For example, it could be valuable to include instruments such as The Five Factor Fire Scale (Ó Ciardha et al., 2015) to see whether arsonists distinguish themselves from other offenders regarding fire-related variables such as serious fire interest, identification with fire, and perceived fire awareness, as Gannon et al. (2013) did. A further advantage of a prospective set up could be the possibility to assemble larger research groups, inter alia to be able to investigate specific recidivism (instead of only general) and explore predictors of recidivism to get a clearer picture of criminogenic needs. Moreover, it could be interesting to compare pyromaniacs with arsonists and other types of offenders to search for potential treatment needs. A larger research sample could also create the opportunity to match participants on possibly confounding characteristics, as Gannon et al. accomplished with variables such as number of total offenses, previous engagement with mental health services, and severity of index offense. Furthermore, it could be informative to compare juvenile and adult firesetters to explore whether age is connected to patterns of characteristics leading to firesetting.

Practical Implications

Contrary to expectations, the results of this study do provide some evidence for a slightly different approach for psychotherapeutic treatment of arsonists. Given their presumed enhanced social isolation and lack of coping skills, arsonists might benefit from more attention to relational and emotion regulation skills. Moreover, the outcome of predominating narcissistic PD traits could promote a therapy style taking into account the client’s vulnerabilities and need for acknowledgment (Links & Prakash, 2014).

The question whether arsonists would benefit from more intensive treatment, with, for example, more frequent sessions and longer treatment duration, needs more exploration. It could be assumed that the nature of generalists calls for a more multifaceted approach (targeting multiple areas such as social network and other environmental factors as well as personality), as suggested for generalist adolescent sexual offenders (Pullman & Seto, 2012). However, the outcome of arsonists proving to be more generalists is not very remarkable, as the literature shows generality to be typical and specialization rather exceptional (Brame, Mulvey, Piquero, & Schubert, 2014). Gannon et al. (2013), on the other hand, found enough significant differences to label arsonists specialists, which would advocate specific treatment as opposed to a more general approach.

Either way, the outcome of this study does not support assuming arsonists require less formal supervision and treatment-oriented programming than non-arson offenders, like Caudill, Diamond, Trulson, DeLisi, and Marquart (2012) found to be the case when comparing 5,223 juvenile felony offenders. Rather, it is justifiable to offer treatment specifically tailored for firesetters, for example, the structured intervention program described by Swaffer, Haggett, and Oxley (2001), incorporating sessions about the dangers of fire, development of skills, insight and self-awareness, and relapse prevention.