Risk Factors for Future Offending in Child and Adolescent Firesetters Following a Fire Service Intervention Program

Abstract

Research has established links between youth firesetting and general antisocial behavior. The current study sought to better understand these links by identifying fire-specific and general risk factors for offending from a national sample of children and adolescent firesetters (N = 1,790), from a New Zealand Fire Service intervention program, up to 10 years after intervention. Most (62%) had committed an offense post-intervention, primarily moderate or severe offending. Only 5% had committed an arson offense post-intervention. Nearly all measures of victimization, psychosocial/emotional problems, previous conduct problem behavior, and child-welfare history were associated with post-intervention offending and/or offending severity. Pre-intervention offending and being older (aged 12+ years) at intervention were the major independent risk factors for offending. A protective factor was firesetting having occurred at home. Findings highlight young firesetters’ broad patterns of pervasive antisocial behavior and multiple adverse family, environmental, and individual factors that accompany and contribute to wide-ranging, non-fire-related offending.

Keywords

antisocial behavior juvenile offenders prediction risk factors youth

Introduction

Intentional firesetting represents one of the most complex and costly antisocial behaviors engaged in by a significant minority of young people. Although estimates vary, community studies indicate that firesetting occurs in between 5% and 38% of children and adolescents (Bowling, Merrick, & Omar, 2014; Chen, Arria, & Anthony, 2003; Dadds & Fraser, 2006; Del Bove, Caprara, Pastorelli, & Paciello, 2008; MacKay, Paglia-Boak, Henderson, Marton, & Adlaf, 2009; Martin, Bergen, Richardson, Roeger, & Allison, 2004; Watt, Geritz, Hasan, Harden, & Doley, 2015). Prevalence rates in clinical populations have been reported to be substantially higher, with between 46% and 67% of young people in inpatient and offending samples reporting a history of firesetting (Kolko, Day, Bridge, & Kazdin, 2001; Watt et al., 2015). Many young people engage in repeated firesetting, with recidivism rates shown to be as high as 59% (Del Bove et al., 2008; MacKay et al., 2009).

Unfortunately, in only a small proportion of suspicious fires are authorities able to identify a specific perpetrator, but where they do identify the perpetrators, a significant number are children and young people. For example, in New Zealand between 2016 and 2017, 50% of suspicious fires with an identified perpetrator were caused by young people under the age of 17 years (Fire and Emergency New Zealand, 2017a). On average, 460 child and adolescent firesetters do a fire awareness education program each year (Fire and Emergency New Zealand, 2017b), further indicating that young people’s firesetting is a considerable problem.

Deliberate firesetting by young people is increasingly conceptualized as occurring within the context of antisocial behavior. Firesetting is a diagnostic criterion of conduct disorder (American Psychiatric Association, 2013), and many risk factors for firesetting behavior, such as family dysfunction, abuse history, impulsivity, and poor school performance, are common to conduct problems (Bowling et al., 2014; Hoerold & Tranah, 2014; Lambie & Randell, 2011; Murray & Farrington, 2010; Walsh & Lambie, 2013). A recent study reported that youth offenders were nearly three and a half times more likely to have engaged in firesetting than young people in the community (Watt et al., 2015). Furthermore, antisocial behavior has been found to be the best predictor of self-reported firesetting (Martin et al., 2004; Watt et al., 2015).

Many firesetters fall within the clinical range of externalizing behavior, such as aggression, defiance, and attention deficit hyperactivity disorder (ADHD; MacKay et al., 2009), and many go on to commit moderate to severe general offenses throughout adolescence and early adulthood (Lambie, Ioane, Randell, & Seymour, 2013). The notion that firesetting may be best understood within the context of antisocial behavior is particularly supported by research which shows that when antisocial firesetters and antisocial non-firesetters are compared, few differences are found in terms of aggression, oppositional and conduct-disordered behavior, and demographics (Forehand, Wierson, Frame, Kemptom, & Armistead, 1991; Martin et al., 2004). Also, the pattern of antisocial behavior of the groups does not differ significantly (Stickle & Blechman, 2002).

Longitudinal evidence has supported more theoretical understanding of risks and trajectories of antisocial behavior in children and young people, including the widely used Moffitt taxonomy (Moffitt, Caspi, Harrington, & Milne, 2002). This classifies young offenders into two broad groups, life-course-persistent and adolescent-limited, with the former showing wide-ranging early features of antisocial behavior seen in some young firesetters (Moffitt, 2018). Adolescent-limited offenders engage in the “ordinary” misbehavior associated with transient risk-taking, typical of many adolescents. In contrast, youth who engage in “life-course-persistent” offending are a minority who have typically experienced higher rates of difficult parenting, family violence and abuse, neurocognitive problems, under-controlled temperament, severe hyperactivity, psychopathic personality traits, and violent behavior, often persistently from a very young age and on into adulthood (Moffitt et al., 2002).

Indeed, a small number of studies indicate that firesetting may be a marker of particularly severe antisocial behavior. Within a sample of offenders, firesetting was associated with early onset and severity of antisocial behavior (Stickle & Blechman, 2002). Extreme antisocial behavior has also been associated with adolescent firesetting (Martin et al., 2004), as has being more socially isolated with poorer coping skills than young offenders more generally (Wilpert, van Horn, & Eisenberg, 2017). Becker, Stuewig, Herrera, and McCloskey (2004) found firesetting predicted both violent and nonviolent later delinquency, and child firesetters were 3 times more likely to be referred to juvenile court in adolescence, even after controlling for conduct disorder. Such studies are indicative of the potential seriousness of the increased risk for, and extreme nature of, antisocial behavior exhibited by firesetters.

The conceptualization of firesetting within the context of antisocial behavior may be somewhat limited. In offending populations, it is possible that the presence of fire-specific or other risk factors may distinguish the etiology of firesetting from that of conduct disorder. For example, in a recent study of firesetting in offending and non-offending adolescents, having a greater emotional affect regarding fire and more interest in fire predicted repeat firesetting (Watt et al., 2015). Moreover, fire-specific variables were the only distinguishing factors between young people who set one or two fires, and those who set three or more. Fire-specific variables may therefore contribute to increased firesetting in community adolescents, and distinguish offending adolescents who set fires from those who do not. In addition, Kolko and Kazdin (1991) reported no interaction effect between firesetting status and conduct disorder, indicating that the presence of a conduct disorder diagnosis was not sufficient to explain differences between firesetters and non-firesetters.

These findings suggest that firesetters are not antisocial as a rule and that such an understanding of firesetting is likely applicable only to those who exhibit a range of other antisocial behaviors. In addition, framing firesetting in the context of antisocial behavior may not sufficiently account for the extensive variability in antisociality among firesetters. Further research is required to ascertain to what extent firesetting may be understood in the context of general antisocial behavior (including life-course-persistent vs. adolescent-limited offending), as well as to what degree fire-specific variables may contribute to the fact that some, but not all, antisocial young people go on to set fires.

There is currently a dearth of literature that considers risk factors specific to firesetters. At this time, factors associated with firesetting behavior include male gender, family dysfunction, history of abuse, victimization, anxiety, depression, conduct disorder, oppositional defiant disorder and ADHD symptoms, interpersonal difficulties, cruelty to animals, and fire interest (Lambie & Randell, 2011). More recently, similar fire-specific variables, such as positive affect toward fire and preoccupation with fire (Watt et al., 2015), and educational variables, such as lower academic achievement and a negative attitude toward school (Bowling et al., 2014), have been identified to predict self-reported firesetting. Impulsivity has also been reported to be a significant predictor of the frequency of deliberate firesetting and is recommended as an area to target in both assessment and treatment (Hoerold & Tranah, 2014).

It is now widely reported that adverse childhood experiences are a significant risk factor for poor outcomes later in life, including antisocial and offending behavior that persists into adolescence and adulthood (Baglivio, Wolff, Piquero, & Epps, 2015; Basto-Pereira, Miranda, Ribeiro, & Maia, 2016; Ford, 2002; Fox, Perez, Cass, Baglivio, & Epps, 2015; Hunt, Slack, & Berger, 2017; Mulder, Brand, Bullens, & van Marle, 2010). Research findings also suggest that traumatic early childhood experiences may be particularly relevant to the development, persistence, and severity of firesetting. Given the cross-over between firesetting and general delinquency, the aims of the current study were to investigate risk factors for ongoing offending and severity of offending in a large sample of child and adolescent firesetters.

Method

Sample

This study comprised data from a larger dataset of all children and adolescents who were referred to the New Zealand Fire Awareness and Intervention Programme (FAIP) between July 1, 2003, and June 30, 2007 (n = 2,696). Data were gathered at the time of assessment. Clients were excluded from this study where the fire-related behavior involved accidental firesetting, or was other than actual engagement in firesetting activity (e.g., malicious emergency call; n = 906). The assessment questionnaire primarily focused on risk factors and psychological needs across a broad range of areas including family circumstances, trauma history, firesetting behavior, and psychological and mental health.

The study sample consisted of 1,790 children and adolescents who had been directly involved in the intentional lighting of a fire and underwent the FAIP. Of these, 45% (n = 809) were children (aged 3-11 years) and 55% (n = 981) were adolescents (aged 12-17 years), with an average age of 12 years (SD = 3.2). Ninety percent of the sample (n = 1,602) were male, 67% (n = 1,191) were European, 30% (n = 539) were Māori or Pacific, and the rest (3%) were Asian/Other.

The New Zealand FAIP

The FAIP is a fire service–operated firesafety education program for young people up to the age of 18 years. The manualized program is delivered by fire fighters specially trained to deliver the program, supervised by a clinical psychologist. The intervention typically takes place in the home of the young person with parents/caregivers present. Young people are referred to the program by a number of sources including schools, police, social workers, and parents/caregivers. The program takes a semistructured interview format with the purpose of gathering assessment information concerning the young person and their behavior, which is followed by a firesafety educational component. Determination regarding the number of sessions is made according to the level of identified need for any given individual or family. Families typically receive two to four sessions as well as follow-up phone calls.

Procedure

Data were provided by or collected from the New Zealand Fire Service, the New Zealand Police National Intelligence Agency (NIA) database, and the Oranga Tamariki—Ministry for Children (CYRAS—Care and Protection, Youth Justice, Residential and Adoption Services) national database. Following approval and commission from the New Zealand Fire Service to undertake this study, ethical approval was granted by the University of Auckland Human Subjects Ethics Committee. Data supplied by the New Zealand Fire Service included all individual, environmental, and fire-specific information collected using a standardized questionnaire during the FAIP intervention interview. This information was provided by the child/adolescent and their parent/caregiver.

Oranga Tamariki—Ministry for Children is the statutory child protection agency for New Zealand. Data collected from the Oranga Tamariki database (CYRAS) were drawn by two postgraduate research assistants from file data entries made prior to the date of FAIP intervention. Forty-five percent (n = 801) of the study sample, including 44% of males (n = 709) and 49% of females (n = 92), had CYRAS records; that is, they had had involvement with the state’s child protection services (such as for maltreatment or abuse).

Offending histories provided by the New Zealand Police included the subjects’ entire offending histories from birth until the time of data collection (October 1 to December 14, 2012). Offense records were drawn from both Youth Aid and Court records and included type of offense and outcome of offending. With the intention of capturing broad indications of conduct problem behavior, all offenses were recorded, regardless of outcome, and even if the offending occurred when the individual was under the legal age of criminal responsibility and therefore unable to be charged. Sixty-eight percent of the study sample (n = 1,217), including 68% of males (n = 1,093) and 66% of females (n = 124), had a record of offending; 37% had a pre-intervention record of offending (37% of males; 36% of females). The median post-intervention follow-up was 7.6 years (SD = 1.2 years, range = 5.5-9.5 years).

Ten percent of the sample (n = 180) was randomly selected for analysis of interrater agreement to determine consistency between the two postgraduate research assistants who were the data collectors of the CYRAS data. Of these, 116 had records in CYRAS and the analysis was therefore limited to these 116 subjects. The Cohen’s Kappa values were all greater than .75, which falls within the range of excellent agreement (Cicchetti et al., 2006).

Risk Factor and Outcome Variables

The principal outcome variable was post-intervention offending. We also investigated severity based on the client’s most severe offense (none, minor, moderate, and severe). The risk factor variables, based on both Moffitt’s taxonomy and the firesetting literature, were broadly classed into four categories: three general (non-firesetting-specific) and one firesetting-specific. The four broad categories were demographic and assessment variables, psychosocial/emotional and environmental variables, conduct problem and offending-related variables, and firesetting-specific variables.

Demographic and Assessment Variables

The following risk factors were drawn from data collected by the FAIP: age at FAIP intervention (dichotomized into 0 to 11 years; 12+ years), gender, ethnicity (Māori or Pacific; neither), current residence (home; nonhome), and deprivation. Deprivation was measured using the NZDep2006, which measures the regional deprivation of each young person’s immediate neighborhood using a weighted average of nine Census indicators of socioeconomic status (Salmond, Crampton, & Atkinson, 2007). These scores range from 1 to 10, where 10 represents the greatest level of deprivation. For this study, we dichotomized the scores into top 40% (scores 7, 8, 9, & 10; i.e., most deprived) and lower 60% (less deprived).

Psychosocial/Emotional and Environmental Variables

These were drawn from data collected by the FAIP questionnaire. “Psychiatric diagnosis ever” comprised specific answers to the question, “Does the client have any of the following diagnoses: Attention-Deficit/Hyperactivity Disorder (ADHD); Conduct Disorder (CD); Oppositional Defiant Disorder (ODD); Obsessive Compulsive Disorder (OCD); Major Depressive Disorder (MDD), as determined by a medical professional?” “Psychosocial/emotional problems” that were specifically recorded were hyperactivity, poor concentration, depressive symptoms, suicidal/self-harm behaviors, anger, anxiety (fear or worrying a lot), and learning difficulties. Victimization variables were created by combining FAIP and CYRAS victimization data. This collation of sources was intended to minimize the effects of underreported victimization and increase accuracy of data. The risk variable “experienced abuse/neglect” was comprised of any record of physical abuse victimization, sexual abuse victimization, and/or neglect. The risk factor of “child-welfare agency involvement” was drawn from CYRAS, representing any referral to government social services for suspected and/or documented history of some form of child abuse, neglect, and/or maltreatment, as assessed by two statutory social workers trained in child-abuse assessment. Also drawn from CYRAS was “domestic violence in family” (violence in the home environment not directly involving the individual, including physical violence and property damage but not verbal violence).

Conduct Problem and Offending-Related Variables

These were drawn from various sources. The FAIP provided data for the variables: aggression to people or animals, deceitfulness or theft, destruction of property, serious violation of rules, and group offending. Sexualized behavior before age 12 (i.e., record of an individual displaying inappropriate sexualized behavior prior to 12 years of age) was derived from CYRAS. All these measures were combined into the factor of “conduct problems.”

Offense variables were created from each individual’s offending history, extracted from the New Zealand Police NIA database. Offense categories were determined in consultation with the police and were based on the extent of damage or harm to an individual and/or property. There were nine offending types (violence, sexual, arson, property damage, drugs and antisocial, dishonesty, traffic, administrative, and miscellaneous), and three severity groups of severe (e.g., violence, sexual, and antisocial offenses), moderate (e.g., violence, sexual, property damage [including arson], drugs and antisocial, dishonesty, miscellaneous, and traffic offenses), and minor (e.g., drugs and antisocial, dishonesty, miscellaneous, administrative, and traffic offenses). More detail regarding the types of offenses that fall into the type and severity groups has been described previously (Lambie et al., 2013). The variable “pre-intervention offending” comprised any of the aforementioned offense categories that had been reported prior to FAIP intervention. If there were any police data records related to mental health, exposure to family violence, or any other concerns not of an offending nature, this was cross-checked with the CYRAS database to ensure any pertinent information was included in the final dataset.

Firesetting-Specific Variables

These were drawn from data collected by the FAIP. There was a basic count of the number of pre-intervention firesetting incidents; one measure pertaining to where the firesetting occurred (firesetting occurred at home), five pertaining to motivation (drawn from a multi-response question asking the client about the reason[s] for their firesetting: antisocial, attention, boredom, experimentation, peer pressure), and one each pertaining to whether the intention of the client’s firesetting was property destruction, whether the firesetting was undertaken as part of a group, and whether accelerant was used. The variable “negative feelings” concerned how lighting the fire made the client feel.

Data Analysis

Data analyses were undertaken using the JMP V12.0 (SAS Inc.) and StatsDirect V3.0 (StatsDirect Ltd.) software. To investigate risk factors of any post-intervention offending (yes/no), contingency table analyses using Fisher’s exact tests were conducted and odds ratios (OR) with 95% confidence intervals (CIs) calculated. Ordinal logistic regression analysis with an effect likelihood ratio test was used to investigate risk factors of post-intervention offending severity. Correlations between general and firesetting-specific risk factors were assessed using the phi (ϕ) coefficient. Correlations were classified according to the Evans (1996) classification: <.20 very weak, .20 to .39 weak, .40 to .59 moderate, .60 to .79 strong, and >.80 very strong. Whole-model multivariable logistic regression was used to provide an investigation of the combined ability of the risk factors included in this study to predict post-intervention offending, and to take into account any confounding effects of risk factors on each other. The overall ability of models to predict post-intervention offending was assessed using the area under the receiver operating characteristic (AUROC) curve.

Results

Post-Intervention Offending

As shown in Table 1, 1,105 (62%) of the sample of firesetters committed an offense in the post-intervention follow-up period. Based on offense-severity groupings, 361 (20%) had engaged in offending where at least one offense was classified as severe. Only 89 (5%) firesetters committed arson post-intervention.

Table 1:

Post-Intervention Offending

Offense	n (%)^a
Any	1,105 (62)
Offense count
Zero	685 (38)
1 or 2	328 (18)
3 or 4	180 (10)
5-10	264 (15)
>10	333 (19)
Severity group
Severe	361 (20)
Moderate	656 (37)
Minor	88 (5)
Outcome
Convicted	665 (37)
Custody	131 (7)
Offense type^b
Violence	522 (29)
Sexual	53 (3)
Arson	89 (5)
Property damage	514 (29)
Drugs and antisocial	590 (33)
Dishonesty	737 (41)
Traffic	535 (30)
Administrative	218 (12)
Miscellaneous	359 (20)

Percentage of subjects who exhibited that offending characteristic in the follow-up period. ^bWhere clients committed more than one offense type, they are counted for each type.

Risk Factors for Post-Intervention Offending

Demographic Risk Factors

Male firesetters were not statistically more likely to offend than females (OR = 1.3, p = .06); however, males were significantly more likely to engage in more severe offending types (p = .001, 21% vs. 12% severe; Table 2). The odds of post-intervention offending among those aged 12+ years were 3.6 times that of younger children (p < .001); adolescents were also significantly more likely to engage in more severe offending (p < .001, 26% vs. 13% severe). The odds of post-intervention offending in firesetters residing in an institution or other nonhome setting were 2.4 times that of those residing at home (p < .001). Not living at home was also a significant risk factor for more severe offending (p < .001, 38% vs. 19% severe). Ethnicity and deprivation were not significant univariable risk factors of post-intervention offending among this group of firesetters, although Māori or Pacific firesetters were marginally more likely to engage in more severe offending (p = .04, 23% vs. 19% severe).

Table 2:

General Risk Factors for Post-Intervention Offending and Offending Severity

Variable	Offending		OR (95% CI)^a	p	R ²	Offending severity^b			χ²	p	R ²
	No(n = 685)	Yes(n = 1,105)				Minor(n = 88)	Moderate(n = 656)	Severe(n = 361)
	n (%)	n (%)				n (%)	n (%)	n (%)
Demographic
Age at intervention
12+ years	244 (25)	737 (75)	3.6 [3.0, 4.4]	<.001	.07	49 (5)	430 (44)	258 (26)	166.4	<.001	.04
0-11 years	441 (55)	368 (45)				39 (5)	226 (28)	103 (13)
Gender
Male	601 (38)	1,001 (62)	1.3 [1.0, 1.8]	.06	.00	68 (4)	595 (37)	338 (21)	10.5	.001	.00
Female	84 (45)	104 (55)				20 (11)	61 (32)	23 (12)
Ethnicity
Māori or Pacific	190 (35)	349 (65)	1.2 [1.0, 1.5]	.09	.00	28 (5)	198 (37)	123 (23)	4.4	.04	.00
Non-Māori/non-Pacific	483 (40)	738 (60)				60 (5)	450 (37)	228 (19)
Deprivation
Top 40%	348 (37)	589 (63)	1.1 [0.9, 1.4]	.30	.00	46 (5)	337 (36)	206 (22)	3.0	.08	.00
Lower 60%	310 (40)	471 (60)				37 (5)	296 (38)	138 (18)
Current residence
Institution or other	21 (21)	77 (79)	2.4 [1.4, 3.9]	<.001	.01	3 (3)	37 (38)	37 (38)	22.6	<.001	.01
Home	663 (39)	1,027 (61)				85 (5)	618 (37)	324 (19)
Psychosocial/emotional and environmental
Psychiatric diagnosis ever
Yes	148 (32)	321 (68)	1.5 [1.2, 1.9]	<.001	.01	12 (3)	175 (37)	134 (29)	28.8	<.001	.01
No	537 (41)	784 (59)				76 (6)	481 (36)	227 (17)
Psychosocial/emotional problems
Yes	352 (34)	677 (66)	1.5 [1.2, 1.8]	<.001	.01	45 (4)	380 (37)	252 (24)	30.4	<.001	.01
No	332 (44)	428 (56)				43 (6)	276 (36)	109 (14)
Experienced abuse/neglect
Yes	191 (28)	487 (72)	2.0 [1.6, 2.5]	<.001	.02	30 (4)	258 (38)	199 (29)	70.2	<.001	.02
No	482 (44)	615 (56)				58 (5)	396 (36)	161 (15)
Child-welfare agency involvement
Yes	231 (29)	570 (71)	2.1 [1.7, 2.5]	<.001	.02	31 (4)	306 (38)	233 (29)	88.4	<.001	.02
No	445 (46)	531 (54)				57 (6)	347 (36)	127 (13)
Domestic violence in family
Yes	85 (25)	249 (75)	2.0 [1.6, 2.7]	<.001	.01	11 (3)	133 (40)	105 (31)	44.7	<.001	.01
No	591 (41)	849 (59)				77 (5)	518 (36)	254 (18)
Conduct problems/offending
Conduct problems
Yes	363 (31)	798 (69)	2.3 [1.9, 2.8]	<.001	.03	60 (5)	441 (38)	297 (26)	89.8	<.001	.02
No	322 (51)	307 (49)				28 (4)	215 (34)	64 (10)
Pre-intervention offending
Yes	112 (17)	552 (83)	5.1 [4.1, 6.5]	<.001	.09	30 (5)	293 (44)	229 (34)	257.5	<.001	.06
No	573 (51)	553 (49)				58 (5)	363 (32)	132 (12)

Note. OR = odds ratio; CI = Confidence Interval.

OR with 95% CIs, for offending versus no offending. ^bSeverity, including no offending, was analyzed as an ordinal measure.

Non-Firesetting-Specific Risk Factors

All psychosocial/emotional and environmental factors were significant risk factors of both post-intervention offending and post-intervention offending severity (p < .001; Table 2). The OR (ranging from 1.5 to 2.0) and R² values (ranging from .01 to .02) for these variables were relatively modest in the context of this study. Conduct problems were also significantly associated with both post-intervention offending (OR = 2.3, p < .001, R² = .03) and post-intervention offending severity (p < .001, 26% vs. 10% severe; Table 2). Similarly, pre-intervention offending was significantly associated with post-intervention offending (OR = 5.1, p < .001, R² = .09) as well as post-intervention offending severity (p < .001, 34% vs. 12% severe).

Firesetting-Specific Risk Factors

Firesetting-specific risk factors most associated with post-intervention offending were that the destruction of property was intended (OR = 2.5, p < .001) and being part of a group at the time of the referral fire incident/s (OR = 1.9, p < .001; Table 3). Firesetting having occurred at home and being motivated by experimentation were negatively associated with post-intervention offending (OR = .3 and .5, respectively), as well as post-intervention offending severity (p < .001, 11% vs. 24% and 13% vs. 23% severe, respectively). Antisocial, attention, or peer pressure motivations were not significantly associated with post-intervention offending or post-intervention offending severity. Negative feelings were negatively associated with post-intervention offending (OR = 0.8, p = .01) and severity (p < .001, 17% vs. 23% severe).

Table 3:

Firesetting-Specific Risk Factors for Post-Intervention Offending and Offending Severity

Variable	Offending		OR (95% CI)^a	p	R ²	Offending severity^b			χ²	p	R ²
	No(n = 685)	Yes(n = 1,105)				Minor(n = 88)	Moderate(n = 656)	Severe(n = 361)
	n (%)	n (%)				n (%)	n (%)	n (%)
How many firesetting incidents pre-intervention
>4	125 (31)	283 (69)	1.5 [1.2, 2.0]	.002	.01	11 (3)	159 (39)	113 (28)	24.9	<.001	.01
2-4	253 (40)	373 (60)	1.0 [0.8, 1.2]			30 (5)	223 (36)	120 (19)
First time	272 (40)	403 (60)	Reference			43 (6)	249 (37)	111 (16)
Firesetting occurred at home
Yes	335 (59)	229 (41)	0.3 [0.2, 0.3]	<.001	.06	20 (4)	146 (26)	63 (11)	148.4	<.001	.03
No	350 (29)	876 (71)				68 (6)	510 (42)	298 (24)
Motivation
Antisocial
Yes	46 (34)	88 (66)	1.2 [0.8, 1.8]	.36	.00	6 (4)	48 (36)	34 (25)	2.1	.15	.00
No	639 (39)	1,017 (61)				82 (5)	608 (37)	327 (20)
Attention
Yes	44 (45)	53 (55)	0.7 [0.5, 1.1]	.16	.00	5 (5)	26 (27)	22 (23)	0.8	.36	.00
No	640 (38)	1,052 (62)				83 (5)	630 (37)	339 (20)
Boredom
Yes	165 (30)	382 (70)	1.7 [1.3, 2.1]	<.001	.01	25 (5)	226 (41)	131 (24)	22.0	<.001	.01
No	519 (42)	723 (58)				63 (5)	430 (35)	230 (19)
Experimentation
Yes	249 (52)	233 (48)	0.5 [0.4, 0.6]	<.001	.01	26 (5)	146 (30)	61 (13)	57.3	<.001	.01
No	435 (33)	872 (67)				62 (5)	510 (39)	300 (23)
Peer pressure
Yes	102 (36)	183 (64)	1.1 [0.9, 1.5]	.39	.00	13 (5)	95 (33)	75 (26)	3.8	.05	.00
No	582 (39)	922 (61)				75 (5)	561 (37)	286 (19)
Destruction of property intended
Yes	49 (22)	178 (78)	2.5 [1.8, 3.5]	<.001	.01	10 (4)	100 (44)	68 (30)	34.1	<.001	.01
No	636 (41)	927 (59)				78 (5)	556 (36)	293 (19)
Part of a group at the time of fire
Yes	472 (35)	896 (66)	1.9 [1.6, 2.4]	<.001	.01	70 (5)	529 (39)	297 (22)	32.2	<.001	.01
No	213 (50)	209 (50)				18 (4)	127 (30)	64 (15)
Accelerant used
Yes	125 (31)	281 (69)	1.5 [1.2, 1.9]	<.001	.01	16 (4)	172 (42)	93 (23)	12.2	<.001	.00
No	560 (40)	823 (60)				72 (5)	484 (35)	267 (19)
Negative feelings when lighting the fire
Yes	336 (41)	475 (59)	0.8 [0.6, 0.9]	.01	.00	46 (6)	292 (36)	137 (17)	11.8	<.001	.00
No	349 (36)	630 (64)				42 (4)	364 (37)	224 (23)

Note. OR = odds ratio; CI = Confidence Interval.

OR with 95% CIs, for offending versus no offending. ^bSeverity, including no offending, was analyzed as an ordinal measure.

Correlation Between Risk Factors

There was a strong correlation between child-welfare agency involvement and experienced abuse/neglect (ϕ = .77). Moderate correlations were found between child-welfare agency involvement and domestic violence in family (ϕ = .53), experienced abuse/neglect and domestic violence in family (ϕ = .48), and between psychosocial/emotional problems and both psychiatric diagnosis (ϕ = .44) and conduct problems (ϕ = .43). There was only weak or very weak correlation (ϕ < .40) between the other combinations of general risk factors.

Among the firesetting-specific risk factors for offending, there were only weak or very weak correlations (ϕ < .40). The exception was a negative correlation between firesetting occurred at home and being part of a group at the time of fire (ϕ = .52). Between firesetting-specific risk factors and general risk factors, all correlations were ϕ < .35, except for weak negative correlations between firesetting occurred at home and both pre-intervention offending (ϕ = .35) and age at intervention (ϕ = .36).

Exploratory Multivariable Analyses

Full-model multivariable analyses were undertaken to assess the independent effect of the risk factor variables on post-intervention offending and to explore the combined ability of the variables to predict post-intervention offending. These analyses were limited to 1,601 clients (89%) who had no missing data for the risk factors investigated. First, we investigated a full-model analysis including all but the firesetting-specific risk factors. Due to the relatively high correlation between child-welfare agency involvement and both experienced abuse/neglect (ϕ = .77) and domestic violence in family (ϕ = .53), the latter two risk factors were also excluded (data not shown). Pre-intervention offending and the age 12+ years at intervention had the strongest independent associations with post-intervention offending (OR > 2.0). Child-welfare agency involvement, conduct problems, male gender, and Māori or Pacific ethnicity also had significant independent associations with post-intervention offending, although these risk factors had smaller B-estimates and ORs (1.5 < OR < 2.0). The overall predictive value of this model was fair (AUROC = .76; Tape, 2006).

The effect of adding firesetting-specific risk factors to the full-model analysis was then investigated. To lessen the complexity of the model, only factors that were significant (p < .05) in univariable analyses were included. Table 4 presents the full-model analysis with risk factors displayed in descending order of the absolute B-estimate and the multivariable OR for each risk factor. As with the previous analysis, pre-intervention offending and age 12+ years had the strongest associations with post-intervention offending (OR > 2.0). Of the firesetting-specific risk factors, firesetting occurred at home had a strong independent negative association with post-intervention offending. There was only modest improvement to the overall predictive value of the model following the inclusion of the firesetting-specific factors (AUROC = .78). The cumulative predictive value of the model was checked as the six strongest associated factors were added one by one, in descending order of the full-model B-estimates, starting with pre-intervention offending and finishing with ethnicity.

Table 4:

Exploratory Multivariable Logistic Regression Model of Post-Intervention Offending With General Risk Factors and Firesetting-Specific Risk Factors

Variables^a	B	SE	χ²	OR (95% CI)^b	p
Intercept	.68	.20	10.7		<.001
Pre-intervention offending [Y]	.47	.07	41.3	2.5 [1.9, 3.4]	<.001
Age at intervention [12+ years]	.37	.07	29.9	2.1 [1.6, 2.8]	<.001
Firesetting occurred at home [Y]	–.36	.09	17.9	0.5 [0.3, 0.7]	<.001
Child-welfare agency involvement [Y]	.27	.06	17.2	1.7 [1.3, 2.2]	<.001
Conduct problems [Y]	.23	.07	10.8	1.6 [1.2, 2.1]	.001
Ethnicity [Māori or Pacific]	.21	.07	9.9	1.5 [1.2, 2.0]	.002
Destruction of property intended [Y]	.20	.10	3.8	1.5 [1.0, 2.2]	.05
Gender [male]	.19	.10	3.6	1.5 [1.0, 2.1]	.05
Motivated by experimentation [Y]	–.12	.07	2.9	0.8 [0.6, 1.0]	.09
Current residence [institutional/other]	.12	.15	0.6	1.3 [0.7, 2.3]	.44
Negative feelings when lighting the fire [Y]	–.09	.06	2.2	0.8 [0.7, 1.1]	.14
Deprivation [top 40%]	.08	.06	1.6	1.2 [0.9, 1.5]	.21
Psychosocial/emotional problems [Y]	.07	.07	1.0	1.2 [0.9, 1.5]	.30
Motivated by boredom [Y]	.06	.07	0.8	1.1 [0.9, 1.5]	.37
Psychiatric diagnosis ever [Y]	.05	.08	0.4	1.1 [0.8, 1.5]	.52
Part of a group at the time of fire [Y]	.03	.08	0.1	1.1 [0.8, 1.5]	.71
How many firesetting incidents pre-intervention [first time]	.00	.09	0.0	0.9 [0.7, 1.3]	.97
How many firesetting incidents pre-intervention [2-4]	–.06	.08	0.6	0.9 [0.6, 1.2]	.45
Accelerant used [N]	.00	.08	0.0	1.0 [0.7, 1.3]	.99

Note. OR = odds ratio; CI = confidence interval; AUROC = 0.78, R² = .18.

Firesetting-specific risk factors are in bold. ^bMultivariable OR with 95% CIs, for offending versus no offending.

These exploratory multivariable analyses indicated that pre-intervention offending and age 12+ years were major, independent, general risk factors for any post-intervention offending in child and adolescent firesetters. The model comprising only those two risk factors had AUROC = .72 (Supplemental Figure S1 is available in the online version of this article) and R² = .12 (data not shown). Firesetting occurred at home was a major protective factor for post-intervention offending, while child-welfare agency involvement and conduct problems were lesser risk factors (1.5 < OR < 2.0) among these child and adolescent firesetters. The model comprising these five risk/protective factors had AUROC = .76 and R² = .16. Māori or Pacific ethnicity, destruction of property intended, and male gender were additional independent general risk factors in this sample; however, the specific effects were relatively weak (multivariable OR = 1.5, .001 < p < .05).

Discussion

This study aimed to investigate risk factors for post-intervention offending in a sample of children and adolescents who had engaged in firesetting and were referred to the New Zealand FAIP program. The findings contribute to a growing body of literature suggesting that firesetting is best understood within the broad context of antisocial and conduct problem behaviors. Particularly striking in this sample of child and adolescent firesetters were the high rates of antisocial and offending behaviors exhibited both prior to and following FAIP intervention.

Prior to FAIP intervention, two thirds of the sample was reported to have engaged in conduct problem behaviors other than firesetting, and more than one third had committed an offense. Antisocial behaviors and offending rates were evident to an even greater extent following FAIP intervention, with the majority of the sample (62%) having committed an offense in the follow-up period, and 57% having committed a “moderate” or “severe” offense. Notably, only a small proportion of the sample (5%) went on to commit an arson offense following FAIP intervention. As such, these findings illustrate the importance of antisocial behavior as a consideration in firesetting research and intervention, as has been evidenced by previous research (Becker et al., 2004; Del Bove & Mackay, 2011; Lambie et al., 2013; Martin et al., 2004; Stickle & Blechman, 2002; Watt et al., 2015). Furthermore, the high prevalence of offending in this sample supports previous suggestions that firesetters fall on the severe end of the antisociality spectrum and represent a subpopulation with particularly high intervention needs (Lyons, McClelland, & Jordan, 2010; Martin et al., 2004; Stickle & Blechman, 2002), fitting within the life-course-persistent offending trajectory (Moffitt, 2018).

As expected, given their established associations with general offending behavior among children and adolescents (Farrington, Loeber, & Ttofi, 2012; Loeber & Farrington, 2001), all of the psychosocial/emotional, environmental, and conduct problem behavior risk factors were associated with post-intervention offending and/or offending severity. Overall, these associations indicated that the risk factors for future offending among young firesetters are comparable with those among general young offenders, highlighting the important need for non-fire specialist interventions for many young firesetters. Most of the firesetting-specific risk factors were associated with post-intervention offending as well as offending severity, with the exception of antisocial and attention-seeking motivations for the referral fire incident. This indicates the importance of firesetting as a key risk variable for subsequent offending in some children.

Pre-intervention offending had the strongest univariable association and largest individual effect size for post-intervention offending and post-intervention offending severity. This is unsurprising as prior antisocial behavior is known to be one of the most robust predictors of later offending behavior (Andrews & Bonta, 2016; Grieger & Hosser, 2014; Moffitt et al., 2002). Being 12 years and older was another significant risk factor of post-intervention offending with a relatively large effect. Generally, an earlier onset of behavioral problems has been found to be related to an increased risk for later offending and antisocial behavior, including to the life-course persistent framework (Farrington, 1995; Moffitt et al., 2002; Odgers et al., 2008).

The findings of the current study likely reflect the development and developing severity of antisocial behaviors throughout childhood and into adolescence, and the increased autonomy, peer group influence, and risk-taking that the adolescent developmental period entails (Monahan, Steinberg, & Piquero, 2015). It could be argued that firesetting is going to be most captivating for an adolescent who, well into their teens, is more likely than a child to have such autonomy, to have established a peer group keen to help set the fires and to be already engaging in high-stakes, group-based risk-taking (Farrington et al., 2012). Nonetheless, these findings do not imply that firesetting in younger children (under 12 years) should be dismissed as benign. The potential risk of significant behavioral issues among younger firesetters may not have been fully captured in this study due to measures of antisocial behavior post-intervention being restricted to offending identified by police, and a restricted follow-up period.

The two other fire-specific variables were being a member of a group and intending to destroy property. These may relate to broader indicators of antisocial behavior in terms of peer influence driving behavior and overriding any individual, prosocial behavioral choices (Monahan et al., 2015). It is likely that if firesetting is taking place in a group, that may indicate that the individual is well-connected to a highly antisocial group, which provides more opportunity to engage in a range of other antisocial behaviors.

The factors with the next largest effects in association with post-intervention offending were conduct problems, experienced abuse/neglect, and child-welfare agency involvement. Conduct problems, like pre-intervention offending, reflect the well-known association between antisocial behavior and later offending (Moffitt et al., 2002). Findings of abuse/neglect and child-welfare agency involvement are consistent with literature showing that both family dysfunction and experience of abuse in childhood are associated with risk for antisocial behavior (Baglivio et al., 2015; Dargis, Newman, & Koenigs, 2016; Farrington, 2005; Fox et al., 2015; Hunt et al., 2017; Smith, Ireland, Thornberry, & Elwyn, 2008; Sousa et al., 2011). Furthermore, previous research has shown that many child and adolescent firesetters experience much more significant trauma than do non-firesetters (Lyons et al., 2010). Similarly, among a sample of adolescent firesetters, young people with more extensive histories of abuse and contact with child-welfare services engaged in significantly more firesetting behaviors than did those who had experienced a more stable upbringing (Del Bove & Mackay, 2011).

All of the risk factors investigated in this study increased the risk for severe post-intervention offending to some extent, but effect sizes were less than for the variable of post-intervention offending overall (which was a dichotomous yes/no variable). This is partly explained by the numbers in each of the severity categories: Minor post-intervention offending was rare in this group (5%); no post-intervention offending (which was unchanged for both measures) comprised 38%; and severe post-intervention offending was committed by 20% of the sample. Overall, there was evidently no additional statistical power in defining post-intervention offending severity in the way that we did for this study group. Nevertheless, the post-intervention offending severity results do highlight and quantify the increased risk for severe post-intervention offending. Given studies examining offending risk factors inform clinical understandings of risk assessment, and that needs-based services are considered best practice (Lipsey, Howell, Kelly, Chapman, & Carver, 2010; Singh et al., 2014), these results contribute to understandings of those who are most in need of intervention.

The exploratory multivariable analyses largely reflected the univariable results in that pre-intervention offending and being age 12+ years at intervention were the major risk factors for post-intervention offending among child and adolescent firesetters. Child-welfare agency involvement and conduct problems were lesser significant risk factors. Firesetting that occurred at home was a major protective factor. Importantly, these multivariable analyses demonstrate that these factors are to some extent independent of each other in this sample, contributing something distinct to the overall risk while in the presence of each other. Most (92%) of the predictive ability of the multivariable model came from just the pre-intervention offending and age 12+ years risk factors. These multivariable analyses help to gain further understanding of the relationship and relative importance of the various measures and highlight the importance and relatively large impact of pre-intervention offending and age 12+ years on future offending. However, the predictive ability of the model is “fair” at best. Thus, the relatively modest effects observed mean these analyses have not provided any sort of “gold standard” set of predictors of future offending that could be considered universally applicable to every individual who engages in firesetting behavior.

Of the general demographic factors, there was only weak evidence of an association of gender or ethnicity with post-intervention offending (association with offending severity only) and no evidence of an association between deprivation and post-intervention offending. Although male gender has generally been found to be strongly associated with child and adolescent rates of delinquency and offending (Esbensen, Peterson, Taylor, & Freng, 2010; Estévez & Emler, 2011; Loeber & Farrington, 2000), the weak association between gender and post-intervention offending in the present study may be attributed to the small proportion of females included in this study. In addition, there were high rates of pre-intervention offending among both males and females in the current study sample. Therefore, given this sample was already a high-risk group and recidivism has been shown to be high for male and female adolescent offenders (Trulson, Marquart, Mullings, & Caeti, 2005), the minimal effect of gender on post-intervention offending could be expected.

In regard to ethnicity, it is also important to acknowledge that just under a third of the sample included Māori or Pacific youth, and this may explain the weak association between ethnicity and post-intervention. In New Zealand, Māori young people are typically overrepresented in adolescent offending statistics, and Pacific young people overrepresented in violent offending (Ioane, Lambie, & Percival, 2016; Ministry of Justice, 2018), but there is no clear picture regarding rates of firesetting. Sample size may also relate to generally lower rates of access to intervention programs for Māori and Pacific young people (owing to cultural and social barriers, compounded by socioeconomic disadvantage) which likely also apply to accessing the FAIP program. Further investigation would be valuable.

There was a weak association for peer pressure as a motivation for post-intervention offending severity. These findings may be affected by the self-report nature of these measures. Understandably, a young person may be less willing to disclose their firesetting intentions if these were antisocial, to gain attention or due to peer pressure. However, considering the intent to destroy property is clearly antisocial, it may be concluded that firesetters with antisocial motivations are at increased risk of future offending behavior. Notably, no mental health or psychosocial/emotional measures included in the current study had large effect sizes in association with post-intervention offending or post-intervention offending severity. This was consistent with the general offending literature which indicates that the presence of mental health difficulties is not a strong or meaningful predictor of recidivism (Andrews & Bonta, 2016; Phillips et al., 2005).

Firesetting occurring at home was the only firesetting-specific risk factor with a relatively large effect size, which was protective in the affirmative. This is a novel finding but may relate to the fact that a fire being set at home is less likely to be done with peers present (therefore, the child is potentially less connected to an antisocial peer group). Being able to set a fire at home may be evidence of a lack of supervision and parental monitoring, or a wish to express anger toward parents/caregivers within a complex relationship, but not to the degree of an older adolescent intending deliberate property harm with a group of antisocial peers.

These findings suggest that for many firesetters, it may not be their firesetting behavior that is of primary concern. Instead, it is their broad pattern of pervasive antisocial behavior that is key, along with the multitude of adverse family, environmental, and individual factors that accompanies and contributes to these behaviors. Thus, this study supports previous research indicating that firesetting may best be conceptualized within the context of generalist rather than specialist antisocial behavior and that, in some children, it serves as a marker of a trajectory of life-course-persistent offending and its complex psychosocial and environmental sources (Dolan, McEwan, Doley, & Fritzon, 2011; Lambie et al., 2013; Martin et al., 2004; Moffitt, 2018; Watt et al., 2015).

Strengths and Limitations

Unlike the majority of existing research concerning firesetting populations, the current study included a broad range of individual and environmental measures, including those pertaining to antisocial behavior. We were thus able to investigate how these were associated with offending over an extended follow-up period within a large sample. Examining risk for future offending among firesetting populations is novel within the existing body of firesetting research and provides further understanding of the extent to which firesetting is embedded within a broad framework of antisocial behavior.

The current study did not include a measure of age of firesetting onset or age of onset of antisocial behavior, which are known to be risk factors for persistent antisocial behavior. Due to the variable age range of the current study’s sample, some high-risk young people may have been relatively young at the end of the follow-up period, and their risk for post-intervention offending may not yet have developed into offending behavior. Furthermore, the data presented are based on a large and diverse cohort of both children and adolescents, and identifies predictors in that group as a whole. Some variables may be more or less relevant for young people at different ages.

The measure of offending severity in this study was somewhat limited with an individual’s offending severity being based on the severity of their most serious offense. Alternative definitions of severity, which incorporated offending frequency and the number of offenses committed within each severity grouping, may have been more sensitive. Also, although this study looked at any contact with the police as a marker for reoffending (broader than solely using conviction data), it is still limited by the fact that a large proportion of deliberate firesetting and antisocial behavior goes undetected by authorities, and the participants were effectively a “treated” sample, as they had done the FAIP program. The data for this study were collated retrospectively. This restricted the choice of measures to those routinely available, and lowered data completeness and quality, in comparison with studies that are more prospective in design.

Clinical Implications and Conclusions

Of the many conduct problem-related behaviors that children and adolescents engage in, firesetting is particularly high risk in terms of its potential harm. Consequently, firesetting may be the behavior which first comes to the attention of agencies as a result of fire-service attendance to the fire incident or referral to a fire service–based intervention, such as the FAIP. Given the complex and high needs of a proportion of firesetters who attend fire-service programs, it is unlikely that such programs can adequately meet the needs of all firesetters. However, these programs may be uniquely positioned to contribute to early intervention by engaging at-risk young people with appropriate, more intensive services through assessment and referral on.

The findings of the current study indicate the need for increased awareness of the identified risk factors by fire-service programs and child and adolescent mental health services, so these factors can be subsequently addressed adequately through intervention. As such, these findings suggest the necessity to shift away from the traditional focus of assessing risk for firesetting recidivism, and a move toward the adoption of validated risk assessment measures for a broader range of difficulties (such as the Strengths and Difficulties Questionnaire, a brief, validated screening measure of emotional, behavioral, and relationship difficulties that has been used with adolescent firesetters to some extent; see Lambie & Krynen, 2017). This is not to say that assessment of risk for firesetting recidivism is not important. For example, given that fire affect, fire interest, and preoccupation with fire have been found to be predictive factors of repeated firesetting in previous research (Watt et al., 2015), fire-specific variables should be addressed in the assessment of firesetters. Nonetheless, a wider perspective of the antisocial behavior and family environment of many firesetters is necessary.

A thorough assessment and referral process, as well as the subsequent engagement of multidisciplinary agencies, may result in reduced antisocial behavior and offending rates of young firesetters more generally. Specifically, the interventions could be structured to first address any risk behaviors—both fire and general offending—then complete the fire education program. Hand-in-hand, assessment and referral on should address the risk, needs, and responsivity factors for the young person and their family (Andrews & Bonta, 2016); that is, ensure that the level of intervention matches the severity of the young person’s risks, particularly in terms of their potential trajectory into life-course-persistent offending. For this FAIP sample, their high level of post-intervention but non-fire-related offending points to the need for much more intensive service input as soon as they get referred to FAIP.

Supplemental Material

CJB842907_Supplemental_Material_CLN – Supplemental material for Risk Factors for Future Offending in Child and Adolescent Firesetters Following a Fire Service Intervention Program

Supplemental material, CJB842907_Supplemental_Material_CLN for Risk Factors for Future Offending in Child and Adolescent Firesetters Following a Fire Service Intervention Program by Ian Lambie, Isabel Randell, Ariana Krynen, Peter Reed and Julia Ioane in Criminal Justice and Behavior

Footnotes

Authors’ Note:

We would like to thank the New Zealand Fire Service Commission, Oranga Tamariki—Ministry for Children, and the Ministry of Education for funding the research. We would also like to thank the staff at the New Zealand Fire Service’s Fire Awareness and Intervention Programme. Thanks to the reviewers for their thoughtful and detailed responses to our work.

Supplemental Material

Supplemental Figure S1Supplemental Material is available in the online version of this article at .

ORCID iD

Ian Lambie

Ian Lambie teaches clinical and forensic psychology at the University of Auckland and is chief science advisor for the Justice Sector, New Zealand Government. His research interests include child and adolescent juvenile justice.

Isabel Randell is a practicing clinical psychologist with a particular interest in adolescent mental health. Her research interests include the intersection of psychology and law, particularly in relation to sexual violence trials. She is currently researching the well-being of young sexual violence complainants in New Zealand’s justice system.

Ariana Krynen is a practicing clinical psychologist and Scientia PhD Scholar at the University of New South Wales. Her research interests include enhancing the detection and assessment of childhood emotional and behavioral difficulties, as well as translating research and best-practice literature into effective interventions for at-risk populations.

Peter Reed is a medical research scientist with 29 years experience. He obtained his DPhil at the Faculty of Clinical Medicine, the University of Oxford. He has been the research statistician for Starship children’s hospital, Auckland, New Zealand, for 13 years and provides research and statistical consultancy services to the wider medical community.

Julia Ioane is a senior lecturer in psychology and registered clinical psychologist. Her area of practice is working primarily with youth in care and protection and youth offending. Her current research interest is looking at the mental health and well-being of this population group.

References

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Washington, DC: Author.

Andrews

D. A.

Bonta

(2016). The psychology of criminal conduct (6th ed.). Newark, NJ: Matthew Bender.

Baglivio

M. T.

Wolff

K. T.

Piquero

A. R.

Epps

(2015). The relationship between Adverse Childhood Experiences (ACE) and juvenile offending trajectories in a juvenile offender sample. Journal of Criminal Justice, 43, 229-241. doi:10.1016/j.jcrimjus.2015.04.012

Basto-Pereira

Miranda

Ribeiro

Maia

(2016). Growing up with adversity: From juvenile justice involvement to criminal persistence and psychosocial problems in young adulthood. Child Abuse & Neglect, 62, 63-75. doi:10.1016/j.chiabu.2016.10.011

Becker

K. D.

Stuewig

Herrera

V. M.

McCloskey

L. A.

(2004). A study of firesetting and animal cruelty in children: Family influences and adolescent outcomes. Journal of the American Academy of Child & Adolescent Psychiatry, 43, 905-912. doi:10.1097/01.chi.0000128786.70992.9b

Bowling

C. H.

Merrick

Omar

H. A.

(2014). Self-reported juvenile firesetting: Results from two national survey datasets. Frontiers in Public Health, 1, Article 60. doi:10.3389/fpubh.2013.00060

Chen

Y. H.

Arria

A. M.

Anthony

J. C.

(2003). Firesetting in adolescence and being aggressive, shy, and rejected by peers: New epidemiologic evidence from a national sample survey. Journal of the American Academy of Psychiatry and the Law, 31, 44-52. Retrieved from http://www.aapl.org/

Cicchetti

Bronen

Spencer

Haut

Berg

Oliver

Tyrer

(2006). Rating scales, scales of measurement, issues of reliability: Resolving some critical issues for clinicians and researchers. The Journal of Nervous and Mental Disease, 194, 557-564. doi:10.1097/01.nmd.0000230392.83607.c5

Dadds

M. R.

Fraser

J. A.

(2006). Fire interest, fire setting and psychopathology in Australian children: A normative study. Australian and New Zealand Journal of Psychiatry, 40, 581-586. doi:10.1080/j.1440-1614.2006.01842.x

10.

Dargis

Newman

Koenigs

(2016). Clarifying the link between childhood abuse history and psychopathic traits in adult criminal offenders. Personality Disorders: Theory, Research, and Treatment, 7, 221-228. doi:10.1037/per0000147

11.

Del Bove

Caprara

G. V.

Pastorelli

Paciello

. (2008). Juvenile firesetting in Italy: Relationship to aggression, psychopathology, personality, self-efficacy, and school functioning. European Child & Adolescent Psychiatry, 17, 235-244. doi:10.1007/s00787-007-0664-6

12.

Del Bove

Mackay

. (2011). An empirically derived classification system for juvenile firesetters. Criminal Justice and Behavior, 38, 796-817. doi:10.1177/0093854811406224

13.

Dolan

McEwan

T. E.

Doley

Fritzon

(2011). Risk factors and risk assessment in juvenile fire-setting. Psychiatry, Psychology and Law, 18, 378-394. doi:10.1080/13218719.2011.559154

14.

Esbensen

F. A.

Peterson

Taylor

Freng

(2010). Youth violence: Sex and race differences in offending, victimization, and gang membership. Philadelphia, PA: Temple University Press.

15.

Estévez

Emler

N. P.

(2011). Assessing the links among adolescent and youth offending, antisocial behaviour, victimization, drug use, and gender. International Journal of Clinical and Health Psychology, 11, 269-289.

16.

Evans

J. D.

(1996). Straightforward statistics for the behavioral sciences. Pacific Grove, CA: Brooks/Cole.

17.

Farrington

D. P.

(1995). The twelfth Jack Tizard memorial lecture. The development of offending and antisocial behaviour from childhood: Key findings from the Cambridge Study in Delinquent Development. The Journal of Child Psychology and Psychiatry, 36, 929-964. doi:10.1111/j.1469-7610.1995.tb01342.x

18.

Farrington

D. P.

(2005). Childhood origins of antisocial behavior. Clinical Psychology & Psychotherapy, 12, 177-190. doi:10.1002/cpp.448

19.

Farrington

D. P.

Loeber

Ttofi

M. M.

(2012). Risk and protective factors for offending. In Welsh

B. C.

Farrington

D. P.

(Eds.), The Oxford handbook of crime prevention (pp. 46-69). Oxford, UK: Oxford University Press.

20.

Fire and Emergency New Zealand. (2017a). National incidents reports. Wellington, New Zealand: Author.

21.

Fire and Emergency New Zealand. (2017b). Fire Awareness and Intervention Programme report. Wellington, New Zealand: Author.

22.

Ford

J. D.

(2002). Traumatic victimization in childhood and persistent problems with oppositional-defiance. Journal of Aggression, Maltreatment & Trauma, 6, 25-58. doi:10.1300/j146v06n01_03

23.

Forehand

Wierson

Frame

C. L.

Kemptom

Armistead

(1991). Juvenile firesetting: A unique syndrome or an advanced level of antisocial behavior? Behaviour Research and Therapy, 29, 125-128. doi:10.1016/0005-7967(91)90040-a

24.

Fox

B. H.

Perez

Cass

Baglivio

M. T.

Epps

(2015). Trauma changes everything: Examining the relationship between adverse childhood experiences and serious, violent and chronic juvenile offenders. Child Abuse & Neglect, 46, 163-173. doi:10.1016/j.chiabu.2015.01.011

25.

Grieger

Hosser

(2014). Which risk factors are really predictive? An analysis of Andrews and Bonta’s “Central Eight” risk factors for recidivism in German youth correctional facility inmates. Criminal Justice and Behavior, 41, 613-634. doi:10.1177/0093854813511432

26.

Hoerold

Tranah

(2014). Correlates of adolescent fire setting: Examining the role of fire interest, fire-related attentional bias, impulsivity, empathy and callous–unemotional traits. The Journal of Forensic Psychiatry & Psychology, 25, 411-431. doi:10.1080/14789949.2014.925137

27.

Hunt

T. K.

Slack

K. S.

Berger

L. M.

(2017). Adverse childhood experiences and behavioral problems in middle childhood. Child Abuse & Neglect, 67, 391-402. doi:10.1016/j.chiabu.2016.11.005

28.

Ioane

Lambie

Percival

(2016). A comparison of Pacific, Maori, and European violent youth offenders in New Zealand. International Journal of Offender Therapy and Comparative Criminology, 60, 657-674. doi:10.1177/0306624x14560725

29.

Kolko

D. J.

Day

B. T.

Bridge

J. A.

Kazdin

A. E.

(2001). Two-year prediction of children’s firesetting in clinically referred and nonreferred samples. The Journal of Child Psychology and Psychiatry and Allied Disciplines, 42, 371-380. doi:10.1111/1469-7610.00730

30.

Kolko

D. J.

Kazdin

A. E.

(1991). Aggression and psychopathology in matchplaying and firesetting children: A replication and extension. Journal of Clinical Child Psychology, 20, 191-201. doi:10.1207/s15374424jccp2002_11

31.

Lambie

Ioane

Randell

Seymour

(2013). Offending behaviours of child and adolescent firesetters over a 10-year follow-up. The Journal of Child Psychology and Psychiatry, 54, 1295-1307. doi:10.1111/jcpp.12126

32.

Lambie

Krynen

(2017). The utility of the Strengths and Difficulties Questionnaire as a screening measure among children and adolescents who light fires. The Journal of Forensic Psychiatry & Psychology, 28, 313-330. doi:10.1080/14789949.2016.1275747

33.

Lambie

Randell

(2011). Creating a firestorm: A review of children who deliberately light fires. Clinical Psychology Review, 31, 307-327. doi:10.1016/j.cpr.2010.12.010

34.

Lipsey

M. W.

Howell

J. C.

Kelly

M. R.

Chapman

Carver

(2010). Improving the effectiveness of Juvenile Justice Programs: A new perspective on evidence-based practice. Washington, DC: Center for Juvenile Justice Reform.

35.

Loeber

Farrington

D. P.

(2000). Young children who commit crime: Epidemiology, developmental origins, risk factors, early interventions, and policy implications. Development and Psychopathology, 12, 737-762. doi:10.1017/s0954579400004107

36.

Loeber

Farrington

D. P.

(Eds.) (2001). Child delinquents: Development, intervention, and service needs. Thousand Oaks, CA: SAGE.

37.

Lyons

J. S.

McClelland

Jordan

(2010). Fire setting behavior in a child welfare system: Prevalence, characteristics and co-occurring needs. Journal of Child and Family Studies, 19, 720-727. doi:10.1007/s10826-010-9361-z

38.

MacKay

Paglia-Boak

Henderson

Marton

Adlaf

(2009). Epidemiology of firesetting in adolescents: Mental health and substance use correlates. The Journal of Child Psychology and Psychiatry, 50, 1282-1290. doi:10.1111/j.1469-7610.2009.02103.x

39.

Martin

Bergen

H. A.

Richardson

A. S.

Roeger

Allison

(2004). Correlates of firesetting in a community sample of young adolescents. Australian and New Zealand Journal of Psychiatry, 38, 148-154. doi:10.1111/j.1440-1614.2004.01318.x

40.

Ministry of Justice. (2018). Youth Justice Indicators summary report April 2018. Wellington, New Zealand: Author.

41.

Moffitt

T. E.

(2018). Male antisocial behaviour in adolescence and beyond. Nature Human Behaviour, 2, 177-186. doi:10.1038/s41562-018-0309-4

42.

Moffitt

T. E.

Caspi

Harrington

Milne

B. J.

(2002). Males on the life-course-persistent and adolescence-limited antisocial pathways: Follow-up at age 26 years. Development and Psychopathology, 14, 179-207. doi:10.1017/s0954579402001104

43.

Monahan

Steinberg

Piquero

A. R.

(2015). Juvenile justice policy and practice: A developmental perspective. Crime and Justice, 44, 577-619. doi:10.1086/681553

44.

Mulder

Brand

Bullens

van Marle

(2010). Risk factors for overall recidivism and severity of recidivism in serious juvenile offenders. International Journal of Offender Therapy and Comparative Criminology, 55, 118-135. doi:10.1177/0306624x09356683

45.

Murray

Farrington

D. P.

(2010). Risk factors for conduct disorder and delinquency: Key findings from longitudinal studies. Canadian Journal of Psychiatry, 55, 633-642. doi:10.1177/070674371005501003

46.

Odgers

C. L.

Moffitt

T. E.

Broadbent

J. M.

Dickson

Hancox

R. J.

Harrington

. . . Caspi

(2008). Female and male antisocial trajectories: From childhood origins to adult outcomes. Development and Psychopathology, 20, 673-716. doi:10.1017/s0954579408000333

47.

Phillips

H. K.

Gray

N. S.

MacCulloch

S. I.

Taylor

Moore

S. C.

Huckle

MacCulloch

M. J.

(2005). Risk assessment in offenders with mental disorders: Relative efficacy of personal demographic, criminal history, and clinical variables. Journal of Interpersonal Violence, 20, 833-847. doi:10.1177/0886260504272898

48.

Salmond

Crampton

Atkinson

(2007). NZDep2006 index of deprivation. Wellington, New Zealand: Department of Public Health, University of Otago, Wellington.

49.

Singh

J. P.

Desmarais

S. L.

Sellers

B. G.

Hylton

Tirotti

Van Dorn

R. A.

(2014). From risk assessment to risk management: Matching interventions to adolescent offenders’ strengths and vulnerabilities. Children and Youth Services Review, 47(1), 1-9. doi:10.1016/j.childyouth.2013.09.015

50.

Smith

C. A.

Ireland

T. O.

Thornberry

T. P.

Elwyn

(2008). Childhood maltreatment and antisocial behavior: Comparison of self-reported and substantiated maltreatment. American Journal of Orthopsychiatry, 78, 173-186. doi:10.1037/0002-9432.78.2.173

51.

Sousa

Herrenkohl

T. I.

Moylan

C. A.

Tajima

E. A.

Klika

J. B.

Herrenkohl

R. C.

Russo

M. J.

(2011). Longitudinal study on the effects of child abuse and children’s exposure to domestic violence, parent-child attachments, and antisocial behavior in adolescence. Journal of Interpersonal Violence, 26, 111-136. doi:10.1177/0886260510362883

52.

Stickle

T. R.

Blechman

E. A.

(2002). Aggression and fire: Antisocial behavior in firesetting and nonfiresetting juvenile offenders. Journal of Psychopathology and Behavioral Assessment, 24, 177-193. doi:10.1023/A:1016062701325

53.

Tape

T. G.

(2006). Interpreting diagnostic tests. Omaha, NE: University of Nebraska Medical Center [Internet book]. Available from http://gim.unmc.edu/dxtests/

54.

Trulson

C. R.

Marquart

J. W.

Mullings

J. L.

Caeti

T. J.

(2005). In between adolescence and adulthood: Recidivism outcomes of a cohort of state delinquents. Youth Violence and Juvenile Justice, 3, 355-387. doi:10.1177/1541204005278802

55.

Walsh

D. P.

Lambie

(2013). “If he had 40 cents he’d buy matches instead of lollies”: Motivational factors in a sample of New Zealand adolescent firesetters. International Journal of Offender Therapy and Comparative Criminology, 57, 71-91. doi:10.1177/0306624x11422224

56.

Watt

B. D.

Geritz

Hasan

Harden

Doley

(2015). Prevalence and correlates of firesetting behaviours among offending and non-offending youth. Legal and Criminological Psychology, 20, 19-36. doi:10.1111/lcrp.12062

57.

Wilpert

van Horn

Eisenberg

(2017). Arsonists and violent offenders compared: Two peas in a pod? International Journal of Offender Therapy and Comparative Criminology, 61, 1354-1368. doi:10.1177/0306624X15619165

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB