Secondary analysis of criminal justice data

The analysis of existing criminal justice data has been an early and relatively dominant method for exploring ‘parent abuse’, since it can produce large quantitative datasets to enable the identification of case characteristics of both parent and child/young person. Such data have been collected at different stages of the criminal justice process, and are dominated by work in the United States. For example, at the policing stage, Evans and Warren-Sohlberg (1988) analysed seventy-three police case reports concerning parents’ formal complaints about ‘parent abuse’, while Walsh and Krienert (Reference Walsh and Krienert2007) examined nearly 18,000 cases of CPV (‘child-to-parent violence’) across twenty-three US states using the incident and arrest recording system (known as NIBRS). At the judicial stage, Cochran et al. (Reference Cochran, Brown, Adams and Doherty1994), Kethineni (Reference Kethineni2004) and Gebo (Reference Gebo2007) all examined US county court records of juveniles convicted of violence against family members, including parents.

One strength of this approach is the availability of comparative groups to enable the identification of mediating variables: for example, Walsh and Kreinert (Reference Walsh and Krienert2009) subsequently compared the CPV group with adolescents who committed parricide, and found distinct differences in terms of age, gender, ethnicity and weapon use. Gebo (Reference Gebo2007) compared her sample with a sample of sixty young people who were committing non-familial violent offences and found that those sentenced in family violence cases were dealt with more leniently by the courts. A similar comparative design by Kennedy et al. (Reference Kennedy, Edmonds, Dann and Burnett2010) found young offenders who were violent towards parents showed greater psychological disturbance and greater association with gang members than those who were not. Outside of the US, Ibabe and Jaureguizar (Reference Ibabe and Jaureguizar2010) examined the case files and judicial proceedings of seventy adolescents in Bilbao, Spain, who had committed violent offences against parents (half of whom had also committed other types of offences) and compared them with a sample of thirty-three adolescents who only committed non-familial offences. Their results suggested that the ‘parent abuse only’ sample constitute a unique group sharing a particular psychosocial profile.

However, it is important to recognise that the group of adolescents who come through the criminal justice system as a result of their abusive behaviour is likely to be a very distinct group, as are the families who report them and the forms of abusive behaviour which are reported. Thus, these data are likely to represent the ‘thin end’ of the wedge; that is, those whose behaviour lies at the severest end of the violence spectrum (with physical abuse more likely to be processed than other forms of abuse) and those with already problematic family relations. Indeed, such methods may tell us little beyond how criminal justice systems process such young people once they have come into contact with them and like other forms of ‘official’ crime data, tend to find males, Black and minority ethnic groups and those from the poorest families over-represented. Furthermore, even within these types of studies, they may not represent equivalent populations, since families with particular characteristics may ‘drop out’ before reaching the next stage (e.g. reaching court) of the justice process. It is therefore problematic to make general conclusions about the characteristics of families involved in ‘parent abuse’ using such data.

Furthermore, it is important to recognise that such approaches only tell us the information which criminal justice agencies deem to be important for their own purposes in the processing of offenders and the reduction of crime. Thus, we learn much about the who and the what of ‘parent abuse’ (or, more accurately, of those young people who are criminalised for parent abuse-related offences) in terms of statistical frequencies, but they tell us little about the how and the why, although they certainly serve as a useful starting point.

Large-scale epidemiological surveys

Large-scale surveys, usually quantitative and sometimes randomised, have also been used to find out about parent abuse. The majority of these have taken place in schools with children and young people, often on the back of larger surveys. For example, Agnew and Huguley's (Reference Agnew and Huguley1989) drew on the US-based National Survey of Youth (1972) to obtain self-report data from 1,395 eleven to eighteen-year olds. While finding no evidence for age or gender ‘risk factors’ from questions concerning violence towards parents, they did find some interesting correlates concerning beliefs, ‘attachment style’ and peer group. Both Peek et al. (Reference Peek, Fischer and Kidwell1985) and Brezina (Reference Brezina1999, Reference Brezina, Litton Fox and Benson2000) reviewed data from the US-based national Youth in Transition (1966–1969) survey of male adolescents, with Peek et al. (Reference Peek, Fischer and Kidwell1985) identifying some surprising risk factors (such as more violence directed towards fathers), while Brezina (Reference Brezina1999) identified correlations to suggest that adolescents may be reacting in response to parental aggression towards them. From Canada, Pagani et al. (Reference Pagani, Tremblay, Nagin, Zoccolillo, Vitaro and McDuff2004, Reference Pagani, Tremblay, Nagin, Zoccolillo, Vitaro and McDuff2009) report on data from the Quebec Longitudinal Study of Kindergarten Children (1986–1996), which followed up 2,908 children from the ages of six to fifteen using questionnaires and interviews with children and their parents, finding a positive correlation between harsh parental punishment and adolescent aggression towards mothers and fathers. More recently, the national Youth at Risk Survey (2000) questioned 2,004 US youths, living at home, on a range of ‘risky behaviours’, and Elliot et al. (Reference Elliott, Cunningham, Colangelo and Gelles2011) found a significant negative correlation between self-reports of ‘hitting parents’ and measures of adolescents feeling that they ‘mattered’ to their families.

Specific crime surveys include Pagelow et al.'s (Reference Pagelow, Ohlin and Tonry1989) survey of 473 US university students, which found 13 per cent admitted violence against family members (mostly parents), and in the UK, Browne and Hamilton's (Reference Browne and Hamilton1998) survey of 232 psychology undergraduates, which found 14.5 per cent admitted to having used violent tactics against a parent. Victim surveys include Edenborough et al.'s (Reference Edenborough, Jackson, Mannix and Wilkes2008) Child to Mother Violence Scale (CMVS),Footnote ² which was posted to 6,000 mothers living in areas of high violence in Australia, producing 1,024 responses. While the quantitative data from this study are yet to be published, the authors produced a small-scale qualitative analysis of the 185 written responses to their final qualitative question regarding their experiences, identifying common themes concerning mothers’ growing fear of violence and their difficulties in accessing support.

However, just as the criminal justice data tells us information that it is pertinent to the criminal justice system, so large-scale surveys correlate parent abuse only with those psychosocial characteristics that are reducible to quantified ‘variables’ (e.g. ‘quality of attachment’, ‘parenting style’, etc.). Furthermore, with the exception of Brezina (Reference Brezina1999) and Pagani et al. (Reference Pagani, Tremblay, Nagin, Zoccolillo, Vitaro and McDuff2004), who compared waves of data to examine longitudinal effects, they all involve cross-sectional analysis. Thus, while such surveys arguably provide us with a static ‘snapshot’ of parent abuse, they tell us little about process or context.

Interview data from intervention groups

A third method for producing knowledge about parent abuse involves collecting data from service users who are already participating in some kind of family intervention programme. Most of this research has been small-scale and qualitative, utilising interviews and/or observations with parents, although some include young people and/or practitioners. Such studies have primarily focused on the experiences and effects of parent abuse, although antecedents have also been identified to enable inductive theory-building. For example, Cottrell and Monk (Reference Cottrell and Monk2004) drew on three studies which included interview and focus group data from parents, young people and counsellors to identify factors (described within ‘ecological levels’) which may be involved in the production of parent abuse. Such methodological approaches have lent themselves to innovative techniques: Eckstein (Reference Eckstein2004) asked twenty parents to narrate one experienced episode of verbal, physical and emotional abuse to enable an exploration of the communication patterns involved in child–parent abusive interactions; Howard and Rottem (Reference Howard and Rottem2008) used genograms to map out the family relationship context with the ten mothers they interviewed.

Often such studies are produced to evaluate the outcome of a programme intervention. For example, in Australia, Paterson et al. (Reference Paterson, Luntz, Perlesz and Cotton2002) used a series of self-report questionnaires and in-depth interviews to measure episodes of violence and change with eighteen mothers who were participating in a group counselling intervention programme. Qualitative experiences of living with abusive adolescents (e.g. ‘walking on eggshells’) are discussed alongside questionnaire data to demonstrate the programmes’ positive impact in both reducing violence and the associated anxiety. Smaller evaluation studies have been conducted by Doran (Reference Doran2007) in Canada and Daly and Nancarrow (Reference Daly, Nancarrow and Ptacek2009) in Australia, which examined restorative justice interventions, and Munday (Reference Munday2009) in the UK, which evaluated a local partnership intervention programme.

In the United States, the increasing use of the Step-Up programme has produced a number of recent evaluations. For example, Routt and Anderson (Reference Routt and Anderson2011) interviewed and observed 268 families in Seattle, Washington who were participating in the Step-Up programme, and combined this with court records, observations and interviews with programme staff. This study is unusual in its use of mixed methods to not simply triangulate the data but to enable the qualitative experiences of parent abuse to be theorised by drawing on the individual, family and systemic antecedents identified using the statistical analyses of both their programme sample and the county records.

However, as with the use of criminal justice data discussed above, such approaches also draw from a very distinct population which produces its own particular problems. In most cases, the families are self-referring, and in any case are only likely to have come forward in areas where such support programmes are available. Even in cases where families are referred by other agencies (such as criminal justice agents), previous research suggests that it is particular kinds of families who are seen as ‘in need’ and are selected for family intervention programmes; that is, lone parents, mothers and those experiencing financial difficulties (Holt, Reference Holt2010).

Of course, the limitations of ‘self-referral’ also apply to participation in research interviews, making the issue doubly problematic. Although interviews traditionally yield a higher response rate compared to questionnaires (70–80 per cent compared to 50 per cent for the most sophisticated questionnaire), Gelles (Reference Gelles and Patterson1990) comments on how these response rates decrease when the topic of discussion involves family violence. Given this difficulty, some studies have elicited qualitative data unintentionally: Jackson (Reference Jackson2003), Stewart et al. (Reference Stewart, Burns and Leonard2007), Holt (Reference Holt2009) and Hunter et al. (Reference Hunter, Nixon and Parr2010) all obtained qualitative data concerning parents’ experiences of ‘parent abuse’ when interviewing parents about a different topic. An alternative approach has been taken by Holt (Reference Holt2011), whose analysis of anonymous postings on online parenting support message boards is particularly illuminating in its consideration of how the emotional fallout of parent abuse is played out within an online arena. While small in number and scale, these studies are important as they access that majority of parents who do not seek help and whose experiences are invisible to institutional number-crunching: that is, the ‘thick end’ of the wedge.

Clinical approaches: case studies and typological analyses

Harbin and Madden's Reference Harbin and Madden1979 paper is often considered the first research publication to have specifically focused on ‘parent abuse’. It is relatively unusual in drawing on the authors’ own case notes and clinical observations of families in treatmentFootnote ³ to highlight how particular analytic concepts (such as ‘parentification’ and ‘denial’) might help us understand how parent abuse can emerge. However, since then, very few published articles have taken this case study approach: notable exceptions are Micucci (Reference Micucci1995) who used a single case study from an adolescent in-patient to illustrate family characteristics and therapeutic principles when working with ‘parent abuse’, Sheehan's (Reference Sheehan1997) analysis of her own observations and case notes from work with sixty families referred for therapeutic intervention and Gallagher's (Reference Gallagher2004a, Reference Gallagher2004b) influential publications examining his own therapeutic work with seventy families.

Instead, clinical approaches have taken a much more typological approach. For example, Nock and Kazdin (Reference Nock and Kazdin2002) examined the profile of 606 adolescent outpatients, 12 per cent of who engaged in parent-directed aggression, and their use of a control group enabled them to distinguish individual and family characteristics (e.g. ‘lower frustration tolerance’, ‘parental stress’) specific to this population. Similar typologies have been developed from Laurent and Derry's (Reference Laurent and Derry1999) analysis of 645 children hospitalised in a child and adolescent psychiatry department in France (finding 3.4 per cent parent ‘batterers’) and Boxer et al.'s (Reference Boxer, Gullan and Mahoney2009) sample of 232 cases taken from a database of families referred for the clinical treatment of emotional and behavioral problems in adolescence. At the furthest end of the clinical spectrum, there are a handful of studies which have focused on the relationship between aggression towards parents and particular clinical ‘syndromes’ such as Tourettes Syndrome (e.g. DeLange and Olivier, Reference DeLange and Olivier2004) and ADHD (e.g. Ghanizadeh and Jafari, Reference Ghanizadeh and Jafari2010).

A key trend within this approach is the development of ‘screening instruments’ to enable identification of ‘parent abuse’, such as Boxer et al.'s (Reference Boxer, Gullan and Mahoney2009) adaptation of the Conflict Tactics Scales (Straus, Reference Straus1979) to operationalise YPA (youth-to-parent aggression) and Ghanizadeh and Jafari's (Reference Ghanizadeh and Jafari2010: 76) Abused Parent Questionnaire. While such moves are positive in that they force researchers to consider the minutiae of definitions, as with the survey data there is a question mark over the validity of such tools, including the decontextualised nature of the measures and the well-recognised problems with relying on participant recall of past incidents.

1. Lengthening the scope of research: relating proximal factors to distal factors

The concept of a ‘funnel of violence’ (Wolfe et al., Reference Wolfe, Wekerle and Scott1997) is frequently invoked to enable the consideration of proximal and distal factors which shape violent outcomes. Distal factors include cultural and political forces which shape the acceptability of violence in terms of respect given to elders or the extent to which violence is tolerated in media representations of the young, for example. Proximal factors include the family and the individual and usually make themselves known in the form of individualised ‘risk factors’ (e.g. gender, ethnicity etc.). However, as is evident from this brief review, there does appear to be a preponderance of research at the proximal end of the spectrum and this can be a problem when researchers draw on such findings from other national contexts to make inferences about their own – not only because findings do not translate well across differing organisational contexts but also because differing contexts shape experiences of parent abuse in the first place. Similarly, while Finkelhor (Reference Finkelhor, Finkelhor, Gelles, Hotaling and Straus1983) is right to observe that power is central to all forms of family abuse, it is important to recognise that power operates distally as well as proximally – hence Archer's (Reference Archer2006) finding that women's violence against men increases in contexts where women have greater political and economic power. This raises interesting but as yet unexplored questions regarding how emerging children's rights discourses in countries such as the UK might be implicated in parent abuse. Such a wider political and cultural analysis which considers the changing structural and political position of children/young people and their parents is currently missing, yet it has the potential to enhance our understanding of parent abuse specifically, and of family abuse more generally.

2. Widening the scope of research: contextualising parent abuse within family abuse

There is already a wealth of evidence which suggests that parent abuse takes place within a wider backdrop of family violence. In particular, research suggests that witnessing inter-parental violence increases the likelihood of engaging in parent abuse (e.g. Ullman and Strauss, Reference Ullman and Straus2003; McCloskey and Lichter, Reference McCloskey and Lichter2003) and that parent abuse co-occurs with parent-to-child abuse (e.g. Brezina, Reference Brezina1999; Boxer et al., Reference Boxer, Gullan and Mahoney2009). Furthermore, research suggests that young people who are abusive to parents may go on to behave abusively towards dating partners (e.g. LaPorte et al., Reference Laporte, Depeng, Pepler and Chamberland2009). Such findings highlight the need for a contextualising, dynamic approach to researching parent abuse, which is important for at least two reasons. First, it would help to identify trajectories from one form of victimisation to another, and therefore in the development of prevention strategies. Second, research into family violence has consistently shown how polyvictimisation produces worse effects in the experience of any one form of violence, particularly in reducing resiliency in overcoming its effects (Finkelhor et al., Reference Finkelhor, Davis, Lurigio and Herman2007) and in increasing the risk of intergenerational transmission (Heyman and Slep, Reference Heyman and Smith Slep2002). Therefore, while also keeping hold of what makes parent abuse unique (to enable appropriate intervention strategies), it is vital that we recognise what parent abuse shares with other forms of family violence to enable theoretical integration.

3. Transforming the scope of research: towards trans-disciplinarity

While this review has highlighted the range of disciplines which has produced knowledge about ‘parent abuse’ (e.g. psychology, psychiatry, social work, criminology), there is little evidence of dialogue between them. This is a problem because positioning ‘parent abuse’ within any one particular discipline can limit the possibilities for both theoretical development and practical intervention. For example, research which analyses criminal justice data inevitably positions the problem – and therefore the solution – within a criminal justice framework, and it may be unhelpful to ‘criminalise’ young people (and their families) in this way, both theoretically and in practice. However, with a hetereogeneity of definitions, populations and outcome measures, joining up the disciplinary dots remains a challenge. As with ‘intimate partner violence’ (IPV), it is likely that the most effective theoretical and practical responses will be produced by contributions from a range of academic and professional disciplines. However, at present we have an approach to parent abuse which is multidisciplinary, and because each discipline offers its own conceptual and methodological perspective, the outcomes are disparate and incoherent. Inter-disciplinarity would be an improvement, with disciplines – and their associated methodologies – combining to provide more coherent and holistic analytical accounts of parent abuse. The final goal might be the development of a new trans-disciplinary approach which establishes an entirely new way of theorising and researching parent abuse.