Research on harm stemming from children’s exposure to multiple forms of victimization, conceptualized as polyvictimization, is relatively new in the research literature [1
]. Not only does polyvictimization appear to be more harmful to children than single victimization, but low-frequency victimization across multiple domains (e.g., bullying at school and abuse at home) appears to produce more distress than high-frequency victimization in a single domain [1
]. Child polyvictimization is associated with more psychopathology, depression, delinquency, low self-esteem, self-blame, suicidal phenomena, health risk behaviors, and anxiety among children, and higher unemployment, substance abuse and mental illness among parents [1
]. Finkelhor et al. argued that there are four pathways to child polyvictimization: a) dangerous families, b) problem-beset families, c) dangerous communities, and d) children with pre-existing emotional problems that increase risky behaviour [1
]. Chan extended the concept to incorporate co-occurrence of multiple forms of violence at the family level, labelling this family polyvictimization [2
]. Family polyvictimization was indicated by child victimization, intimate partner violence (IPV), and elder abuse, and was associated with parental addiction and poor health in a Chinese household sample [2
]. Despite Chan’s initiative, polyvictimization remains understudied in an East Asian context. Moreover, research on the dangerous families pathway [1
] generally focuses on individual characteristics of family members rather than family systemic characteristics, little work has been done to link the concept of polyvictimization with its measurement, and the problem of how to incorporate high frequency single-victimization with polyvictimization has not been resolved. Specifically, despite a large literature on the co-occurrence of IPV and child maltreatment [8
], little literature has examined the effects of types of IPV on child polyvictimization. Similarly, no existing measure integrates polyvictimization incidence and frequency into a common scale of polyvictimization severity. This paper examines the relationship between polyvictimization and a new typology of violent family organization in a probability proportional to size random cluster sample of rural South Korean children. We hypothesize anarchic and totalitarian family types are at higher risk for polyvictimization. The paper argues for and implements a measure of polyvictimization that allows single-victimization frequency and polyvictimization to be combined in the same scale that captures the severity of polyvictimization. We argue this measure is closer to the conceptual rationale for the study of polyvictimization.
Although Chan [2
] has stressed the importance of considering emergent qualities of the family as a unit in considering polyvictimization, and a 30-year research tradition documents conceptual and empirical links between intimate partner violence and child maltreatment [8
], only a handful of studies searchable via google scholar explicitly examine intimate partner violence and child polyvictimization. Chan’s research examined exposure to parental IPV as one component of polyvictimization and found that polyvictimized children had lower self-esteem, higher rates of aggression, PTSD, addiction, and lower quality of life [2
]. Pereda and Gallardo-Pujol similarly included IPV exposure as polyvictimization and found that it predicted re-victimization in adulthood [13
]. Using a similar concept of polyvictimization, other research found that the relationship between child polyvictimization and PTSD were mediated by child attributions [14
] and that polyvictimization was correlated with later disordered eating [15
1.1. Disorder vs. Deviant Order: Totalitarian and Anarchic Family Types as Risk Factors for Polyvictimization
Although the foregoing literature on child polyvictimization has considered IPV exposure as a form of polyvictimization [5
], it has insufficiently examined parental IPV in an etiological role with respect to polyvictimization via the dangerous families pathway [1
], neither has it conceptualized IPV as having a profound effect on emergent family processes. Emery [16
] argues that the introduction of physical violence into an intimate relationship shifts the foundation on which power rests. According to this argument, power dynamics between intimate partners in never-violent relationships rest on walk-away costs. However, when an act of physical violence occurs between partners the base of power may shift from walk-away costs to force. In that case power may rapidly shift to the partner who can command the most physical force [16
], permanently changing family power dynamics. In Emery’s typology, IPV exists on dual continua of order and power, ranging from anarchic type (low order, chaotic, no consistent rules or legitimate power) at one extreme and totalitarian type (highly ordered, asymmetric power) at the other [16
]. Totalitarian families are characterized by power asymmetry and elaborate systems of rules extending control to “mundane areas of everyday life …not normally thought of as norm- or rule-governed” [17
]. Emery argues that the anarchic type is chaotic, unpredictable, and fits more with the social disorganization conceptualization of IPV common in criminology, while the totalitarian type reflects the traditional feminist conceptualization of IPV. IPV at both extremes, however, is likely to create a more dangerous family environment for children [16
]. IPV at the extremes of these continua can put children at higher risk not only because the IPV may be more frequent and qualitatively severe [16
], but also because maltreatment may be more likely to be legitimized as punishment for rule-breaking in the totalitarian type while family rules (norms) aimed at protecting children may be absent in the anarchic type. This suggests families with any history of IPV and characterized by anarchy or totalitarian style control may put children at particular risk for polyvictimization.
Although some research [18
] has begun to use the concept of the totalitarian type, no research has used a measurement specifically aimed at capturing anarchic or totalitarian type families, and no research has examined these types as risk factors for polyvictimization. Indeed, research on types of IPV in an etiological role for child polyvictimization broadly is lacking. Although arguably the most common typology of IPV is Johnson’s [20
] intimate terrorism typology, a google scholar search of “intimate terrorism” and “polyvictimization” returns only 47 results, none of which feature intimate terrorism in a causal role for child polyvictimization. The intimate terrorism typology is defined on the basis of control motive [20
], with the result that whether control attempts succeed or not (achieved control) is ignored. The intimate terrorism typology has been critiqued on this basis as ignoring power, for which reason this paper focuses on the anarchic/totalitarian IPV typology in etiological relation to child polyvictimization.
1.2. Rationale for Polyvictimization and Its Measurement: Capturing Severity
The burgeoning literature on polyvictimization [1
] rests on a largely implicit rationale: polyvictimization must be distinct in both etiology and impact from high frequency victimization of a single type. If this were not the case, multiple experiences of victimization of different forms could lumped together additively in a single measure without reference to type. For example, experience of acts of physical abuse, acts of neglect, and witnessing acts of IPV could simply be summed. The rationale for polyvictimization as a unique subject is empirically supported by findings that even low-frequency polyvictimization appears to have a more severe impact than high frequency single-victimization [1
]. The problem for the polyvictimization field to date [1
] is that researchers are forced to choose between examining the impact of frequency for single-victimization or sums of dichotomous indicators for various types of single-victimization that add up to polyvictimization. Continuous measures generally have more statistical sensitivity than dichotomous measures, and Emery et al. [21
] have long argued for and implemented continuous measures of violent victimization weighted by the log-odds of injury. Although the continuous measure creates a right-skewed distribution, this problem can now easily be handled via monotonic transformation, robust standard errors, or both [23
]. Moreover, a continuous measure can capture the severity, rather than simply the fact, of polyvictimization.
We argue that the empirical logic of the rationale for the study of polyvictimization is one of interaction effects. That low frequency victimization in the context of a second form of victimization has a larger impact on child well-being than high frequency single-victimization [1
] suggests that the effect of one form of victimization on well-being differs depending on whether a second form is present or absent. Interaction effects capture just such an effect, as when the effect of witnessing IPV on externalizing behavior problems depends on the child’s age [24
]. The standard approach to handle this problem in linear models is to model multiplicative relationships between variables, (e.g., age X witnessing IPV) rather than only additive relationships [23
]. In this paper, we present a continuous measure of polyvictimization which is additive when a single form of victimization is present but multiplicative when multiple forms are present. This extends the scale to higher levels when more than one kind of victimization is present. For example, if our measure of physical abuse severity is a 5 and only physical abuse is present, then the continuous polyvictimization scale will also be a 5. However, if the child has a physical abuse score of 5 and also a neglect score of 2, then the polyvictimization score will be 2 × 5 = 10. Our models compare the continuous severity measure with a more standard categorical measure of polyvictimization.
1.3. Current Study
Based on the IPV literature, two hypotheses were made:
Polyvictimization and polyvictimization severity will be positively associated with totalitarian type IPV.
Polyvictimization and polyvictimization severity will be positively associated with anarchic type IPV.
The aim of the current study was to extend the literature on polyvictimization by examining whether emergent family types (anarchic and totalitarian) from the IPV literature [16
] are related to polyvictimization and comparing a new severity measure of polyvictimization with the standard approach. The current study also contributes to the IPV literature. The full models control for a measure typically used to capture a type of IPV known as intimate terrorism [20
]. Intimate terrorism is commonly studied in the IPV literature, and is defined as violence in the context of systematic attempts to control the victim [20
]. The intimate terrorism concept has been critiqued as inadvertently ignoring power [16
]. If the anarchic and totalitarian family types predict polyvictimization when intimate terrorism is held constant this would suggest some support for the added value of the disorder to deviant order continuum theory (anarchic/totalitarian typology) [16
]. Models also control for age and sex of the focal child, sex of the parent, and household income.