3. Sex-Role Theory
A subtype of social learning, sex-role theory suggests that early sex-role socialization teaches boys to be the dominant partner, major wage earner, and head of the household, while women are socialized to accept male dominant relationships and taught to meet the needs of others through their main roles as wives and mothers. These roles may leave males and females vulnerable to becoming offenders and victims of marital violence. Most empirical studies have failed to validate a sex-role interpretation of marital violence (Hotaling and Sugarman 1986; Mihalic and Elliott 1997). Walker (1984), contrary to her original supposition, found no evidence in a clinical sample that battered women had traditional sex-role attitudes. Instead, they perceived themselves as more liberal; however, they perceived their mates as traditional. Her research suggested that the discordance in perceived sex roles might lead to conflict within the marriage and hence to marital violence. This hypothesis was tested by Coleman and Straus (1986), who found that equalitarian couples had the lowest rates of conflict and violence, while male- or female-dominant couples had the highest rates. Consensus about the legitimacy of the power structure reduced the rate of conflict and violence in male- or female-dominated families, but when conflict did occur in these families, it was associated with a much higher risk of violence than that of equalitarian families encountering the same level of conflict. This suggests that disagreement over sexrole orientations may be a bigger factor in marital aggression than the actual orientation held.
4. Is Aggression Generalizable?
A key element of social learning theory concerns its generalizability. Is violence learned in one context generalizable to other contexts? Social learning theory predicts a generalized pattern of learned aggression that may be modeled in both family and nonfamily relationships. Bandura (1971, 1973) proposes that aggressive models transmit general lessons, as well as specific ones, and that observers learn general aggressive strategies that go well beyond the specific modeled examples. The perspective of generalized modeling has much empirical support in both the family violence literature and the delinquency literature (McCloskey and Lichter 2003; Mihalic and Elliott 1997; Thornberry 1994). A review of twenty-three articles on the effects of observing parent aggression provided evidence that children observers are at risk for a variety of externalizing behaviors, including increased aggression at home and school and in the community (Fantuzzo and Lindquist 1989). The effects on child witnesses of domestic violence are not confined only to behavioral development, but also affect emotional development, although meta-analyses suggest that these links may be weak (Kitzman et al. 2003).
Data from the 1975 and 1985 National Family Violence Surveys and a 1972 university student survey demonstrated that children assaulted by parents were more violent toward brothers, sisters, parents, and persons outside the family. They were also more likely to be involved in property crimes and with the police (Hotaling, Straus, and Lincoln 1990). This study also found that adult offenders and victims of family assault had higher rates of violent and nonviolent crime outside the family. The relationship existed even with controls for socioeconomic class, gender, and severity of violence, although the relationship was, in general, stronger for males and blue-collar families. These authors suggest that it is not just the direct experience of being assaulted that leads to violence, but the experience of living in a multi-assaultive family (i.e., the highest rates of outside family violence were reported by those respondents who were from families where they witnessed violence between their parents and were directly assaulted by a parent). These findings suggest that there are common links in all types of violence.
A prospective sample of 299 children, ages six to twelve, were interviewed with their mothers in 1991 to examine gender differences in adolescent delinquency five years later against a backdrop of witnessing marital violence and being a victim of child abuse (Herrera and McCloskey 2001). This study indicated that 31 percent of children who experienced abuse and 33 percent who witnessed marital violence, compared with 18 percent of those children without abuse in their childhoods, were referred to juvenile court at least once. Additionally, 17 percent of abused children and 17 percent of those who witnessed violence were referred for a violent offense, compared with 5 percent without a family background of violence. Being a victim simultaneously of both forms of abuse failed to predict delinquency above and beyond that of either of the other two categories. There was an interaction between sex and child abuse, with girls at higher risk of arrest for violence if they had a prior history of physical child abuse.
Evidence that exposure to violence in childhood is related to antisocial behavior outside the home also comes from longitudinal delinquency surveys. The Rochester Youth Development Study, which tracks 1,000 seventh- and eighth-grade students in the Rochester public school system, found that a history of substantiated cases of physical or sexual abuse or neglect prior to age twelve increased the chances of youth violence by 24 percent. Adolescents growing up in homes with partner violence or a family climate of hostility also exhibited higher rates of self-reported violence. Exposure to multiple forms of family violence doubled the risk of self-reported youth violence. These analyses controlled for gender, race/ethnicity, family structure, and social class (Thornberry 1994). In the National Youth Survey, females who witnessed marital violence had higher rates of minor adolescent violence and felony assault (Mihalic and Elliott 1997).
Another study followed 1,575 cases from childhood through young adulthood, comparing 908 substantiated cases of childhood abuse or neglect with a group of 667 matched children not officially recorded as abused or neglected. Being abused as a child increased the likelihood of arrest as a juvenile by 59 percent, as an adult by 28 percent, and for a violent crime by 30 percent (Widom and Maxfield 2001).
A question that arises is whether there is a threshold in one’s early experience of violence that must be surpassed before the aggressive lessons become salient. To answer this question, a national representative sample was used to compare the effects of minimal, moderate, and frequent spanking on children’s physical aggression against siblings and parents. A linear relationship for preschoolers, preadolescents, and adolescents was found for all levels of spanking. This supports the idea that ‘‘violence begets violence’’ and that any punishment that uses violent means may be harmful. It is important to note that the degree to which the parent reasoned with the child moderated the effect in several models (Larzelere 1986).
The cycle of violence theory assumes that if physically aggressive parents end up with aggressive children, it is because the child has learned a patterned response to violence. An alternative explanation is that the child has a predisposition toward aggressive behavior and that the punitive parental behavior is a response to the child. Parents often cite child misbehaviors as leading to greater use of severe corporal punishment. Thus, corporal punishment may be a response to aggressive child behavior, rather than its cause. A test of the social learning model against the temperament model provided support for the social learning model, which suggests that temperament does not adequately explain the process by which corporal punishment is passed on intergenerationally (Muller, Hunter, and Stollak 1995).
While there appears to be support for a link between family violence and youth violence, others have argued that the claim that child maltreatment is the leading cause of delinquency relies upon methodologically flawed studies and that the few rigorous studies are inconclusive or offer only a weak connection, which often disappears when other variables are controlled in the analyses (Schwartz, Rendon, and Hsieh 1994).