Surveys of the literature reveal that there are few differences between the psychological maladjustment associated with each of the five types of maltreatment. The broad range of adjustment problems related to each type makes the search for outcomes specific to each maltreatment type difficult. This is particularly true in the case of sexual abuse, in which the list of sequelae is immense (Higgins & McCabe, 1994). The lack of outcome specificity leads to a major problem: when multiple types of maltreatment are not assessed simultaneously, the association between a specific type of maltreatment and adjustment cannot be accurately described unless all other types of maltreatment have been excluded as a potential explanation for the observed maladjustment.

However, only a few studies have actually conducted a simultaneous examination of the impact of each type of maltreatment (e.g., McGee, Wolfe, Yuen, Wilson, & Carnochan, 1995; Ney et al., 1994). McGee et al. (1995) found that 90% of their sample of maltreated adolescents had experienced more than one type of maltreatment. The results of the studies by McGee et al. (1995) and Ney et al. (1994) suggest that neglect may be the most damaging type of child maltreatment.

Research is conducted as if the different types of abuse or neglect occur in isolation (Ney et al., 1994). Rosenberg (1987) noted that 'there has been a shifting emphasis from one type of maltreatment to another with little energy directed toward integrating findings across maltreatment areas' (p. 166). Due to the concurrent nature of family violence and dysfunction, research needs to evaluate dysfunctional family systems, family violence and maltreatment as a whole, rather than focus on one aspect. It is important to consider all of the types of violence occurring simultaneously in the one family, as violence toward an abused child may take a number of different forms (Stanley & Goddard, 1993). Hamilton and Browne (in press) cite a number of studies as support for their contention that maltreated children are at greater risk of any type of further abuse on neglect, not just revictimization by the same type of maltreatment.

Critics have noted that single types of abuse or neglect do not generally exist in isolation (Finkelhor, 1995; Higgins & McCabe, 1994; McGee et al., 1995; Sanders & Becker-Lausen, 1995). The interrelatedness of all maltreatment of children requires an integrated approach to both research and treatment (Briere & Runtz, 1988; Hamilton & Browne, in press; McGee et al., 1995O'Hagan, 1995). Ecological and family dysfunction models of child maltreatment not only recommend assessing maltreatment but also family characteristics which are related to either the occurrence of maltreatment or the adjustment of children. An ecological framework (e.g., Krishnan & Morrison, 1995) outlines the contribution of both familial and societal systems--and the interaction between the two--to the incidence of child maltreatment. It is unlikely that the experience of multiple forms of abuse will merely have a cumulative effect, as each of the traumas affect similar domains, with almost total overlap in the potential range of symptoms/reactions. The effect on a child of the interaction between two or more types of trauma will most likely have a unique impact.

Hamilton and Browne (in press) suggest a two-dimensional taxonomy of victimisation terms which distinguishes between single victimisation (single incident, single perpetrator), multiple victimisation (single incident, multiple perpetrators), repeat victimisation (multiple incidents, single perpetrator), and re-victimisation (multiple incidents, multiple perpetrators). However, a third dimension also exists: the type of maltreatment to which the incident refers (i.e., sexual abuse, physical abuse, psychological maltreatment, neglect, witnessing family violence). In order to distinguish between children subjected to only one of these maltreatment types and those subjected to more than the type of abuse and neglect, it is suggested that the prefixes single-type or multi-type be added to the nomenclature proposed by Hamilton and Browne (in press). They ask 'What is it about that person... that makes them more likely to be abused for a second time - either by the same or a different perpetrator?' (p. 22). Further to this question, we can also ask: what factors occuring in a child's familial environment distinguish multi-type repeat victimisation and multi-type re-victimisation from single-type single and multiple victimisation?

It is only when each form of maltreatment is assessed simultaneously and the effects of each type on an individual partitioned statistically that conclusions can be drawn about the relationships between maltreatment types and measures of adjustment. However, difficulties emerge when considering how to obtain respondents for a study which would encompass all maltreatment types (see Higgins & McCabe, submitted). For example, child abuse treatment centres (most commonly treating sexual abuse victims) may not routinely screen for other types of maltreatment - so the degree to which other types of abuse or neglect are present go undetected, limiting their usefulness as a source for research subjects. Also, children's problems are often not adequately identified or are under-reported when children's self-reports are used (Higgins & McCabe, submitted). Parent ratings have frequently been used to assess the presence of psychopathology (usually defined in terms of emotional and behavioural problems) in abused children (e.g., Mannarino, Cohen, & Berman, 1994). As well as evaluating the child's functioning, parent reports can include other information on the family, and the child's experience of maltreating behaviours being directed towards them by parents or others (Higgins & McCabe, submitted).

The development of a measure to assess children's experiences of maltreatment, mediating variables and current adjustment was reported by Higgins and McCabe (submitted). The Family and Life Experiences Questionnaire - Parent Version (FLEQ-P) is the only integrated measure (of which we are aware) which assesses parent perceptions of sexual abuse, physical abuse, psychological maltreatment, neglect, and witnessing family violence experienced by children aged 5 to 12 and to examine the interrelationships between the different forms of child maltreatment. The relationship of maltreatment to family background variables and measures of behavioural adjustment was also evaluated. It was hypothesised that there would be a high degree of overlap between each of the five types of child maltreatment being assessed. It was further hypothesised that (1) family background variables would provide significant prediction of both child maltreatment and adjustment scores; (2) family background characteristics would distinguish between single-type and multiple-type victimisations.

Parents (n=50) of children aged 5 to 12 years were recruited from community organisations and human service agencies in metropolitan and regional centres. Eighty-six percent of respondents were female, 12% were male, and the gender of 2% was not known. The mean age of respondents was 40.6 years, although there was considerable variability (sd = 16.0). Forty-six respondents were biological parents (92%), two were adoptive parents (4%), one a step-parent, and one did not indicate the relationship to the child.

Measures

Family and Life Experiences Questionnaire - Parent Version. The FLEQ-P is a wide-ranging measure developed by the first named author which assesses four broad classes of variables. Demographic variables assessed were Gender; Age; Racial/ethnic background; Birth position; With whom the child mainly lives; Parental divorce; Family income; Father's education; Father's occupation; Father's alcohol/drug abuse; Mother's education; Mother's occupation; and Mother's alcohol/drug abuse. Family characteristics were measured in terms of: Traditional family values of father and mother (Finkelhor, 1979); Parental sexual punitiveness (adapted from Finkelhor, 1979); Interparental relationship satisfaction (adapted from McCabe, 1995); and Family cohesion and adaptability (FACES II) (Olson, Bell & Portner, 1982). The child's experiences in the following areas were rated by the parent: Sexual abuse; Physical abuse; Psychological maltreatment; Neglect; Witnessing family violence; and Physical and verbal affection. The child's current psychological adjustment was determined using five outcome measures: Proneness to emotional upset (5 items from Spivack & Spotts, 1966); Self-esteem (1 item); Gender identity (1 item); Sexual behaviours (Child Sexual Behaviour Inventory - CSBI) (Friedrich, Grambsch, Broughton, Kuiper, & Beilke, 1991); and Behavioural adjustment (Child Behaviour Checklist-CBCL) (Achenbach, 1991). The quality of the child's relationships was also assessed by multiplying each of the ratings of the child's closeness (on a five-point scale of -2 through to +2) to a list of target persons. The total quality score for each relationship had a possible range of -10 through to +10. These were summed for the child's relationships with: mother, father, peers, siblings/cousins, teachers, and other significant adults.

Procedure

Parent reports of child maltreatment (sexual abuse, physical abuse, psychological maltreatment, neglect, witnessing family violence), family functioning, mediating factors, and behavioural and psychosocial adjustment of children aged 5 to 12 were assessed using the FLEQ-P. Where there was more than one child in the family of primary school age, one child was selected by the respondent to be the target of study. Parent were not selected on the basis of the child's maltreatment status; therefore the sample includes 'unreported' and 'unsubstantiated' cases of maltreatment. Only one parent reported that an official report of abuse or neglect had ever been made to the police or a child protection authority concerning the child.

Nearly fifty organisations were contacted by letter outlining the purpose and nature of the study and requesting their participation in the research project, which was followed up by telephone or face-to-face meeting with the organisation's manager/director. Organisations not wanting to participate were three medical or health clinics, three counselling agencies, and one child care centre, fitness centre, youth hostel, large manufacturing company, state social security department and state correctional service.

Thirty-four organisations agreed to participate in the study and allowed a distribution/collection point for the questionnaires to be set up. However, as no completed questionnaires were returned to eight of these organisations, subjects were effectively recruited from 26 different centres. Posters advertising the study were placed above a pile of questionnaires and a box in which to anonymously return completed questionnaires. Data were collected from medical centres (n=8); counselling agencies (n=7); child care centres (n=4); fitness centres (n=4); and maternal and child health centres (n=2), and students completing first year psychology at one regional campus of a multi-campus Australian university (no class credit was given).

Descriptive Sample Statistics

Demographic characteristics. The referent child had an average year level at school of grade 3. The mean age of the child was 104.4 months (8 years, 8.4 months). Male children were more frequently selected by parents as the referent child (n=30; 60%) than females (n=20; 40%). Mean rating on family income was 2.71 (where 2=$15,000-$30,000; 3=$30,000-$45,000). Sixteen children (32%) had experienced parental divorce, occurring at a mean age of 4.17 (sd=2.83). Most children (n=39) lived with either two or three other children (M=2.56; sd=0.97); eight had step-siblings. The child's racial/ethnic background was described as 'Australian' by just over half of the sample (n=26), with almost half coming from a wide variety of racial or ethnic origins. While the majority of the referent children lived with two biological parents (n=30), a sizeable number lived with a biological mother only (n=18). Two children lived with adoptive parents.

Maltreatment characteristics. Parents rated the frequency with which children were subjected to maltreating behaviours. However, to allow comparison to be drawn with previous research, it is helpful to describe the proportion of children who could be classified as having experienced each particular type of maltreatment. Therefore the continuous frequency scores on each of the maltreatment variables were rendered dichotomous by way of a mean-split. Mean-split was preferable to a median-split as the median score was lower than the mean for all five maltreatment types. The median for both neglect and sexual abuse was zero, and so all subjects who scored above zero would be classified as 'maltreated' when using a median-split, even though they may have experienced sexual abuse or neglect only infrequently. This was a more conservative estimate of the proportion of children who could reasonably be classified as 'maltreated'. Creating a dichotomous variable allows the percentage of children who have experienced moderate or high levels of each type of abuse and neglect to be calculated. However, it should be noted that making a distinction between low and high scores on the maltreatment variables is purely arbitrary and masks some of the variability within the two groups. Using the dichotomous maltreatment variables, the number of children whose maltreatment scores were above the mean was calculated. The most frequently experienced maltreatment types were neglect (36%) and witnessing family violence (36%), followed by psychological maltreatment (32%) and physical abuse (30%), with the least frequently experienced type being sexual abuse (10%). The frequencies with which children were above the mean on none, one, two, three, four or five maltreatment types is shown in Figure 1 - Percentage of Subjects Who Had Scores Above the Mean on More Than One Maltreatment Variable

Relations between the five maltreatment types assessed (physical abuse, sexual abuse, psychological maltreatment, neglect, witnessing family violence) are presented in the matrix of intercorrelations presented in Table 1 - Intercorrelation Matrix of Maltreatment Variables

The correlations between sexual abuse and other types of maltreatment were not significant, except with psychological maltreatment (r = .28, p = .023). Intercorrelations between the other four maltreatment variables were all significant and ranged from .52 (neglect and physical abuse) to .73 (neglect and psychological maltreatment). This indicates a large degree of overlap between each of these four maltreatment types. Parents reported the highest amount of overlap between psychological maltreatment and the other three maltreatment types (physical abuse, neglect, witnessing family violence).

Outcome measures. The correlation between the five maltreatment variables and adjustment measures are presented in Table 2. The highest number of correlations between a maltreatment type and the six measures of adjustment was found for neglect, with all correlations except that between neglect and CSBI being significant. High levels of neglect were reported by parents as demonstrating higher levels of internalising and externalising behaviour problems, greater proneness to emotional upset (i.e., anger and fear), a more negative view of their self-worth, and greater gender dissatisfaction.

Table 2 - Correlations Between Maltreatment Variables and Measures of Adjustment and Family Background Characteristics

Physical abuse, psychological maltreatment and witnessing family violence all had significant correlations with four of the adjustment measures. The only outcome measure correlating with sexual abuse was the CSBI (r = .28, p = .024); also the CSBI was not significantly correlated with any of the other four maltreatment variables. Gender identity was the only outcome measure which did not correlate significantly with physical abuse or psychological maltreatment. The only maltreatment type not strongly associated with internalising behaviour problems, proneness to emotional upset and self-esteem was sexual abuse. The frequency with which children experienced maltreating behaviours across all five maltreatment types (TotMal) was significantly correlated with CBCL Internalising (r = .70), CBCL Externalising (r = .54), self-esteem (.41), proneness to emotional upset (.39) and gender identity (.31) (see Table 6). The correlation between TotMal and CSBI was not significant.

Family background. The relationship of the five maltreatment variables to family background are also indicated by the correlations presented in Table 2. Interparental relationship satisfaction and parental drug/alcohol abuse were not significantly associated with any of the maltreatment variables. Large correlations (p < .001) were also found between CBCL Internalising Behaviour Problems and: CBCL Externalising Behaviour Problems, CSBI, Proneness to emotional upset, gender identity disturbance, low self-esteem, low family cohesion, sexual punitiveness, and low levels of maternal education. The variables other than Internalising Behaviour Problems highly correlated with CBCL Externalising Behaviour Problems were Proneness to emotional upset, gender identity disturbance and low family cohesion. Gender identity was strongly associated with both low self-esteem and low maternal educational level, while divorce and self-depreciation were significantly correlated. Large correlations were found between family cohesion and all variables except CSBI, parental drug use, Father's educational level and Traditionality of father. Family adaptability had only one significant correlation with any measure of adjustment (CBCL Internalising r = -.32, p < .01). Parental sexual punitiveness was highly correlated with low paternal education. The correlations of both parental drug/alcohol abuse and quality of child's relationships with all other key variables were not highly significant (p > .001).

Discriminant Function Analysis

A stepwise discriminant function analysis was performed using family background variables (parental sexual punitiveness, traditionality of parental family values, parental divorce, quality of child's relationships, family income, family adaptability, family cohesion, child's gender) to predict group membership: nonmaltreated; one-to-two different maltreatment types; three-to-five different maltreatment types. All cases were included for both prediction and classification (n = 50). After four steps, family cohesion, parental sexual punitiveness, traditionality of parental family values and parental divorce were the predictor variables included in the analysis.

The data yielded two discriminant functions (k-1) with a combined c2(8) = 20.26, p = .0094. After removal of the first function, however, the association between groups and predictors was not significant. The first function accounted for 81% of the between group variability, and maximally separated the group that experienced three-to-five maltreatment types from the other two groups: those who had experienced no maltreatment or one-to-two different types. The loading matrix of correlations between predictors and the discriminant function suggests that the best predictors for distinguishing between children classified as having experienced many different kinds of maltreatment and those children experiencing none, one or two types of abuse or neglect were family cohesion, parental sexual punitiveness, traditionality of parental family values and parental divorce. For classification, prior probability of group membership was set at sample proportions for the subjects entered into the analysis. With the use of a classification procedure (without jackknifing) for the total useable sample of 50 students, 52.0% were classified correctly. Prediction of the second group (one-to-two different maltreatment types) was the poorest, with only 26.7% classified correctly. Children who had experienced three-to-five different maltreatment types less likely to be correctly classified (53.3% correct classifications) than nonmaltreated children (70% correct classifications) using the current set of predictor variables.

Multiple Regression Analyses

Predicting maltreatment types from family background variables. Standard multiple regression analysis was used to assess the contribution of family background variables (parental sexual punitiveness, quality of interparental relationship, quality of child's relationships, parental divorce, positive affection, traditionality of mother and father, family adaptability and cohesion) to each of the five maltreatment types. Family background variables did not significantly predict either sexual abuse (F(9, 40) = 0.50, p = .86) or physical abuse (F(9, 40) = 1.77, p = .10). Using psychological maltreatment as the dependent variable, R for regression was significantly different from zero, F(9, 40) = 5.74, p = .0000. Three independent variables uniquely contributed significantly to prediction of psychological maltreatment: family cohesion (sr2 = .07), parental sexual punitiveness (sr2 = .05), and parental divorce (sr2 = .04). In combination, the nine predictor variables accounted for 56% (47% adjusted) of the variance in psychological maltreatment.

Using neglect as the dependent variable, R for regression was significant, F(9, 40) = 4.12, p = .0008. While in combination the nine predictor variables accounted for 48% (36% adjusted) of the variance in neglect, none provided unique prediction. Using witnessing family violence as the dependent variable, R for regression was significant, F(9, 40) = 4.25, p = .0007. Divorce was the only independent variable which contributed significant unique prediction to witnessing family violence (sr2 = .07). In combination, the nine predictor variables accounted for 49% (37% adjusted) of the variance in witnessing family violence.

Predicting adjustment from maltreatment variables. Standard multiple regression analysis was used to assess the contribution of the five maltreatment types to three measures of adjustment: CBCL Internalising Behaviour Problems; CBCL Externalising Behaviour Problems and the CSBI. Using the CBCL Internalising Behaviour Problems scale as the dependent variable and sexual abuse, physical abuse, neglect, psychological maltreatment, and witnessing family violence as independent variables, R for regression was significant, F(5, 44) = 14.81, p = .0000. Neglect was the only independent variable which significantly contributed unique prediction of Internalising Behaviour Problems (sr2 = .20). In combination, the five predictor variables accounted for 63% (58% adjusted) of the variance in Internalising Behaviour Problems.

Using the CBCL Externalising Behaviour Problems scale as the dependent variable and the same independent variables as above, R for regression was significantly different from zero, F(5, 44) = 5.53, p = .0005. Again, neglect was the only independent variable which significantly contributed unique prediction of Externalising Behaviour Problems (sr2 = .06). In combination, the five predictor variables accounted for 39% (32% adjusted) of the variance. CSBI was not able to be predicted from the five maltreatment variables (F(5, 44) = 1.20, p = .33).

Predicting adjustment from family background variables and maltreatment. Standard multiple regression analysis was used to assess the contribution of both family background variables (parental sexual punitiveness, quality of interparental relationship, closeness and quality of child's relationships, parental divorce, positive affection, traditionality of parents, family adaptability and cohesion) and maltreatment (sexual abuse, physical abuse, neglect, psychological maltreatment, witnessing family violence) to three measures of adjustment: CBCL Internalising Behaviour problems; CBCL Externalising Behaviour problems and the CSBI. Using the CBCL Internalising Behaviour Problems scale as the dependent variable, R for regression was significant, F(14, 35) = 6.03, p = .0000. Two independent variables significantly contributed unique prediction of Internalising Behaviour Problems: neglect (sr2 = .14) and family cohesion (sr2 = .05). In combination, the fourteen predictor variables accounted for 71% (59% adjusted) of the variance in Internalising Behaviour Problems.

With CBCL Externalising Behaviour Problems as the dependent variable, R for regression was significant, F(14, 50) = 3.44, p = .0015. Family cohesion (sr2 = .09) and family adaptability (sr2 = .05) were the two independent variables which uniquely predicted Externalising Behaviour Problems, and in combination the fourteen predictor variables accounted for 58% (41% adjusted) of the variance. CSBI scores were not able to be predicted by either family background or maltreatment variables (F(14, 50) = 1.14, p = .36). However, regression analyses used to predict adjustment from family background and maltreatment variables should be treated with caution. The ratio of cases to variables was barely sufficient to satisfy the assumptions of the test and this which may have compromised statistical power.

These results suggest that neglect and family functioning (low cohesion and low adaptability) were the most important predictors of adjustment problems. Family characteristics were also important in predicting three of the five maltreatment types (but not physical or sexual abuse). Family background factors improved prediction of externalising behaviour problems from that predicted by maltreatment.

The results showed that all maltreatment types were strongly associated with each other, except for sexual abuse. A large degree of overlap was found between physical abuse, psychological maltreatment, neglect and witnessing family violence, indicating that if a child has experienced one type of abuse or neglect, there will be a high probability that they will also have experienced one or more other maltreatment types. Psychological maltreatment had the strongest level of association with the other maltreatment types. As the associations between psychological maltreatment and physical abuse, neglect and witnessing family violence are so strong, it can be assumed that the child victim of psychological maltreatment is also the victim of one or more other types of abuse or neglect. This supports the hypothesis that rather than occurring in isolation, multiple forms of maltreatment co-exist.

The relative isolation of sexual abuse is surprising, given the current literature on the overlap between sexual abuse and other types of maltreatment, particularly physical abuse (e.g., Harrison, Edwall, Hoffman & Worthen, 1990; Lanktree, Briere & Zaidi, 1991). This finding can be explained by the low number of primary school-aged children who were described by their parents as having experienced sexual maltreatment. The self-selecting sample bias may account for the relatively low frequency of sexual abuse when compared to both the frequency of other maltreatment types, and to national prevalence data. While willing to disclose other maltreating behaviours directed by themselves or others toward the child, the social stigma and high level of negative publicity associated with child sexual abuse perpetrators would make such disclosure difficult.

Use of a mean-split on the maltreatment continuous scores to render them dichotomous was believed to be a conservative measure of maltreatment. Yet, many children were described by their parents as being subjected to these maltreating behaviours at a frequency above the sample mean. With the exception of sexual abuse, which was reported by only 10% of respondents, other types of maltreatment were experienced by between 30% and 36% of children, with only 40% of children being classified as nonmaltreated. Only 18% of all children experienced only one type of maltreatment, which is one-third of the maltreated children; the remaining two thirds experienced between two and five different types of maltreatment. These results appear consistent with the extent of overlap between the maltreatment types described in a recent review of the literature on revictimisation (Hamilton and Browne, in press).

Neglect had a strong impact on adjustment. Neglect was strongly associated with both other maltreatment types and with five of the six measures of adjustment. This type of abuse was the most strongly related to poor adjustment in children, followed by psychological maltreatment. This supports the findings of previous studies (e.g., Ney et al., 1994) that the worst combinations of child maltreatment include 'neglect'. Neglect was the only maltreatment variable which provided unique prediction of children's adjustment and was significantly correlated with more of the measures of adjustment problems than the other maltreatment types. The damaging effects of neglect (alone, or in combination with other types of maltreatment) needs to be given greater recognition in our understanding of the process by which child maltreatment impacts on the adjustment of children. Neglect accounted uniquely for more than one-third of the adjusted R2 in predicting CBCL Internalising Behaviour Problems.

Maltreating families are likely to inflict multiple forms of abuse or neglect on children. The most important implication of this finding is that when an assessment is made of only one type of maltreatment, erroneous conclusions may be made about the specific impact of that form of abuse or neglect. The more types of maltreating behaviours experienced by children, the worse their adjustment. As the number of different types of maltreatment experiences increased, children experienced more internalising, externalising and sexual behaviour problems, had a greater proneness to emotional upset, and had lower levels of satisfaction with their gender. The relationship of multi-type victimisation to lowered self-esteem was not significant. Maltreating families are characterised by low cohesion, divorce, and sexual punitiveness towards children. Strong relationships with these aspects of family background were found for psychological maltreatment, witnessing family violence and neglect. Family factors were not good predictors of physical abuse and sexual abuse. This appears inconsistent with past research, particularly studies which have demonstrated a strong relationship between both structural and dynamic aspects of families and the risk of sexual abuse (e.g., Finkelhor, 1979). However, the lack of relationship between sexual abuse and family characteristics may be due to the very low frequency with which sexual abuse was reported in the current study.

These results also provide some preliminary answers to the question posed by Hamilton and Browne (in press) how family cohesion, high parental sexual punitiveness, traditional values held by parents concerning women and children, and parental divorce were factors which provided significant discrimination between children classified as having experienced many different types of maltreatment and those classified as experiencing few or no different types. While results of the discriminant function analysis classification were only modest (52% correct), the function on which the above four family background factors all loaded was statistically significant. It appears that children from intact families, where there is high degree of cohesiveness, where parents are less traditional in their beliefs about the role of women and children, and where parents are less punitive in their responses to children's and behaviours concerning sexuality are - to a limited degree - protected from multiple types of victimisation. These were essentially the same types of family background characteristics that provided unique prediction of three of the maltreatment types: psychological maltreatment, neglect and witnessing family violence. These family background characteristics, as well as mediating the risk of an initial experience of certain maltreatment types, also mediate the ongoing risk of further multi-type maltreatment.

Family background variables did not contribute greatly to prediction of internalising behaviour problems above that predicted by maltreatment characteristics. However, for externalising behaviour problems, family background variables provided a significant additional contribution to explaining adjustment, with both family adaptability and cohesion providing unique prediction. It seems that many family factors influence adjustment through their impact on the risk of maltreatment. Rather than having a direct influence on adjustment, the impact of family characteristics (e.g., divorce, rigid family dynamics) is largely mediated through their relationship to the various types of abuse and neglect.

While divorce was not significantly correlated to internalising, externalising or sexual behaviour problems, it was strongly related to disturbances to both gender identity and self-esteem. It was also associated with an increased risk of maltreatment types (except sexual abuse). Similarly, while low family adaptability (rigidity) was significantly correlated with only one measure of adjustment (internalising behaviour problems), it was also associated with a higher degree of psychological maltreatment, neglect and witnessing family violence. These results are interpreted as further support for the mediating role of the family environment in the relationship of maltreatment and the adjustment of children. The findings of the current study highlight these structural and dynamic aspects of families (e.g., parental divorce, low cohesion, low adaptability) and the role they play in increasing the risk of maltreatment occurring (both single-type and multi-type), and the likelihood of adjustment problems.

Briere, J., and Runtz, M. (1988). Multivariate correlates of childhood psychological and physical maltreatment among university women. Child Abuse and Neglect, 12, 331-341.

Finkelhor, D. (1979). Sexually victimised children. New York: Free Press.

Finkelhor, D. (1995). The victimization of children: A developmental perspective. American Journal of Orthopsychiatry, 65, 177-193.

Friedrich, W. N., Grambsch, P., Broughton, D., Kuiper, J., and Beilke, R. L. (1991). Normative sexual behaviour in children. Pediatrics, 88, 456-464.

Hamilton, C.E., & Browne, K.D. (in press). The repeat victimisation of children: Should the concept be revised? Agression and Violent Behaviour, 2.

Harrison, P. A., Edwall, G. E., Hoffman, N. G., and Worthen, M. D. (1990). Correlates of sexual abuse among boys in treatment for chemical dependency. Journal of Adolescent Chemical Dependency, 1, 53-67.

Higgins, D. J., and McCabe, M. P. (1994). The relationship of child sexual abuse and family violence to adult adjustment: Toward an integrated risk-sequelae model. The Journal of Sex Research, 31, 255-266.

Higgins, D. J., and McCabe, M. P. (submitted). Assessment of child maltreatment: The Family and Life Experiences Questionnaire. Child Abuse and Neglect.

Krishnan, V., and Morrison, K. B. (1995). An ecological model of child maltreatment in a Canadian province. Child Abuse and Neglect, 19, 101-113.

McCabe, M. P. (1995). Sexual Function Scale. Melbourne: School of Psychology (Burwood Campus), Deakin University.

McGee, R. A., Wolfe, D. A., Yuen, S. A., Wilson, S. K., and Carnochan, J. (1995). The measurement of maltreatment: a comparison of approaches. Child Abuse and Neglect, 19, 233-249.

Mannarino, A. P., Cohen, J. A., and Berman, S. R. (1994). The relationship between preabuse factors and psychological symptomatology in sexually abused girls. Child Abuse and Neglect, 18, 63-71.

Ney, P. G., Fung, T., and Wickett, A. R. (1994). The worst combinations of child abuse and neglect. Child Abuse and Neglect, 18, 705-714.

O'Hagan, K. P. (1995). Emotional and psychological abuse: Problems of definition. Child Abuse and Neglect, 19, 449-461.

Olson, D. H., Bell, R. Q, and Portner, J. (1982). FACES II: Family adaptability and cohesion evaluation scales. St Paul, MN: Family Social Science.

Rosenberg, M. S. (1987). New directions for research on the psychological maltreatment of children. American Psychologist, 42, 166-171.

Sanders, B. and Becker-Lausen, E. (1995). The measurement of psychological maltreatment: Early data on the child abuse and trauma scale. Child Abuse and Neglect, 19, 315-323.

Spivack, G., and Spotts, J. (1966). Devereaux Child Behaviour Rating Scale. Devon, PA: Devereux Foundation.

Stanley, J., and Goddard, C. (1993). The association between child abuse and other family violence. Australian Social Work, 46, 3-8.