ChiMAbuse & Neglect,Vol. 16, pp. 647--659,1992

0145-2134/92$5.00+ .00 Copyright© 1992PergamonPressLtd.

Printed in the U.S.A.All rightsreserved.

CHILDHOOD HISTORY OF ABUSE AND CHILD ABUSE SCREENING JOHN A. CALISO Franklin Lakes Public Schools, Franklin Lakes, NJ

JOEL S. MILNER Department of Psychology, Northern Illinois University, DeKalb, IL

Abstract--Although a childhood history of abuse is related to parental child abuse, many parents with a history of abuse are not abusive. To determine the effects of a childhood history of abuse on adult child abuse potential, a modified Conflict Tactics Scale (CTS) and the Child Abuse Potential (CAP) Inventory were administered to matched groups of physically abusive mothers with a childhood history of abuse, nonabusive comparison mothers with a childhood history of abuse, and nonabusive comparison mothers without a childhood history of abuse. The modified CTS asked about childhood events and was used to confirm a childhood history of abuse. As expected, the CTS verbal and violence scales were higher for the abusive and nonabusive mothers with a childhood history of abuse. None of the CTS scores were different for the abusive and nonabusive mothers with a childhood history of abuse. In contrast, the CAP abuse scores distinguished between all three study groups. However, on the CAP factor scales, only the rigidity and unhappiness factors discriminated between abusive and nonabusive mothers with a childhood history of abuse. Nonabusive mothers with a childhood history of abuse were less rigid in their child expectations and were happier in their interpersonal relationships than abusive mothers with a childhood history of abuse. Key W o r d s - - C h i l d h o o d history, Child abuse screening.

A RELATIONSHIP BETWEEN the childhood experience of abuse and parental child abuse has been mentioned frequently in the family violence literature (Altemeier, O'Conner, Vietz, Sandler, & Sherrod, 1982; Baldwin & Oliver, 1975; Berger, 1980; Daniel, Hampton, & Newberger, 1983; Fontana & Bernard, 1971; Friedrich & Wheeler, 1982; Heifer & Kempe, 1987; Johnson & Morse, 1968; Martin, 1976; Wasserman, 1967). Although a large number of studies support the intergenerational hypothesis, Widom (1989), in a recent review, points out that existing studies suffer from a number of methodological flaws and suggests "that conclusions about the strength of the cycle of violence be tempered by the dearth of convincing empirical evidence" (p. 3). As we attempt to determine the extent of intergenerational effects on the occurrence of child abuse, we also need to determine which factors related to the childhood abuse experience contribute to violence transmission. One popular view is that early social learning of aggressive responses to others, especially other family members, through the experience and observation of aggression during childhood provides a link to later parental child abuse (Straus, Gelles, & Steinmetz, 1980). In addition to aggression, other parental characteristics that may contribute directly or indirectly to the intergenerational transmission of abuse inReceived for publication May 2, 1990; final revision received September 27, 1990; accepted November 2, 1990. Requests for reprints may be sent to Joel S. Milner, Family Violence Research Program, Department of Psychology, Northern Illinois University, DeKalb, IL 60115-2892. 647

64~

J. A Calisoand J. S. Milner

clude parental attachment problems, rigid expectations of child behavior, unique attributional styles, the lack of parental coping and parenting skills, and parent psychopathology. Egeland, Jacobvitz, and Sroufe (1988) found that mothers who continue the cycle of violence, relative to those who do not, experience more stress and are more "anxious, dependent, immature, and depressed" (p. 1080). Each of these parental characteristics may produce negative consequences for the child who is the target of child abuse, albeit the consequences will likely vary as a function of the child's developmental stage. Helfer (1987) has discussed the "World of Abnormal Rearing" that extends beyond physical child abuse and includes childhood situations that do not provide adequate environments for the learning of basic interpersonal skills. These inadequate conditions are viewed as "a never ending cycle, passing from one generation to another" (p. 71 ). Parental factors combined with sociological problems (e.g., inadequate housing, poor nutrition and medical care, unemployment, drug use, large families, and single-parent families) may contribute to parent-child interactional difficulties and to a social milieu in which children develop inadequate cognitive schema (e.g., beliefs, perceptions, values, attitudes, and expectations). Problems in the abused child's socialization (e.g., peer relationships and school adjustment) and later interpersonal and parenting problems, including child abuse, may arise from the combination of these negative experiences and conditions. Less is known about childhood experiences that may serve as buffering, attenuating, or countervailing variables for the transmission of child abuse. Although data are generally lacking on this question, Egeland et al. (1988) reported that mothers who break the cycle of violence, relative to those who continue the cycle, received emotional support from an adult during childhood and have successful, emotionally supportive relationships with a mate. Another study reports correlational evidence indicating a significant inverse relationship between the self-reported childhood experience of a caring adult or a caring friend and adult child abuse potential (Milner, Robertson, & Rogers, 1990). At present, however, the meaning of an emotionally supportive adult or spouse during adulthood and a caring adult or friend during childhood remains to be fully specified. It is also possible that children who establish caring relationships during childhood vary on other characteristics that mitigate the transmission of abuse across generations so that no direct relationship exists between the childhood experience of caring relationships and later child abuse potential. In addition to the studies reporting a relationship between the childhood experience of abuse and the later occurrence of parental child abuse, expected relationships between a childhood abuse history and child abuse measures have been found. For example, a positive relationship between the occurrence and the severity of childhood abuse and Conflict Tactics Scale (CTS) scores have been reported (Straus et al., 1980). In another study, Chan and Perry (1981) found that students who indicated a childhood history of abuse earned significantly higher Child Abuse Potential (CAP) Inventory abuse scores than students who did not report a childhood history of abuse. Supporting these findings, Milner et al. (1990) found significant, albeit modest, relationships between the observation, receipt, and severity of childhood physical abuse and CAP Inventory abuse scores. Related to the understanding of the childhood experience of abuse, there is a need to delineate adult personality factors that distinguish between individuals who experience childhood abuse and do not engage in child abuse and those who experience childhood abuse and abuse their children. The determination of personality characteristics that distinguish these groups would further define our child abuse models and provide information for the differential assessment and treatment of child abusers. Since a childhood abuse history is correlated with parental child abuse and adult child abuse potential, but not all abuse parents abused their children, the present study investigated the effects of a childhood history of abuse on responses to a modified CTS questionnaire

649

Child abuse screening Table 1. Demographic Characteristics of the Parent Groups Parent Group Demographic Characteristic Ethnic background (%) Black White Age (years) M SD

Education (years) M SD

Marital status (%) Single Married Separated Divorced/Widow Children (number) M SD

Abusers w/History

Comparisons w/History

Comparisons w/o History

46.7 53.3

23.3 76.7

36.7 63.3

30.4 6.3

34.0 6.6

30.2 6.3

12.1 1.5

13.2 2.0

12.5 1.1

16.7 30.0 23.3 30.0

6.7 40.0 26.7 26.7

10.0 50.0 26.7 13.3

2.1 1.2

2.5 1.1

2.6 1.2

and to the CAP Inventory in a group of physical child abusers with a childhood history of physical abuse and two nonabusive comparison groups, one with and one without a childhood history of physical child abuse. A number of analyses were conducted on the CTS and CAP scores between and across the study groups. Of special interest was the assessment of the CAP Inventory's ability to distinguish physical child abusers with a childhood history of abuse from nonabusive comparison parents with a childhood history of abuse. As noted by Milner (1989a) and reviewers of the CAP Inventory's psychometric research, studies such as the present study that challenge the limits of the CAP Inventory's discrimination abilities in different groups of abusive and nonabusive groups are needed.

METHOD

Subjects Ninety female parents took part in the study, including 30 physical child abusers who reported a childhood history of physical abuse, 30 nonabusive comparison parents who reported a childhood history of physical abuse, and 30 nonabusive comparison parents who reported no childhood history of physical abuse. Nonabusing comparison parents in the history and no history of child abuse groups were individually matched on demographic variables with each of the abusive parents. The demographic characteristics of each of the three study groups are presented in Table 1. Univariate analyses of variance (each with 2 and 87 degrees of freedom) were conducted to explore the possibility of group differences in age, education, and number of children. No significant group differences (p > .01) were found. Additional analyses (using frequencies) indicated that the three study groups were not significantly different in ethnic background, X2(2,N = 90) = 3.59, p > .01, or marital status, ×2(6,N = 90) = 4.98, p > .01. Although repeated nonindependent testing usually requires some adjustment (e.g., Bonferroni's procedure) to avoid excessive power, no adjustment was made in the alpha level of the tests of demographic differences because the .01 alpha level was used. It might be noted that the

650

J. A. Calisoand J. S. Milner

unadjusted .01 alpha level provides the same level of actual power as use of the .05 alpha level with Bonferroni's correction (i.e., .05/5 = .01). These results indicate that even though exact individual subject matching was not achieved, group matching across the three study groups was achieved on each of five demographic variables.

Subject Selection The physical child abusers were physically abusive mothers identified as the perpetrator by child welfare agencies in New Jersey that were involved in the identification and treatment of child abusing parents. Participants were recruited from the Division of Youth and Family Services and the Correctional Institution for Women. At the time of research participation each of the abusive mothers admitted the physical child abuse. Although admission of child abuse may decrease the likelihood of false-positive classifications in the child abuse group, the fact that all abusive mothers admitted their abuse limits the generalizability of the study results. Abusive mothers who admit their abuse may differ from abusers who do not admit abuse on factors (e.g., guilt) that may affect study results. In the present study, physical child abuse was defined in a manner that was congruent with the definition used by the Division of Youth and Family Services in New Jersey. A child was considered to be physically abused when the mother inflicted bruises, welts, cuts, broken bones, skull fractures, burns, poisoning, internal injuries of soft tissue and organs, and injuries to the bone and tissue joints of her child under 18 years of age. Known cases of sexual abuse and child neglect were not included in the present study. The nonabusive comparison parents who reported a childhood history of physical abuse were obtained from community health resource programs that included: Shelter Our Sisters, Teaneck, N J; Alternatives to Domestic Violence, Hackensack, N J; and Parents Anonymous, Parisppany, NJ. Participants were recruited by caseworkers on the basis of a review of intake records that indicated the parents had been physically abused in childhood but had not abused their children. Other participants were recruited from newspaper advertisements. The advertisements asked for volunteers who believed that their parents had used excessive physical punishment regularly but who did not use the same methods with their own children. Those responding to the advertisements were interviewed by the senior author in an attempt to confirm the childhood history of physical abuse. Each interview was conducted in the respondent's home and lasted approximately 2 hours. The interview included questions regarding the type and frequency of childhood abuse, a description of the perpetrator, and the role of the other parent. The nonabusive comparison parents who reported no childhood history of abuse of any type were obtained by the senior author from community organizations such as: Paterson Day Care 100 Program, Paterson, NJ; and Our Lady of Mercy, Jersey City, NJ. Mothers were included in this nonabusive comparison group based on interview data that indicated that no childhood history of abuse had occurred and that the mother was not abusive. Scores on a modified CTS were used as a procedure check to verify the presence or absence of a childhood history of abuse in each group.

Test Instruments Child Abuse Potential Inventory.The CAP Inventory is a 160-item questionnaire designed to assist in the screening of physical child abuse reports (Milner, 1986a). The questionnaire is most appropriately used in social service and similar settings where base rates for physical child abuse among physical child abuse reports range from 35-50%. The questionnaire is client administered and is answered in a forced-choice, agree-disagree format. The CAP Inventory contains a physical child abuse scale and six descriptive factor scales. The factor scales are

Child abuse screening

651

the distress scale, the rigidity scale, the unhappiness scale, the problems with child and self scale, the problems with family scale, and the problems from others scale. The CAP Inventory factor structure has been replicated in cross-validation research (Milner, 1986a). The most recent version of the CAP Inventory (Milner, 1986a) also contains three validity scales (i.e., lie, random response, and inconsistency scales) that are used in various combinations to form three response distortion indexes: the faking-good index, the faking-bad index, and the random response index. The version of the CAP Inventory used in the present study contained the same 77 physical abuse scale items as the most recent version of the CAP Inventory. However, the version of the CAP Inventory used in the present study did not contain the items necessary to allow computation of any of the validity indexes. Therefore, no measures of response distortion were available. The development of the CAP Inventory is described in two initial studies (Milner & Wimbedey, 1979, 1980) and in a technical manual (Milner, 1986a). Supportive cross-validation data on the concurrent validity of the CAP Inventory are available (e.g., Milner, 1989a, 1991a; Milner, Gold, Ayoub, & Jacewitz, 1984; Milner & Robertson, 1989) and different applications and limitations of the CAP Inventory have been discussed (e.g., Melton & Limber, 1989; Milner, 1986b, 1989b, 1990, 199 lb). Predictive validity data from a longitudinal study of at-risk parents have been published. These data indicate a significant relationship between elevated CAP abuse scores and later physical child abuse (Milner, Gold, & Wimberley, 1986). Extensive construct validity data are also available and are summarized in the technical manual (Milner, 1986a). CAP Inventory abuse scale reliabilities appear to be adequate across a variety of groups (Milner, 1986a). For example, internal consistency (KR20) reliabilities for the abuse scale range from .92-.96 for groups of abusive (n -- 152) and comparison subjects (n = 2,062). Temporal stability estimates for l-day, l-week, l-month and 3-month intervals for the abuse scale are .9 l, .90, .83, and .75, respectively. For the present study sample (n -- 90), the CAP Inventory abuse scale internal consistency (alpha) reliability was .93.

Conflict Tactics Scale. A modified CTS, designed to assess family conflict resolution techniques (Straus et al., 1980), was chosen as an additional measure of a childhood history of abuse. The CTS questionnaire (Form N) used in the present study contained 18 items (i.e., items a through r) and an "other" category (i.e., item s). Since the CTS was used as an independent procedure check of the subject selection and group assignment process, the instructions for completing the CTS were modified. CTS instructions were similar to the original instructions except that the respondents were asked to indicate how disputes with their parents were handled in any 1 year that most described their childhood interactions with their parents. No changes were made in the individual CTS items. The CTS scale (Form N) contains a number of subscales that can be scored (Straus & Gelles, 1990). In the present study, the CTS reasoning scale, that contains three items (i.e., items a through c), was scored; the CTS verbal aggression scale, which consists of seven items (i.e., d through j), was scored; and the CTS violence scale, which consists of eight items (i.e., k through r), was scored. On each of the three CTS scales, the number (i.e., 0 through 6) indicating the seven-point Likert-type response for each item was summed across the scale items to obtain the CTS scale score. No additional weighting of the item responses was used. For the present study sample (n -- 90), the CTS reasoning scale (three items) internal consistency (alpha) reliability was .47, the CTS verbal abuse scale (seven items) internal consistency (alpha) reliability was .83, and the CTS violence scale (eight items) internal consistency (alpha) reliability was .85. In a supplemental analysis, the CTS factor structure was analyzed for the present study sample. This exploratory factor analysis was conducted because the CTS instructions were

¢~52

J.A. Calisoand J. S. Milner

modified to gain information on the parents" childhood history of family interactions rather than the parents' interactions with their own children and it was not known how this change might affect the CTS factor structure. The total study sample was used in an attempt to maintain an adequate n/variable ratio for the analyses. Even using the total sample, the n/variable ratio was only 5.0. Two-, three-, four-, five-, and six-factor maximum likelihood factor solutions were estimated for the total sample. Using the eigenvalues and the conceptual content of the rotated factors to evaluate the different factor solutions, it appeared that the four-factor solution was the most meaningful solution. A three-factor solution was expected because Straus et al. (1980) indicated that the CTS measures reasoning, verbal abuse, and violence. However, the four-factor solution most closely resembled the factor structure suggested by Straus's three CTS scales. That is, three of the four factors suggested the reasoning, verbal abuse, and violence scales developed by Straus. The fourth factor was a second violence dimension that consisted of the last two items on the violence scale. This factor was also present in the three-factor solution where the verbal abuse and violence factors combined into one factor. Conceptually the last two items in Straus's violence scale appear to deal with more severe violence. The items indicate that the respondent's parent had "threatened with a knife or gun" and that the parent had "used a knife or gun." The loadings on the fourth factor for these two items were .86 and .92, respectively, with cross loadings on the violence factor of only. 18 and .03, respectively, suggesting that the new factor is measuring a different type of violence. As previously indicated, one factor was a reasoning factor and as expected all three of the CTS reasoning items had their highest loadings (all above .40) on this factor. The only crossloaded factor was the "cried" item from the verbal abuse scale that loaded highest on the reasoning scale. A second factor was the verbal abuse factor and this factor contained five of the eight expected CTS verbal abuse items. The exceptions were the "cried" item, which loaded on the reasoning scale, and two other items ("did or said something to spite me" and "threatened to hit or throw something at me") that had slightly higher loadings on the violence scale. A third factor was the violence factor with the first six of the eight CTS violence items loading as expected on the violence factor. The last two items, as previously mentioned, formed the new more severe violence factor. For all four factors, all primary factor loadings not previously mentioned were above .40 loadings. Although the exploratory factor analysis suggested the possibility of a fourth factor, the CTS was analyzed using the three scales suggested by Straus et al. (1980). This decision was made because of the exploratory nature of the factor analysis, the limited n/variable ratio, and the low frequency of response to the last two items by subjects in each of the study groups. Only 20% (n = 6), 17% (n = 5), and 0% of the abusive mothers, the nonabusive comparison mothers with a childhood history of abuse, and the nonabusive comparison mothers without a childhood history of abuse, respectively, indicated any affirmative response to even one of the last two CTS items. Procedure Case workers administered the CTS and CAP Inventory to the child abusing parents. To increase the likelihood of a standard administration, an inservice training session was provided for primary professionals, supervisors, and administrators at each research site. The nonabusive comparison mothers with a childhood history of abuse who responded to the newspaper advertisement were given the study materials by the senior author. In addition, seven case workers at the community health resource programs were trained to administer the scales to nonabusive comparison parents with a childhood history of abuse. The nonabusive comparison parents with no childhood history of abuse were given the scales by the senior

653

Child abuse screening

Table 2. Group Means (Standard Deviations) for the Modified Conflict Tactics Scale (CTS) Reasoning, Verbal Abuse, and Violence Subscales Parent Group CTS Subscale Reasoning M SD Verbal Abuse M SD Violence M SD

Abusers w/History

Comparisons w/History

Comparisons w/o History

4.6 4.3

5.4 5.3

6.5 3.4

20.4 13.1

25.5 10.5

8.1 6.4

25,4 7,2

26.6 9.1

5.2 4.3

author. During the testing, each subject was presented a statement of confidentiality and an informed consent. After the subject completed a demographic sheet, the CTS and CAP Inventory were presented. Following completion of the two questionnaires, the subject was thanked and questions concerning the research were answered.

RESULTS The CTS reasoning, verbal abuse, and violence scores that were obtained as a procedure check on the case history/interview data for the abusers and the two nonabusive comparison groups are presented in Table 2. A MANOVA was used to test for group differences in the three CTS scale scores. The MANOVA was significant, Wilks' lambda = .31, F(6,170) = 22.79, p < .001, therefore follow-up univariate analysis of variances were conducted (each with 2 and 87 degrees of freedom). Analysis revealed no significant differences (p > .05) between groups on the CTS reasoning scale. However, significant group differences (p < .01) were found on the CTS verbal abuse and violence scales. Newman-Keuls' tests performed on the CTS verbal abuse scores revealed no significant differences (p > .05) between the abusers with a childhood history of abuse and the nonabusive comparison subjects with a childhood history of abuse, but both of these groups showed significantly higher (p < .01) levels of verbal abuse than the nonabusive comparison subjects without a childhood history of abuse. Newman-Keuls' tests conducted on the CTS violence scales revealed no significant differences (p > .05) between the abusers with a childhood history of abuse and the nonabusive comparison subjects with a childhood history of abuse, but both of these groups showed significantly higher (p < .01) levels of violence than the nonabusive comparison subjects without a childhood history of abuse. The CAP Inventory abuse score and descriptive factor scale scores for the abusers and the two nonabusive comparison groups are presented in Table 3. An analysis of variance indicated significant (p < .01) group differences in CAP physical child abuse scores. A NewmanKeuls' test revealed that each of the group mean abuse scores were significantly different (p < .01) from each of the other group means with the highest abuse scores obtained by the abusive mothers, the intermediate abuse scores obtained by the nonabusive comparison mothers with a childhood history of abuse, and the lowest abuse scores obtained by the nonabusive comparison mothers without a childhood history of abuse. A MANOVA was used to test for group differences in the six CAP Inventory descriptive factor scale scores presented in Table 3. The MANOVA was significant, Wilks' lambda

654

J. A. Caliso and J. S. Milner

Table 3. Group Means (Standard Deviations) for the Child Abuse Potential (CAP) Inventory Abuse Scales Parent Group CAP Scale

Abusers

Comparisons

Comparisons

w/History

w/History

w/o History

304.4 67.6

160.4 85.5

90.2 51.9

205.2 42.1

101.1 58.9

51.0 35.3

21.0 15.5

9.3 8.7

13.4 13.4

33.7 16.3

17.6 14.0

11.1 8.0

7.1 6.9

4.9 5.8

1.4 3.1

19.2 15.5

14.8 14.1

5.6 9.7

18.2 6.9

12.8 8.1

7.7 7.3

Physical abuse M SD

Distress M SD

Rigidity M SD

Unhappiness M SD

Problems with child M SD

Problems with family M SD

Problems from others M SD

= .27,F( 12,164) = 12.60, p < .01, therefore follow-up univariate tests were conducted (each with 2 and 87 degrees of freedom). The univariate Ftests indicated significant (p < .01) group differences for each of the six descriptive factors. Differences in group means were evaluated using the Newman-Keuls' test. On the distress and the problems from others scales, all three groups were significantly different (p < .05) from each other. On the problems with child and the problems with family scales, abusive and nonabusive comparison mothers with a childhood history of abuse were not different (p > .05), but each had significantly higher scores (p < .05) than the nonabusive comparison mothers without a childhood history of abuse. On the rigidity and the unhappiness scales, abusers had significantly higher scores (p < .05) than nonabusive comparison mothers with or without a childhood history of abuse, who did not differ (p > .05) from each other. The correlations between the CTS scales and the CAP Inventory scales are presented in Table 4. The CTS reasoning, verbal abuse, and violence scales were significantly correlated (p < .01) with the CAP Inventory abuse scale. However, as Table 4 indicates, most of the CTS scales were also intercorrelated. To better understand the relative importance of the relationships between the three CTS scales and the CAP abuse scale, a stepwise regression analysis was conducted across the three CTS scales using the CAP abuse scores as the dependent variable. In the regression analysis, the CTS violence scale was the only CTS factor that accounted for a significant (p < .0 l) amount of variance in the CAP abuse scores. The CTS violence scale, that measured the self-reported degree of childhood violence experienced, accounted for approximately 23% of the variance in the CAP physical child abuse scale. Using the CAP Inventory standard scoring procedure, the CAP Inventory abuse scale correct classification rates across the study groups were determined. Using the 215-point abuse scale cutoff score (Milner, 1986a), 87.7%, 73.3%, and 100% of the abusers, nonabusive comparison parents with a history of abuse, and nonabusive comparison parents without a history of abuse, respectively, were correctly classified. Using the alternate 166-point abuse

Child abuse screening

655

Table 4. Correlation Matrix for Conflict Tactics Scale (CTS) and Child Abuse Potential (CAP) Inventory Scales

CTS Scales

Reasoning

Verbal Abuse

Violence

CTS Verbal Abuse CTS Violence CAP PhysicalAbuse CAP Distress CAP Rigidity CAP Unhappiness CAP Problem withChild CAP Problem with Family CAP Problem fromOthers

-. 16 -.33** -.25** -.26** .06 -.23" -.02 -.25** -.20*

-.66* .25** .24* -.09 .15 .27** .34* .20*

--.48* .48* .02 .36* .35* .35* .43*

*p < .05; **p < .01; *p < .001.

scale cutoff score, that is based on signal detection theory (Milner, 1986a), 96.7%, 60.0%, and 83.3% of the abusers, nonabusive comparison parents with a history of abuse, and nonabusive comparison parents without a history of abuse, respectively, were correctly classified. All protocols were included in the determination of the abuse scale selectivity and specificity because the validity indexes were not available on the CAP Inventory used in the present study. Most research (e.g., Milner, 1986a, 1989a; Milner et al., 1986) indicates that meaningful improvements in the overall classification rates occur when the validity indexes are used to eliminate invalid protocols and the interpretation rules are followed (Milner, 1986a). It is not known how the classification rates reported here would have changed if the invalid protocols could have been removed.

DISCUSSION The CTS scores were congruent with the case history/interview data across the three study groups, confirming the subject placements with respect to a childhood history of abuse. Although these data support the reliability of the childhood history of abuse recall data, we do not know the degree to which the recall data were accurate. When the CTS and CAP abuse scores were compared, the simple correlations indicated relationships between each of the three CTS and the CAP Inventory physical child abuse scale. However, the regression analysis revealed that the CTS reasoning and verbal abuse scales did not account for any significant variance after the variance accounted for by the CTS violence scale was removed. The finding of a relationship between reported childhood violence and abuse scores is consistent with other studies that report a relationship between a childhood history of abuse and CAP abuse scores (Chan& Perry, 198 l; Milner et al., 1990). However, we report the first data that confirm this relationship in a sample that included physically abusive parents. The present findings also indicate a stronger relationship than has been previously reported. Milner et al. (1990) found an overall correlation of .29 between a childhood history of abuse and CAP abuse scores in a college sample compared to an overall correlation of .48 between the CTS childhood history of physical violence measure and the CAP physical child abuse scores obtained in the present study. The finding that the CTS childhood violence measure was related to parental physical child abuse potential is consistent with the view (e.g., Straus et al., 1980) that early social learning is related to the intergenerational transmission of violence. The data also indicate that the

656

J.A. Calisoand J. S. Milner

relationship may be relatively specific because only the experience of violence, rather than the experience of verbal abuse or the lack of parental reasoning, was related to adult physical child abuse potential. However, the fact that similar high levels of violence existed in the abusive mothers and nonabusive comparison mothers with a childhood history of abuse supports the view that an abuse history is not sufficient to adequately predict the occurrence of physical child abuse. Although the present study does not provide an explanation of what childhood or adult factors may moderate the transmission of violent behavior, it is evident from the analysis of CTS data presented in Table 2 that the abusers and nonabusive comparison parents that experienced childhood abuse each experienced similar levels of verbal abuse and parental reasoning. These data suggest that the degree of verbal assault and parental reasoning were not related to abuse or nonabuse comparison group membership for mothers who experienced childhood abuse. Thus for the present study, the CTS data did not reveal differences that might support the view that moderating effects may result from the increased use of parental reasoning in individuals who were abused but are not abusive. The CAP Inventory scales, however, revealed both similarities and differences between the study groups. The CAP abuse scores were different for each group with the abuse group earning the highest abuse scores, the nonabusive comparison group with a childhood history of abuse earning the second highest abuse scores, and the nonabusive comparison group without a childhood history of abuse earning the lowest abuse scores. The nature of the group differences is revealed through an analysis of the descriptive factor scores. The abusers and nonabusive comparison group that had a childhood history of abuse were similar in the degree of child problems and family problems with both their scores being higher than the nonabusive comparison group without a childhood history of abuse. In contrast, the rigidity and unhappiness scores differentiated the abusers from the nonabusive comparison subjects with a childhood history of abuse, and these scales were the only measures that were not different for the two nonabusive comparison groups. These findings indicate that nonabusive mothers with a childhood history of violence are less rigid toward their children and are happier than abusers with a childhood history of abuse; and, the nonabusive comparison mothers with a childhood history of abuse are similar on these measures to nonabusive mothers without a childhood history of abuse. This is the first report of personality differences in matched groups of abusive and nonabusive mothers who have similar levels of self-reported childhood violence. The data suggest that lack of parental rigidity and the presence of parental happiness n'lay be moderating characteristics or are the result of other variables that serve as buffering factors for the intergenerational transmission of abuse. It is important to note that the CAP rigidity factor measures child related rigidity (Milner, 1986a), that represents rigidity in the parent's attitudes toward the appearance and behavior of their children. This parental characteristic is evident in opinions that children should always be neat and clean, should be orderly in their behavior, should be obedient, should never cause trouble, should never disobey, should be seen and not heard, should be quiet and attentive, and should not talk back. A second rigidity theme involves beliefs that a home should be spotless with everything in its place. As they relate to abuse, these beliefs may be expressed through the forceful treatment of children to make them fit a rigid mold defined by the parent. The CAP unhappiness factor is described by a variety of happiness and unhappiness items with high scores indicating depression and difficulty in maintaining close adult relationships (Milner, 1986a). Thus, nonabusive comparison mothers with a childhood history of physical abuse were differentiated from abusers with a childhood history of abuse by their lower levels of child related rigidity and their higher levels of happiness, that included more satisfying intimate adult relationships.

Child abuse screening

657

The finding that satisfying intimate adult relationships are related to nonabusive group membership in those with a childhood history of abuse is congruent with the conclusion of Egeland et al. (1988) that mothers who break the cycle of violence have emotionally supportive relationships with mates. Egeland et al. (1988) also noted that these mothers received support from a caring adult during childhood. Although data on supportive childhood relationships are not reported in the present study, it is interesting to note that Milner et al. (1990) found an inverse relationship between CAP abuse scores and a caring adult or a caring friend during childhood. Inspection of the classification rates for the standard scoring procedure using two established cut-offpoints of 166 and 215 (Milner, 1986a) reveals that the CAP abuse scale classification rates are affected by a history of childhood abuse. For the 215-point cut-off score, the correct classification rates (i.e., 87.7%) for all protocols for the physical child abusers are superior to those previously reported using the standard CAP scoring procedure (Milner, 1989a). In a previous study (Milner, 1989a) comparable high rates for abusers were only achieved after the validity scales and interpretation rules were used. Although in the present study the abusers were correctly classified more often, the correct classification rates (i.e., 73.3% for nonabusive comparison mothers with a childhood history of abuse and 100% for nonabusive comparison mothers without a childhood history of abuse) were better for nonabusers without a childhood history of abuse and lower for nonabusers with a childhood history of abuse than general population rates previously reported. Correct classification rates for general population parents have typically been in the 90% range. Assuming that the nonabusive comparison group with a childhood history of abuse contained no undetected physical child abusers, the present data indicate that more false-positive classifications will occur when the CAP Inventory is used with nonabusive comparison mothers with a childhood history of abuse. The data also indicate that fewer misclassifications will occur if the higher abuse scale cut-off score (i.e., 215) is used. Thus, only the 215 cut-off score should be used in cases where a childhood history of abuse is known to exist. It is also evident that a new (higher) CAP abuse scale cut-offscore should be considered for use in cases where a childhood history of abuse is reported to determine if a higher cut-off score would reduce the false-positive classification rate without dramatically increasing the false-negative classification rate. The determination of a new cut-off score would require the testing of additional nonabusive parents with a childhood history of abuse and use of the CAP validity scales to eliminate those subjects who might be faking good and faking bad. Possibly important for differential classification purposes are the findings that the rigidity and unhappiness factor scales differentiated abusive and nonabusive comparison parents with a childhood history of abuse. These findings suggest that for parents with a severe history of childhood abuse who earn elevated abuse scores, the rigidity and the unhappiness factor scales may be useful in a secondary client evaluation because the nonabusive comparison group with a childhood history of abuse earned lower scores on these scales, and the scores were similar to nonabusive comparison parents without a history of abuse. Thus, these factor scales may have utility in reducing the number of false-positive classifications in nonabusive parents with a childhood history of abuse in cases where the overall abuse score is elevated. As is often the case in family violence research, we must mention several additional caveats related to the present study findings. Additional research is needed before the present findings can be used in clinical practice, including the investigation of abusive subjects without a childhood history of abuse. The study is retrospective in nature and the use of self-report childhood history of abuse data that are obtained years after the the abuse is problematic without independent confirmation. However, the fact that similar data were obtained from two different childhood history of abuse assessments (i.e., case history/interview data and CTS data) conducted at different times suggests that the recall of the childhood abuse was

65g

J.A. Caliso and J. S. Milner

reliable on a g r o u p basis. A n o t h e r limitation was that the study samples were c o n v e n i e n c e samples with n u m e r o u s study inclusion/exclusion criteria. A l t h o u g h a strength o f the present study was the use o f three demographically m a t c h e d groups, the d e m a n d s o f m a t c h i n g c o m bined with the various criteria for study inclusion/exclusion and the use o f only female volunteer parents limits the meaningfulness a n d the generalizability o f the study findings. Thus, replication o f the present study results is needed with other child abuse groups with a n d without a childhood history including those w h o do not a d m i t the abuse and male physical child abusers. In conclusion, the present findings suggest specific cognitive a n d affective factors m a y be i m p o r t a n t in reducing abuse potential in abusive parents with a c h i l d h o o d history o f abuse. Cognitive treatments that focus on rigid thinking patterns related to children's b e h a v i o r and skill training that increases the likelihood o f the parent developing a n d maintaining intimate adult relationships m a y decrease the abuser's potential for child abuse.

REFERENCES Altemeier, W. A., O'Conner, S., Vietz, P. M., Sandier, H. M., & Sherrod, K. B. (1982), Antecedents of child abuse. Journal of Pediatrics, 100, 823-829. Baldwin, J. A., & Oliver, J. E. (1975). Epidemiology and family characteristics of severely abused children. British Journal of Preventive Social Medicine, 29, 205-221. Berger, A. M. (1980). The child abusing family: II. Child and child-rearing variables, environmental factors and typologies of abusing families. American Journal of Family Therapy, 8, 52-68, Chan, D. A., & Perry, M. A. ( 1981). Child abuse, discriminating factors toward a positive outcome. Paper presented at the biennial meeting of the Society of Research in Child Development, Boston, MA. Daniel, J. H., Hampton, R. L., & Newberger, E. H. (1983). Child abuse and accidents in black families: A controlled comparative study. American Journal of Orthopsychiatry, 53, 645-653. Egeland, B., Jacobvitz, D., & Sroufe, L. A. (t 988). Breaking the cycle of abuse. Child Development, 59, 1080-1088. Fontana, V. J., & Bernard, M. L. (1971). The maltreated child. Springfield, IL: Charles C. Thomas. Friedrich, W. N., & Wheeler, K. K. (1982). The abusing parent revisited: A decade of psychological research. Journal of Nervous and Mental Disease, 170, 577-587. Heifer, R. E. (1987). The developmental basis of child abuse and neglect: An epidemiological approach. In R. E. Helfer & R. S. Kempe (Eds.), The batteredchild(4th od.) (pp. 60-80). Chicago, IL: The University of Chicago Press. Heifer, R. E., & Kempe, R. S. (Eds.). (1987). The battered child (4th od.). Chicago, IL: University of Chicago Press. Johnson, B., & Morse, H. A, (1968). Injured children and their parents. Children, 15, 147-152. Martin, H. P. (1976). The abused child. Cambridge, MA: Ballinger Publishing. Melton, G. B., & Limber, S. (1989). Psychologists' involvement in cases of child maltreatment: Limits of role and expertise. American Psychologist, 44, 1125-1233. Milner, J. S. (1986a). The Child Abuse Potential Inventory: Manual (2nd ed.). Webster, NC: Psytec Corporation. Milner, J. S. (1986b). Assessingchild maltreatment: The role of testing. Journal of Sociology and Social Welfare, 13, 64-76. Milner, J. S. (1989a). Additional cross-validation of the Child Abuse Potential Inventory. Psychological Assessment: A Journal of Consulting and Clinical Psychology, 1, 219-233. Milner, J. S. (1989b). Applications and limitations of the Child Abuse Potential Inventory. In J. T. Pardeck (Ed.), Child abuse and neglect: Theory, research, and practice (pp. 83-95). New York: Gordon and Breach Science Publishers. Milner, J. S. (1990). An interpretive manual for the Child Abuse Potential Inventory. Webster, NC: Psytec Corporation. Milner, J. S. ( 199 la). Medical conditions and Child Abuse Potential specificity. PsychologicalAssessment. A Journal of Consulting and Clinical Psychology, 3, 208-212. Milner, J. S. (1991b). Additional issues in child abuse assessment. American Psychologist, 46, 80-8 I. Milner, J. S., Gold, R. G., Ayoub, C., & Jacewitz, M. M. (1984). Prediction and explanation of child abuse: Cross-validation of the Child Abuse Potential Inventory. Journal of Consulting and Clinical Psychology, 54, 865-866. Milner, J. S., Gold, R. G., & Wimbedey, R. C. (1986). Predictive validity of the Child Abuse Potential Inventory. Journal of Consulting and Clinical Psychology, 54, 865-866. Milner, J. S., & Robertson, K. R. (1989). Inconsistent response patterns and the prediction of child maltreatment. Child Abuse & Neglect, 13, 59-64. Milner, J. S., Robertson, K. R., & Rogers, D. L. (1990). Childhood history of abuse and adult child abuse potential. Journal of Family Violence, 5, 15-34.

Child abuse screening

659

Milner, J. S., & Wimberley, R. C. (1979). An inventory for the identification of child abusers. Journal of Clinical Psychology, 35, 95-100. Milner, J. S., & Wimbedey, R. C. (1980). Prediction and explanation of child abuse. Journal of Clinical Psychology, 36, 875-884. Straus, M. A., & Gelles, R. J. (Eds.). (1990). Physical violence in American families. New Brunswick, N J: Transaction Publishers. Straus, M. A., Gelles, R. J., & Steinmetz, S. K. (1980). Behind closed doors. New York: Doubleday. WMserman, S. (1967). The abused parent of the abused child. Children, 14, 175-179. Widom, C. S. (1989). Does violence beget violence? A critical examination of the literature. Psychological Bulletin, 106, 3-28.

R ~ n m ~ - B i e n que les antb~kients de maltraitance au cours de l'enfance soient li6s ~ la maltraitance parentale, beaucoup de parents ayant ~ 6 eux-m~mes des enfants maltrait6s ne maltraitent pas pour autant leurs enfants. Pour d6terminer I'impact d'ant6c&lents d'abus sur le potentiel de violence d'un parent ~ l'6gard de son enfant, deux 6chelles dont le CTS modifi6 (Conflict Tactics Scale) et le CAP (Child Abuse Potential Inventory) ont 6t6 test6es sur des groupes apafi6s de ,~'res physiquement abusives avec des ant6cbtents de maltraitance au cours de l'enfance, des m~res nonabusives avec les m~mes ant6c6dents et des m~res nonabusives sans ant6c6dents de maltraitance au cours de l'enfance. Le CTS modifi6 donne des renseignements sur les 6v/~nements marquants de l'enfance e t a 6t6 utilis6 pour confirmer les ant6c6dents de maltraitance. Comme pr6vu, les 6chelles CTS verbales et de violence ~taient plus 61ev6es pour les m~res abusives que pour les m~res nonabusives, ayant des ant6c6dents d'abus au cours de l'enfance. Aucune 6chelle CTS ne difl'6rait pour ies m~res abusives et nonabusives avec des ant6c6dents d'abus au cours de l'enfance. ?alsles scores d'abus du CAP difl'6rentiaient bien les trois groupes d'6tudes. Sur l'6cheile CAP, les seuls facteurs discri,inatifs entre les m~res abusives et nonabusives, ayant des ant6c6dents d'abus au cours de l'enfance, 6taient la rigidit6 et le malheur. Les m~res nonabusives, ayant 6t~ maltraitb~s au cours de l'enfance 6taient moins figides clans leurs attentes et 6taient plus heureuses dens leurs relations ~tautrui que les m~res, ayant 6t6 maltrait6es au cours de l'enfance. ltemmmc~---A pesar de que un historial de abuso en la infancia est~ relacionado con abuso en la infancia parental, muchos padres con historial de abuso no son abusivos. Para determinar los efectos de tener un historial de abuso en la infancia en el potencial de abuso al menor del adulto, se le administr6 el "Conflict Tactics Scale" (CTS) modificado y el "Child Abuse Potential Inventory" (CAP) a grupos pareados de madres fisicamente abusivas con un historial de abuso en la infancia, y madres no abusivas sin historial de abuso en la infancia como control. El CTS Modificado preguntaba sobre eventos de la infancia y se usaba para confirmar el historial de abuso. Como era de esperarse, el CTS verbal y Ins escalas de violencia result6 mayor para las madres abusivas y no abusivas con un historial de abuso en la infancia. Ninguno de los puntajes de ia escala CTS fueron diferentes tanto en las madres abusivas como no abusivas con un historial de abuso en la infancia. Por el contrario, los puntajes de abuso del CAP diferenciaban entre los tres grupos del estudio. Sin embargo, en las escalas factoriales del CAP, solo los factores de rigidez e infelicidad discriminaron entre madres abusivas y no abusivas con un historial de abuso en la infancia. Las madres no abusivas con un historial de abuso en la infancia eran menos rigidas en sus expectativas infantiles y eran m~s felices en sus relaciones interpersonales que las madres abusivas con un historial de abuso en la infancia.

Childhood history of abuse and child abuse screening.

Although a childhood history of abuse is related to parental child abuse, many parents with a history of abuse are not abusive. To determine the effec...
1MB Sizes 0 Downloads 0 Views