C~ildAbuse&Neglect,Vol, 16, pp. 661-671, 1992

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

Printedin the U.S.A.All rightsreserved.

SUBSTANTIATION OF REPORTED CHILD ABUSE OR NEGLECT: PREDICTORS AND IMPLICATIONS HELEN R. WINEFIELD Department of Psychiatry, University of Adelaide, South Australia

PATRICK

W.

BRADLEY

Program Planning Division, South Australian Department for Family and Community Services, Adelaide, South Australia

Abstract--Underreporting and overreporting of suspected child abuse and neglect cases reduce the efficiency of child protection services. We used all the reports in South Australia for 1988-1989 (N = 3,228) to study the determinants of the decision by child protection workers to register a reported incident as being one of child abuse and neglect. Logistic regression showed that registration (substantiation) was predictable from two variables: the age of the alleged victim and the caseworker's estimate of severity. This latter variable needs investigation, as despite its crucial role we have no information on how caseworkers form a judgment about severity.

Key Words--Child abuse and neglect, Child protection services, Mandatory reporting.

PRESSURE HAS INCREASED for agencies to undertake systematic evaluation of their procedures as a result of increased public attention to, and funding for, child protection. The ethical and practical difficulties in gathering relevant information are, however, serious (Graham, Dingwall, & Wolkind, 1985). Confidentiality issues make access to ongoing cases problematic (Kinard, 1985). As a first step to evaluating the outcome of child protection services, we therefore used computerized records to study the process by which reports are or are not substantiated. In so doing, we aimed to contribute information to the current debate about under- and overreporting. "Underreporting" refers to the failure, usually by professionals with mandatory obligations, to report suspicions of child abuse and neglect to the relevant authorities. The introduction in many states in Australia and America of mandatory notification was intended to ensure that professionals such as physicians, teachers, and therapists would pass on their suspicions of child abuse to specialists in child protection. At the same time, there has been considerable concern about negative consequences of mandated notification--that it breaks the confidentiality of the professional-client relationship, that it will overload welfare authorities with ill-informed reports from inadequately trained sources, and that professionals will Research funded by a grant to the authors from the Channel 7 Children's Research Foundation in South Australia. Part of the data in this paper was presented at the 12th Congress of the International Association for Child and Adolescent Psychiatry and Allied Professions in Kyoto, Japan, July 17, 1990. Received for publication September 29, 1990; final revision received October 28, 199 l; accepted November 4, 199 I. Requests for reprints may be sent to Dr. H. R. Winefield, Department of Psychiatry, University of Adelaide, GPO Box 498, Adelaide, Australia 5001. 661

~62

H. R Winetieldand P. W Bradley

continue to report only when they choose to because the sanctions against failing to report cannot or will not be enforced (Butz, 1985; Nelson, Dainauski, & Kilmer, 1980). Underreporting has been discovered in teachers (Lombard, Michalak, & Pearlman, 1986), general practitioners (Winefield & Castell-McGregor, 1986), clinical psychologists (Stadler, 1989; Winefield, 1988), mental health clinicians (Kalichman, Graig, & Follingstad, 1988), residential care workers (Rindfleisch & Bean, 1988), and nonprofessionals (Dukes & Kean, 1989). Uncertainties over the definition of abuse, and fears of the impact of the notification upon the family or upon the family's relationship to the notifier, have commonly been found to cause hesitation. Zellman (1990), for example, took a large stratified random sample of U.S. mandated notifiers and asked them about reasons for reporting and for failing to report suspected cases of child abuse. While 44% of her 1,196 respondents indicated that they always reported suspected cases, the second commonest pattern (33% of respondents) was discretionary reporting. The most frequently endorsed reason for failure to report was lack of sufficient evidence (related to obscurity in defining what counts as abuse and what counts as reasonable suspicion), followed by fears that the report would disrupt or interfere with treatment. If professionals' fears of inadequate services and family trauma as consequences of reporting are unfounded, potential notifiers need more education. On the other hand, if their fears are well-founded, then the child protection service's procedures need modification. A quantitative analysis of intervention outcomes in a large sample of child protection cases offers the possibility of settling such questions. To date, no such study has, to our knowledge, been published, and the present paper reports the first stage of such a project. "Overreporting" refers to allegations of child abuse or neglect being found upon investigation to be unsubstantiated. Fears have been expressed that such cases, whether due to ignorance or malice on the part of the reporter, clog the child protection services and impede their effective response to genuine cases, and also cause unnecessary distress and loss of privacy for affected families (Carment, 1990; Freeman, 1983; Newberger, 1983). The other charge is that, due to their visibility in the welfare system, disadvantaged social groups will be unfairly targeted as subjects of abuse and neglect reports, adding to their burdens of stress and discrimination. In New York, professional reports were more likely to be substantiated than nonprofessional, especially for charges of neglect (Eckenrode, Powers, Doris, Munsch, & Bolger, 1988). Anonymous reports of neglect were substantiated at the lowest rates of all, but anonymous reports of abuse were substantiated at the same rate as those of unrelated laypersons. Thus, it may be hasty for anonymous reports to be automatically discounted, such as by Barone, Adams, and Tooman (1981). Zuravin, Watson, and Ehrenschaft (1987) found that more anonymous than professional reports were unsubstantiated in Baltimore, mostly because they referred to less serious incidents. Substantiation itself may be more a reflection of the caseworker's expectations than of what actually happened (Kotch & Thomas, 1986). Support for this perspective came from Eckenrode et al. (1988), who found that different substantiation rates for different sources of report were associated with different frequencies of undertaking court action. However, they could not tell from their data whether these associations were related to features of the abuse such as its severity, or to the caseworkers' tendency to give more credibility to reports from professional sources. Finkelhor (1990) noted a rise in the proportion of substantiated reports, from 43% in 1980 to 53% in 1986. Equivalent definitions and sampling procedures were used at the two times, by the Second National Incidence Study. In reply, Besharov (1990), who maintains that the U.S. national rate of unfounded reports is 60-65%, offered explicit guidelines for professionals about what evidence should count as "reasonable" grounds for suspicion of child

Substantiation of reported child abuse

663

abuse, and which cases should be automatically rejected for consideration by the well-trained intake workers, who are essential. Underreporting and overreporting decrease the efficiency of child protection services. In South Australia, in common with many other places in the world, official notifications of suspected child abuse or neglect may be made to the child protection service by any member of the community. In addition, a wide range of professionals who come into contact with children are required by law to report suspected child abuse. There is evidence of significant professional underreporting (Winefield, 1988; Winefield & Castell-McGregor, 1986). Our first goal in the present study was to measure the extent of any overreporting in South Australia by examining the frequency with which notified cases were found to be unsubstantiated upon investigation. Second, we aimed to compare substantiation rates across sources of report. The connection between source of notification and outcome (in the sense of substantiated or not) is of special interest where, as in South Australia, the net of mandated notifiers is wide. Are or should all reports be taken equally seriously by the welfare workers? Third, we sought to explore outcome according to reported type of abuse and victim characteristics (age, sex, ethnicity, and social status). If high proportions of unsubstantiated neglect reports were received about poor or minority families, for example, it could be taken as evidence of the kind of discrimination feared by those opposed to mandatory reporting. Fourth, it was hoped that findings would clarify the education and training needs of those required by law to report suspected child abuse and neglect.

METHOD Data Base

The Department for Community Welfare (DCW) was the statutory body with responsibility to devise and implement child protection policies in South Australia. Modifications to the computerized record system for child abuse cases were introduced by DCW in July 1988, which therefore formed the beginning of our 12-month data collection period. During 19881989, 3,228 separate incidents of alleged child abuse were reported to the DCW in South Australia, of which 2,366 cases (73.3%) fell within the metropolitan area, which is where 74% of the population lived at the time according to census data. There were no significant differences between country and metropolitan victims in the percentage of cases that were substantiated, or in the age or sex of victims or the percentage who had been the subject of previous notifications. More country reports referred to aboriginal children, because more aboriginal people live outside the metropolitan area, but there was no relationship between aboriginality and likelihood of substantiation. There were some differences between metropolitan and rural areas in the source and type of reports. Further analysis has been confined to incidents reported to metropolitan offices, because we wished to geographically restrict the casenotes read in the second phase of our project, to be reported subsequently, which aimed to study the impact of the report on the child and family. Once a notification of suspected child abuse or neglect has been received by DCW, caseworkers may (a) register the case as one of child abuse or neglect which amounts to substantiation of the report, (b) refer the child for help with a nonabuse problem, or (c) close the case. One of these actions should be taken by the end of 28 days after the original receipt of the allegation, on the basis of investigations during the interim. Registration of the case is the equivalent of a substantiation of the suspected abuse; the other two responses constitute nonsubstantiation of the alleged abuse.

664

H.R. Winefield and P. W. Bradley

Table 1. Percentage of all Reports that Came from Each Source, and Percentage of Reports from each Source that Were Registerd Source

% of Reports

% Registered

Victim Police Social workers Teachers

6.6 6.0 8.9 20.9

83.2 75.7 70.7 62.4

Paramedics Anonymous Doctors Relatives Nonprofessional Nurses

6.3 2.1 4.7 25.5 17.6 1.3

(DCW worker, hospital or other social worker) (Teacher, student counselor, day-care giver, kindergarten worker) (Psychologist, paramedic, other health/welfare) (Hospital or other doctor) (Maltreater, sibling, parent, guardian, other relative) (Friend, neighbor) (Hospital or CAFHS or other nurse)

55.4 53.8 52.8 50.8 4 |. 1 30.4

The criteria for registration/substantiation in South Australia are that there is acceptable evidence that the child has been abused or neglected. The abuser need not be known, and the child need not require ongoing protective services, nor must the type of abuse necessarily be that of the original allegation. The evidence need not be as conclusive as that needed in court, but should "be conclusive enough to result in a professional judgment that harm to the child has occurred," according to the 1988 DCW manual of standard procedures. The computerized records contained little information about the grounds on which the decision was reached, except for two pieces of information collected at the same time as the registration decision was recorded, in other words, after up to 28 days of investigation after the initial report: (a) "tick-the-boxes" details of up to six signs of abuse/neglect, and (b) the worker's rating of the severity of the abuse/neglect, categorized as minor, moderate, severe, or life-threatening.

RESULTS The overall substantiation rate, that is, cases that were registered as cases of child abuse or neglect, was 56.3%; 13.2% were not registered but were referred on with other problems and 30.6% were closed without registration or referral (figures rounded to one decimal place). In the analyses that follow, numbers vary somewhat due to the exclusion of cases with missing data, which was more common for some variables than others.

Sources and Types of Abuse Notifications The sources of notification can be seen in Table 1. Categories can be summarized as lay, 51.8%, medical and social work, 21.2%, teachers, 20.9%, and police 6.0%. Reports alleged physical abuse in 40.9% of the cases, sexual abuse, 28.5%, emotional abuse, 6.6%, and neglect, 24%.

Victim Characteristics The age of children in these reports averaged 7.5 years. Victim age differed significantly by analysis of variance according to type of abuse, F(3, 2374) = 48.03, p < .001. Pairwise comparisons using the Newman-Keuls post-hoc test with significance level set at .05 showed that the reported victims of neglect were the youngest group at average age 5.46 years, and victims of sexual abuse were the oldest at average age 8.47 years. Reported victims were 53.6% female; 29.6% had been the subject of a previous notification,

Substantiation of reported child abuse

665

and ethnic background was recorded as aboriginal for 5.1%, anglo/Australian for 87.6%, and non-English-speaking background for 7.3%. The head of the household was employed in 46.2% of the cases, with 83.7% in blue-collar occupations, but the large amount of unavailable information on these two variables (over half the data missing) reduces the reliability of analyses including them.

Results of Investigations by ChiM Protection Workers Signs of physical injury, including bruising, fractures, and burns, were found in 24.1% of cases, while 15.9% had no visible injury. Definite signs of sexual abuse were found in 22% of cases, of emotional abuse in 6.1%, and of neglect in 18.4% of the cases. In 45.3% of cases, the severity of the abuse/neglect was judged to be minor, 39.9% moderate, 13.7% severe, and 1.2% was life-threatening.

Effect on Registration Decision of Source of Report and Type of Abuse Registration was significantly related to source of report, ×2(9) = 109.30, p < .001. As shown in Table 1, registration occurred at higher levels than the average when reports came from victims, police, social workers, or teachers; reports were least likely to result in registrations when they came from nurses and unrelated nonprofessionals. Anonymous reports were substantiated at comparable rates to those for doctors and relatives. There was no significant difference in the proportion of cases registered according to type of abuse/neglect, ×2(3) = 11.34, p < .01. Teachers showed a different pattern again, being more likely to have reports of physical abuse and neglect registered, and reports of sexual and emotional abuse unsubstantiated, ×2(3) = 15.21, p < .01.

Victim and Family Character&ticsand Reg&tration Children whose cases ended up as registered were older than children whose cases were closed: average ages 8.16 and 6.48 years, respectively, t(2,001) = 7.81, p < .001. Having been the subject of a previous notification, increased the chances of the case being registered rather than closed, X2(2) = 17.12, p < .001, but sex, ethnicity, and social status were not associated with the overall likelihood of registration. Households where the caregiver had a white-collar job had a lower registration rate of physical abuse cases and a high rate for the few cases of emotional abuse and neglect, x2(3) = 11.91, p < .01.

Severity Judgments and Registration The 32 categories describing signs of abuse were difficult to combine meaningfully, as they varied in explicitness and in implied objectivity of the evidence, and thus cannot be used in further analyses. There was a significant relationship between judged severity and the likelihood of registration, x 2(2) = 228.61, p < .001, with 46.2% of minor cases and 87.5% of severe or life-threatening cases being registered.

Multivariate Prediction of Report Substantiation As the predictor variables (age, source, and type of report) were related to each other as well as to the dependent variable, the univariate analyses described above were difficult to integrate to form an overall picture. Multivariate analysis was therefore undertaken to remove the effects of this collinearity, and sought to predict outcome (registration) as parsimoniously as possible. Stepwise logistic regression is the statistical procedure of choice where, as here, the goal is prediction of a dichotomous outcome (registered or not) using independent variables

t~66

H R. Winefield and P. W. Bradley

Table 2. Prediction of Registration by Logistic Regression, in Part (a) Using Only Variables Known or Potentially Known at the Time of the Report, and in Part (b) Using Age and the Caseworker's Judgment of Severity Variable

Regression Coefficient

Standard Error (SE)

(Coeffacient/SE)

Ratio

0.05 -0.01 - 1.23 1.81 0.01 0.11 -0.16 0.23 0.48

0.02 0.23 0.28 0.40 0.03 0.03 0.04 0.07 0.17

2.84 -0.06 -4.34 4.50 0.56 3.88 -3.99 3.34 2.85

0.26 0.95 0.08 0.43

0.10 0.14 0.01 0.12

2.73 6.91 6.32 3.69

Part (a)* Age Source a Source b Source c Age x source a Age × source b Age × source c Previous notification Constant

Part (b) Severity ~ Severity b Age Constant

* Source effects are comparisons against the d u m m y variable lay source. Source: "medical and social workers; bteachers; a n d Cpolice.

that may not be normally distributed, and many of which are categorical rather than continuous. Variables with a known relationship to registration, and that had reasonable levels of available data, were entered into logistic regression analyses. These possible predictors were source of report (recoded to four categories to avoid empty cells), type of abuse/neglect reported, the alleged victim's age, sex, ethnicity, and previous notification, and the worker's severity judgment. As the severity judgment was more of a conclusion based upon the investigation process, while the other variables were more truly predictors, being known or potentially known at the time of the report, in the first logistic regression analysis severity was excluded to study outcome prediction based on initial knowledge only. Logistic regression using only variables known or potentially known at the time of the report showed that registration could be predicted on the basis of victim age, previous notification, source of report, and the interaction between age and source of report (See Table 2(a)). The likelihood of registration was higher for older children; children who had been the subject of a previous notification; those reported by police, especially if of younger age; and lower for those reported by teachers, especially if younger. The logistic regression formula fitted the data well, as shown by Hosmer-Lemeshow x 2(8) = 4.17, p = . 84; C. C. Brown X2(2) = .27, p = .88. In a second regression, severity was added, and it displaced several of the other variables: F(severity) = 50.16, F(age) = 21.17, F(previous notification) = 3.48, F (source) = 2.70; the latter two terms failed to reach statistical significance. As before, sex, ethnicity, and type of abuse did not gain entry to the regression equation. A third and final regression, therefore, used as predictors only age and severity and their interaction, and the last term did not enter the equation. Results are shown in Table 2(b) (the ratio of the regression coefficient to its standard error may be read as approximately a t statistic). The obtained Hosmer-Lemeshow ×2(8) = 8.27, p = .41, and C. C. Brown X2(2) = .18, p = .91, indicate that the data fitted the logistic model well. As illustrated by Figure l, the probability of registration of a reported case

667

Substantiationof reported child abuse -o- min severity .41. rnod severity 41- high severity

1.0

0.8

~-

0.6

o.

0.4

0.2

I

0

10

I

20

age Figure 1. Prediction of registration decisions by age and severity (n = 1629).

increased with the age of the alleged victim, and was also higher for moderate and severe cases than for those rated as minor in severity.

DISCUSSION This paper reports analyses of the computerized records of all the child abuse and neglect notification made to South Australian child protection authorities over a 1-year period. It examines the frequency and the correlates of the case being registered as one of child abuse or neglect. Subsequent work on this project will involve manual searches of casenotes for more detailed information on the process of decision making and the outcomes for the family of the report. Unfortunately we can only present and analyse data for reported cases, while being aware that a high proportion of abuse and neglect cases are never reported. The considerable amounts of missing data, which we discovered in the archival computer record, reduced the research utility of that data base. This common problem increases the pressure on researchers to conduct time-consuming file searches or face-to-face interviews with clients, both of which pose greater ethical difficulties. It would therefore be advantageous for outcome measurement, if computer data were comprehensive as well as relevant. The overall substantiation rate of 56% in South Australia in 1988-1989 is similar to that for the second National Incidence Survey in America (53%) (Barone, Adams, & Tooman, 1981), and higher than that found in two other locations where mandatory reporting occurs, namely New York (Eckenrode et al., 1988), and New South Wales (Carment, 1990). No conclusions can be drawn about the effects of the legal obligation to report per se, as such provisions were in force throughout the survey period and for several years beforehand; during this time reporting rates have risen dramatically as the subject of child abuse gained higher community awareness. Nearly 33% of cases not registered as child abuse were judged to need some form of intervention and were referred to welfare services other than child protection. In the metropolitan cases studied here, source of report affected substantiation probability. Four out of five victim reports, and three out of four police reports, were substantiated; after these two groups came social workers and teachers, then other nonprofessional and anony-

668

H . R . Winefield and P. W. Bradley

mous reports, with nurses having the lowest substantiation rate. It is thus misleading to contrast the substantiation rates of professional reporters, who have the legal obligation to report suspected cases, with nonprofessionals, who do not. Those categories are too broad. Again, some sources were more accurate or more credible, according to what type of abuse they reported: nonprofessional reports were most likely to be registered when they concerned alleged emotional abuse and neglect, unless the sources were relatives of the victim, when registration probability was greatest for physical and sexual abuse. Teachers were most credible about physical abuse and neglect. It may seem feasible that these differences in registration rates are related to the opportunities that members of different groups have to observe, not only the physical signs of abuse, but the interactions between victims and abusers. If, as seems likely, it is distressing for families to undergo investigations for alleged abuse, the goal of reducing unsubstantiated reports yet further would seem desirable. However, it follows from the principles of signal detection theory that the more "true positives" are detected, the more "false alarms" will also be reported (Winefield, 1988), if decision-making rules remain the same. Education for mandated notifiers about their responsibilities must therefore include information about what levels of maltreatment go beyond that acceptable in our society, and constitute registrable abuse. Decision rules need to become more sensitive, or more congruent with those of child protection workers, rather than increases in the number of reports being the sole aim of training. In this manner, reporters could reduce their "false alarm" rate as well as reduce their "false negatives" (failures to report registrable cases). Law enforcement agency reports had high substantiation rates, as had been found before (Groeneveld & Giovannoni, 1977). Possibly the cases that are brought initially to police, or that the police decide to report, are more serious than those reported by other groups; possibly workers assume them to be more serious. Yet another possibility is that caseworkers assume the police to be highly trained in investigation and to do it thoroughly; a related hypothesis that deserves research is that worker training levels may affect the probability of registration. In any case, high substantiation rates result from stringent criteria at the cost of nonreport of borderline cases, some of which would be shown upon investigation to be registrable. Decision rules cannot be formulated without making value judgments about the relative costs and benefits, for the victims and others, of accurate and inaccurate reports. These matters need community debate and, if possible, consensus. It is unrealistic to expect culturally normative and firmly entrenched attitudes about children as the property of their parents, to disappear quickly, especially as such disappearance threatens those with greater social power than children (Kaye & Winefield, 1988). The possibility of conflict between the interests of adults and of children suggests the need for a children's advocacy body that is independent of the welfare system but that can review its operations and investigate complaints. The fact that almost half of even professional reports are not substantiated raises several questions. Are the criteria for substantiation too demanding (e.g., the definition of "acceptable" evidence made too stringent), or are there inadequate efforts to achieve substantiation, or is there a high proportion of inappropriate or inaccurate allegations? As suggested by Besharov (1990), intake workers need to be highly trained to screen out clearly unsuitable reports. In addition, the potential reporters (both mandated and not) need extensive training and feedback about which reports are appropriate. Few systematic evaluations of training programs for mandated notifiers have been published, and this issue clearly needs high quality research efforts. Research that provided useable feedback to workers and policy planners could justify the dedication of resources not only to itself but also to child protection efforts as a whole. Older children, and those who have been the subject of previous notification, were more

Substantiation of reported child abuse

669

likely to be registered. In white-collar households, emotional abuse and neglect were more likely to be registered, perhaps because the few reports concerned extremely obvious cases. There was no evidence for the claim that unsubstantiated neglect reports are more common in relatively powerless groups. Unfortunately, lack of substantiation/registration does not necessarily mean that the welfare workers felt no abuse had occurred. It might also mean that there were not enough workers available to investigate allegations accorded low priority. Staff shortages, in turn, are related to inadequate funding and supervision, and to high turnover and low morale. Finkelhor (1990) has compared the child protection and criminal justice systems, which have similar "substantiation" rates. The public tolerates the inefficiency and intrusions of the criminal justice system, because people place high priority on crime control. Finkelhor argues that child protection, which is also endorsed as important by the public, deserves more funding and more public esteem rather than less. Putting together all the predictor variables and statistically removing their intercorrelations, as was achieved through the use of the logistic regression technique, allowed prediction of registration from just two pieces of information: the age of the child and the worker's judgment of the severity of the abuse/neglect. This means that the effects of the other variables that were related to registration when considered one at a time were summarized within the age and severity factors. The age of the child may be influential because of the greater facility of verbal expression in older children, and the correspondingly fewer difficulties in collecting information from them about the abuse. Preconceptions about the greater reliability of older children's memory may also be operating; these are not necessarily accurate if interviewing is skilled (Loftus & Davies, 1984). Another possibility is that workers understimate the vulnerability of young children to abuse and neglect. The worker's severity rating is a more mysterious variable, in that we really have no understanding of how such judgments are arrived at, or how much consistency there would be between different workers in reaching a judgment about the same case. To illustrate the apparent inconsistencies in how ratings were applied; verbal abuse was ticked as the most serious sign of abuse in 32 cases. Six of these attracted a minor severity rating, 19 were moderate, six severe, and one life-threatening. Four of the five cases with skull fracture were rated as moderate in severity. The severity rating presumably represents a global judgment, which takes account of many contributory pieces of evidence, perhaps including intentionality. Although it was intended to reflect the severity of the injury, another possibility is that it functionally reflects the worker's assessment of the case's registrability. There is no evidence that workers' ratings of severity after 1 month of investigation validly predict the impact of the abuse/neglect upon the victim's development, which is now realized to be complex (Augoustinos, 1987; Wolfe, 1987). As the caseworker's rating of the severity of the abuse/neglect has been shown here to be critically important to whether the notification is substantiated or not, we need to understand better how this judgment is formed, and to what extent it is congruent with official definitions and policies in that area. Worker training needs to be explicit and its effectiveness monitored. In this as in other areas of the child protection endeavor, the need for systematic quantitative research is clear. In conclusion, our study of computerized records of a complete 12 months of reported child abuse and neglect in the main city of South Australia provides important information about the process of reporting by lay and professional notifiers, and also about the process of decision making in the caseworkers. The ratio of substantiated to unsubstantiated cases is similar in this state to that in comparable child protection systems, with no evidence of overreporting selectively affecting those who are more "visible" to welfare authorities. The likelihood of registration does vary by source of report, but a simple division into lay and professional

~)7 0

H. R. Winefield and P. W. Bradley

report sources is not helpful. Registration probability also varies by victim age and previous notification, although not by sex, ethnicity, or type of alleged abuse. The strong association between registration and caseworker conclusions about severity of the case means that the latter variable deserves close scrutiny, and has important implications as do many of the other findings, for worker training as well as for the education of mandated notifiers.

REFERENCES Augoustinos, M. (1987). Developmental effects of child abuse: Recent findings. ChildAbuse & Neglect, 11, 15-27. Barone, N., Adams, W., & Tooman, P. (1981). The screening unit: An experimental approach to child protective service. Child Welfare, 60, 198-204. Besharov, D. J. (1990). Gaining control over child abuse reports. Public Welfare Spring Issue, 34-40. Butz, R. A. (1985). Reporting child abuse and confidentiality in counseling. Social Casework, 66, 83-90. Carment, A. (1990). Mandatory notification of child abuse and neglect. Australian Journal of Social Issues, 25, 15-26. Dukes, R. L., & Kean, R. B. (1989). An experimental study of gender and situation in the perception and reportage of child abuse. ChiMAbuse & Neglect, 13, 351-360. Eckenrode, J., Powers, J., Doris, J., Munsch, J., & Bolger, N. (1988). Substantiation of child abuse and neglect reports. Journal of Consulting and Clinical Psychology, 56, 9-16. Finkelhor, D. (1990). Is child abuse overreported? Public Welfare, Winter Issue, 22-29. Freeman, M. D. A. (1983). The rights and wrongs of children. London: Pinter. Graham, P., Dingwall, R., & Wolkind, S. (1985). Research issues in child abuse. Social Science and Medicine, 21, 1217-1228. Groeneveld, L. P., & Giovannoni, J. M. (1977). Disposition of child abuse and neglect cases. Social Work Research and Abstracts, 13, 24-30. Kalichman, S. C., Graig, M. E., & Follingstad, D. R. (1988). Mental health professionals and suspected cases of child abuse: An investigation of factors influencing reporting. Community Mental Health Journal 24, 43-51. Kaye, M., & Winefield, H. R. (1988). Child sexual abuse: A cybernetic description and its implications for professionals. Australian and New Zealand Journal of Family Therapy, 9, 131-138. Kinard, E. M. (1985). Ethical issues in research with abused children. ChiMAbuse & Neglect, 9, 301-31 t. Kotch, J. B., & Thomas, L. P. (1986). Family and social factors associated with substantiation of child abuse and neglect reports. Journal of Family Violence, 1, 167-179. Loftus, E. F., & Davies, G. M. (1984). Distortions in the memory of children. Journal of Social Issues, 40, 51-67. Lombard, F. K., Michalak, M. J., & Pearlman, T. A. (1986). Identifying the abused child: A study of reporting practices of teachers. University of Detroit Law Review, 63, 657-681. Nelson, G. K., Dainauski, J., & Kilmer, L (1980). Child abuse reporting laws: Action and uncertainty. (;him Welfare, 59, 203-212. Newberger, E. H. (1983). The helping hand strikes again: Unintended consequences of child abuse reporting. Journal t f Clinical ChiM Psychology, 12, 307-311. Rindfleisch, N., & Bean, G. J. (1988). Willingness to report abuse and neglect in residential facilities. Child Abuse & Neglect, 12, 509-520. Stadler, H. A. (1989). Balancing ethical responsibilities: Reporting child abuse and neglect. Consulting PsTchologist, 17, 102-110. Winefield, H. R. (1988). Psychologists' experiences and views in relation to child sexual abuse. A ustralian P~ychologist, 23, 13-23. Winefield, H. R., & Castell-McGregor, S. N. (1986). Experiences and views of general practitioners concerning sexually abused children. Medical Journal tfAustralia, 145, 311-313. Wolfe, D. A. (1987). Child abuse." lmplicationsfi>r child development and psychopathology. Newbury Park, CA: Sage. Zellman, G. L. (1990). Child abuse reporting and failure to report among mandated reporters. Journal of Interpersonal Violence, 5, 3-22. Zuravin, S. J., Watson, B., & Ehrenschaft, M. (1987). Anonymous reports of child physical abuse: Are they as serious as reports from other sources? Child Abuse & Neglect, 1 I, 521-529.

Rrsumr--Lorsque que les cas de maltraitance ne sont pas signalrs ou, dans le cas contraire, lorsque des signalements non justifirs parviennent aux services de protection de I'enfance, l'el~cacit6 de ces services se trouve diminu6e. A

Substantiation of reported child abuse

671

partir de registres de protection de renfant dans l'~tat de South Australia, les auteurs ont 6tudi~ tous les 3,228 cas de signalements parvenus durant l'ann6e 1988-1989, afin de voir quels sont les facteurs d~terminants qui m~nent les travailleurs dans ces services ~tinscrire les signalements de maltraitance clans leurs registres. Une analyse de r~ression logistique d~montra que les travailleurs validaient les cas ~tpartir de deux variables: l'~ge des pr6tendues victimes et la gravit~ de ia situation telle que perque par le travaiUeur. Cette deuxi~me variable, si importante, n6cessite une ~tude plus approfondie puisqu'il n'existe aucune donn6e fi savoir comment les travailleurs forment un jusement par rapport fi la gravit6 des cas. Resumen--La eficiencia de los servicios de protecci6n infantil seve reducida por los casos de sospecha de abuso y negligencia a los nifios que se reportan de m~s o de menos. Utilizamos todos los reportes de 1988-1989 del Stir de Australia (N = 3,228), para estudiar los determinantes de esta decisi6n de los empleados de protecci6n infantil para registrar un incidente reportado como que es abuso o negligencia a los nifios. La regresi6n logistica demostr6 que el registro (fundamentaci6n) podia predecirse desde dos variables: la edad de la alegrada victima y el grado de severidad que juzga el encargado del caso. Esta filtima variable neeesita investigaci6n, ya que, a pesar de su importancia, no tenemos informaci6n de como los encargados forman su juicio de severidad.

Substantiation of reported child abuse or neglect: predictors and implications.

Underreporting and overreporting of suspected child abuse and neglect cases reduce the efficiency of child protection services. We used all the report...
812KB Sizes 0 Downloads 0 Views