Accuracy of Data Submitted to an Automated Children's Protective Services Central Registry (Michigan).

Abstract

Child abuse/neglect case data submitted to a state central registry were studied. The purpose was to examine the reliability of these data, which are the basis for most administrative decisions and current theory regarding child abuse and neglect. Over 300 cases which had been investigated by Children's Protective Services in Wayne County, Michigan, were read. The information reported to the central registry was compared to the information which should have been submitted, based on the casefile. Specific items and different types of information (concrete/abstract; routine/seldom used) were found to have distinctly varied levels of accuracy. Type of case (denied/open/closed; court/noncourt; abuse/neglect/both) also influenced accuracy. Information often used for program evaluation and casework decisions was found to have error rates exceeding 20%. The presence of both abuse and neglect in the family, and of men in the home, were severely underreported. A 70% total error rate was found in the "name" field, which is a fatal database flaw. Type and direction of error is discussed. All supervisors (20) and section heads (5) in the county completed a questionnaire probing general background, self-reports of personality, attitudes towards paperwork, MIS staff, and data submission requirements, and specific supervisory practices. Significant differences were found in the accuracy of supervisory and administrative units, but theories about the effect of such supervisory differences, based on computer science and business administration users, were not supported with human service users. A least squares regression revealed five supervisory areas which accounted for 60% of the difference in accuracy found: (a) self-perceptions as more logical than intuitive and (b) more willing to take risks than peers, (c) belief that the MIS information may be used against them in the future, (d) self-reports that they checked the accuracy of the MIS form 100% of the time with routine casereadings and (e) have done some computer or information system design or training.