The article, "Improving Decision Making in Forensic Child Sexual Abuse Evaluations", appears in the February 2005 issue of the journal Law and Human Behavior, Vol. 29, No. 1, pages 87-120. Copies of the article are available upon request from the author at drsteveherman@gmail.com.

Courts and child protection agencies often look to child protective service social workers and other child professionals to conclude that child sex abuse allegations have or have not been "substantiated," that is, shown to be true. Herman begins his analysis by considering the scientific reliability of these substantiation decisions. In science, a procedure is reliable if it produces the same result each time it is performed on the same information (the same case), even if the person performing the procedure changes. Herman focuses on reliability because rarely can the accuracy of a child sex abuse allegation be conclusively established. He notes that a procedure for assessing allegations cannot be highly accurate without being highly reliable.

After a detailed statistical analysis of data gathered in many previous studies, some of child sex abuse cases and some in related fields, Herman concludes that the minimum error rate for substantiation decisions is 24%. This means at least 25,000 incorrect substantiation decisions (false positives and false negatives) nationally each year by CPS workers and thousands more by other evaluators.

Herman observes that this error rate calls into question whether courts should admit substantiation decisions in evidence. They may not qualify under the Daubert case and other standards for admitting expert testimony.

Herman also cites studies showing that substantiation decisions are affected by factors which logically cannot relate to the truth of the allegation. One example is whether the evaluator has ever been sexually abused. Other examples of such factors are the evaluator's age, gender, and discipline (social work, psychology, etc.).

Herman then asks how the error rate for substantiation decisions can be improved. He discounts increased training in child interviewing. Studies have shown that it produces lasting improvement in interview technique only when followed by permanent intensive supervision. Such supervision, Herman believes, is not going to happen, at least not in many places or for very long.

Herman also discounts improving guidelines for professional practice. He says that research on doing this, though scanty, suggests practice guidelines do not much affect the way child professionals work.

Herman instead suggests using an actuarial procedure-- substituting objective factors for a clinician's judgment. To illustrate the benefit of the actuarial approach, Herman proposes a simple rule for making substantiation decisions: the allegation would be accepted as true when the child has reported abuse before formal investigation and when the child has also reported abuse in one of the child's first three interviews in the formal investigation. Otherwise, according to this rule, the allegation would be deemed false.

Herman applies this rule to cases which, he says, were evaluated by teams of exceptionally qualified mental health professionals, operating under optimal conditions. His rule yielded the same substantiation decision that those teams reached in a large majority of the cases.