Carobeth Wheeler
Marshall University

Huntington, WV–Behavior analysts and others have long known that projective tests (ink blots, human figure drawings, etc.) are, essentially, invalid as measures of human behavior, thoughts, and feelings. However, a recent meta-analysis (West, 1998) attempted yet again to show that these techniques are successful in diagnosing sexual abuse in children. West concluded that "projective techniques have the ability to discriminate between children who have been sexually abused and those who are not abused sexually" (p. 1151).

But now researchers Howard N. Garb of Pittsburgh V.A. Health Care System, James M. Wood from the University of Texas at El Paso, and M. Teresa Nezworski of the University of Texas at Dallas have undone West's conclusions by reanalyzing her results. Projective techniques are not successful in detecting child sexual abuse they say.

Garb and his colleagues pointed out three major flaws in the West study. First, West selectively reported results in her study. Instead of including all of the data she had gathered, West left out the studies with small effect sizes, which made the overall usefulness of the projective techniques appear larger than they really were.

The second flaw in West's meta-analysis was more serious. She failed to rule out the research that was plagued by the examiner's knowledge of the child's abuse status. West included studies in which the examiner knew the abuse status of the children while testing them. Clearly, West did not rule out the strong likelihood that projective interpretations were influenced by prior knowledge of whether the child had been abused.

The third flaw in West's study was the inappropriate use of responses by non-abused children in other studies for comparison. That is, several of West's twelve studies used no control group of their own, instead relying on control group data (non-abused children's responses) from other studies.

Along with identifying these errors in West's study, Garb and his colleagues reanalyzed West's data. They looked at the same twelve studies and came to different conclusions. The usefulness of the tests had been overestimated by more than one hundred percent by West.

Garb, Wood, and Nezworski concluded that projective techniques are not sufficiently effective to identify child sexual abuse. Behavior analysts might well consider an analysis of variables that maintain viability of projectives in the absence of validity.

Garb, H.N., Wood, J.M., and Nezworski, M.T. (2000). Projective techniques and the detection of child sexual abuse. Journal of the American Professional Society on the Abuse of Children, 5, 161-168.

West, M.M. (1998). Meta-analysis of studies assessing the efficacy of projective techniques in discriminating child sexual abuse. Child Abuse and Neglect, 22, 1151-1166.

Back to the table of contents for this issue

Return Home