Computerized Test Interpretation of the MMPI-2 in the Forensic Context: A Time to Use Your Head and Not the Formula?

Computer-based test interpretations (CBTI) of the MMPI-2 are ubiquitous in forensic evaluations despite limited evidence that they are either reliable or valid. Not only are they unable to account for the test taker’s lived experiences, thus yielding generic narrative statements that may or may not be accurate, but the primary focus of CBTIs of the MMPI-2 on psychopathology also yield a predominance of negative descriptors as well as inappropriate attributions of causality. This situation can be problematic, particularly in the hands of evaluators who lack the necessary training and experience to interpret them properly. This is of particular concern in forensic evaluations, in which the individual test taker’s credibility and character are directly at issue. Due to the proprietary nature of CBTI programs, it is not possible for independent researchers to fully examine the algorithms upon which they rest. Such programs are not subjected to peer review, and there are no known error rates associated with them; thus, it is therefore arguable whether CBTIs meet the Daubert standard, which requires these and other indicia of scientific rigor. In the present paper, we subjected the responses of a plaintiff involved in a sexual harassment lawsuit to five popular MMPI-2 CBTI programs and their within and between variability were examined; we also discuss the prejudicial implications of descriptions and predictions based on the incomplete information available to automated programs. Finally, we discuss validity concerns and conclude with suggestions for whether and how these programs can be improved.