Commentary on Butcher, Arbisi, Atlis, and McNulty (2003) on the Fake Bad Scale

Abstract

Butcher, Arbisi, Atlis, and McNulty (2003) misunderstand the Fake Bad Scale (FBS) and make naive, inaccurate assumptions about malingering. Their reasoning and methodology do not support their conclusions. Their methodology and reporting are very unusual for a research study. Not even basic descriptive statistics were reported, for example, means, standard deviations, or ranges. Five of their six samples were not in the setting for which the FBS was designed, and only one—their smallest sample—was personal injury litigants. Applying the FBS in other contexts is analogous to applying the Marital Distress Scale to single persons. The authors used data in a National Computer Systems archival database containing 119,672 cases but excluded 89,675 cases (75%) without explanation and without stating the exclusion criteria. They excluded another 10,881 cases because their profiles appeared invalid, leaving 19,116 cases to study. Excluding 10,881 cases with invalid profiles when evaluating a scale designed to detect invalid profiles is analogous to testing the MMPI-2’s capacity to detect depression after first removing depressed patients from the sample. The selection bias in this methodology is calculated to produce a misleadingly low impression of the scale being evaluated. Their cutoff criteria were not justified and they are so unusually high that many people normally thought to be exaggerating were presumed to be valid cases instead of being excluded. For example, the authors used a cutoff of T = 110 for the F and F(b) scales. As noted by Greene (2000, p. 70), such scores represent “extreme” distortion and scores on these scales as low as 81 may represent invalid profiles due to symptom exaggeration. Butcher et al. used dramatically different standards to criticize the FBS than Butcher and his colleagues use in the MMPI-2 manuals (Butcher, Dahlstrom, Graham, Tellegen, & Kaemmer, 1989; Butcher et al., 2001). The first MMPI-2 manual indicates malingering as the first in the list of suspected sources of elevation of profiles with an F scale over 70, and says that such scales are of