Deception is different: Negative validity test findings do not provide “evidence” for “good effort"

Abstract

Objective: The purpose of this paper is to determine whether negative validity test findings should be used in the Bayesian aggregate along with positive test findings for the determination of malingering as the condition of interest (COI). Method: Evidence-based diagnostic methods for conditions in neuropsychology and medicine were reviewed for comparison with their use in cases of malingering. Logical and Bayesian analyses of these cases were applied. A case study showed that negative validity test findings did not indicate “good effort”. Results: Deception about illness is fundamentally different from other constructs/diseases in evidence-based medicine and neuropsychology. This is because deception involves a deliberate process that may involve coaching, claimant research, and/or focusing the deception on one aspect (e.g., slowness) as opposed to other neurocognitive problems (e.g., memory). Comparatively, other conditions in medicine and neuropsychology are unlikely to be manipulated by the patient. Conclusions: The assertion by Frederick (2015) and Black, Necrason, and Omasta (2016) that both positive and negative validity test findings must be used together in the aggregate does not stand up to this comparative scrutiny. The fundamental assumption by these authors that a negative test finding concerning malingering represents “good effort” is flawed; it simply represents lack of evidence of malingering, which cannot be construed as evidence of lack of malingering. We recommend that in forensic determination of malingering, negative validity test findings should not be used in a Bayesian aggregation. This conclusion is consistent with current practices in the field.

Keywords:

• Evidence-based practice
• validity tests
• likelihood ratios
• malingering
• Bayesian probabilities

Acknowledgements

Thomas Frazier's workshop at NAN in 2006 on evidence-based methods was instrumental in opening up the Bayesian world to me. I am grateful to Glenn Larrabee for considerable help in applying these methods to the COI of malingering and for specific advice on these methods for my 2011 paper. Glenn Larrabee also provided critical commentary on a late version of the current paper, which was helpful. Christoph Leonhard provided an early critical reading of this manuscript, which was most helpful in communication of concepts. I am also indebted to Paige Haley for her early discussions of this paper, supplying calculations for the tables, and subsequent readings and commentary. I appreciated the reading and comments on the early manuscript by Christian Oldenburg. Darcy Cox and Ryan Schroeder provided insights on a later draft, for which I am grateful. The TCN reviewers for this manuscript made this a far better paper, and I am indebted to them. I appreciate the National Academy of Neuropsychology allowing me the opportunity to give an online workshop on these topics in 2020.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Larrabee (2008) uses the term positive predictive power (PPP), but I am using PPV for consistency with the rest of the paper and the cited authors.

2 Unfortunately, in the Meyers et al. (2014 study), the table showing sensitivities and specificities of the validity tests actually showed positive and negative predictive values, an error reported in Larrabee et al. (2019).

3 For a simplified presentation, and in keeping with the theme of this article, I am using only positive findings.

4 Even though the motive in this study was positive (e.g., to reduce invalid data in veterans), the effects are the same.