Examining M-FAST item-level specificity and alternative scoring procedures in a forensic psychiatric sample.

Abstract

The recommended screening cutoff of ≥6 on the Miller Forensic Assessment of Symptoms Test (M-FAST) has been found to produce high false-positive rates in forensic populations. However, comprehensive testing of feigned psychopathology may be unfeasible, especially in resource- and time-limited forensic contexts. Using protocols from 172 forensic patients who were administered the M-FAST and Structured Interview of Reported Symptoms/Structured Interview of Reported Symptoms-Second Edition, this study examined whether there were nonspecific or nondiscriminant M-FAST items. This study also explored whether the removal of poor performing items from the scoring could result in improved classification accuracy compared to the standard 25-item scoring procedure. Results showed that removing four to six items that were not specific or did not discriminate between presumed honest and presumed feigning patients reduced false-positive error rates while maintaining similar sensitivity levels to the 25-item M-FAST. Using a pool of 19 or 21 M-FAST items in the scoring could increase confidence in assertions about noncredible presentations (i.e., invalid test response) and reduce the need for unwarranted comprehensive testing among forensic patients with severe mental illness. (PsycInfo Database Record (c) 2026 APA, all rights reserved).