Date of Award

2011

Degree Type

Dissertation

Degree Name

Ph.D.

Department

Psychology

First Advisor

Shore, Douglas (Psychology)

Keywords

Psychology, Clinical.

Rights

CC BY-NC-ND 4.0

Abstract

The present study sought to generate an embedded effort index within the CVLT-II yes/no recognition trial using a known-groups design. Four types of recognition foils--i.e., novel/semantically unrelated (UN), novel/semantically related (PR), list B/semantically unrelated (BN), and list B/semantically related (BS)--as well as two composites--i.e., easy to reject foils (ETR) and difficult to reject foils (DTR)--were evaluated on their ability to distinguish between a group of 82 outpatients with moderate-severe traumatic brain injuries (TBI) and a group of 31 litigants meeting Slick et al. (1999) criteria for malingered neurocognitive dysfunction (MND). Separate multiple logistic regression analyses were performed. The full model based on the 4 foils correctly classified 88.5% of cases (61.3% sensitivity/98.8% specificity). The full model based on the composites correctly classified 81.4% of cases (45.2% sensitivity/95.1% specificity). With respect to univariate predictors, UN correctly classified 51.6-64.5% of MND cases and 90.2-100% of TBI cases depending on the diagnostic cut-off used. ETR also showed good classification accuracy (25.8-51.6% sensitivity/90.2-100% specificity). Three different ratios were generated from the original analyses--UN/PR, UN/(PR+BN+BS), and ETR/DTR. All three ratios yielded good to excellent diagnostic accuracy (87% sensitivity/98.4% specificity, 70.4% sensitivity/97% specificity, and 38.5% sensitivity/95.5% specificity, respectively). In addition, UN, ETR, and the multivariate equations were cross-validated with a group of 19 patients with complicated mild TBI supplying adequate effort (MTBI) and a group of 23 patients with complicated mild TBI performing poorly on effort measures (SE), resulting in high specificity values depending on the cut-offs used. Finally, previous research using the CVLT and CVLT-II (Coleman et al., 1998; Curtis et al., 2006; Millis et al., 1995; Millis et al., 2007; Sweet et al., 2000) was replicated. Overall, UN, ETR, both multivariate equations, and all three ratios derived from the foils of the CVLT-II yes/no recognition trial show considerable merit as embedded effort indices. Predictive power values are provided for all predictors at various diagnostic cut-offs across 5 hypothetical base rates to facilitate generalization of findings to different settings. Clinical and forensic implications are discussed with a focus on differential diagnoses.

Share

COinS