|
Citation
|
Finkelman MD, Lowe SR, Kim W, Gruebner O, Smits N, Galea S. Psychol. Trauma 2017; 9(3): 379-389. |
|
Abstract
|
OBJECTIVE: To investigate the potential of customized computer-based testing procedures to reduce the mean test length of the Posttraumatic Stress Checklist for DSM-5 (PCL-5).
METHOD: A retrospective analysis was conducted using responses from 942 adults who had completed the full-length (20-item) PCL-5 in the aftermath of Hurricane Sandy. The abilities of 2 testing procedures, curtailment and stochastic curtailment, to lessen the instrument's mean test length while maintaining the same result as the full-length PCL-5 ("positive" or "negative") were evaluated in a post hoc simulation. Curtailment and stochastic curtailment track a respondent's answers as she takes the instrument and stop the test if future items are unable or unlikely to change the result. The performance of each procedure was recorded under 2 scoring methods: a total-score-based method and a cluster-based method. Each procedure's sensitivity, specificity, and overall agreement with the full-length PCL-5 were computed.
RESULTS: Curtailment reduced the mean test length by 40% under the total-score-based method, and by more than 70% under the cluster-based method, while exhibiting 100% sensitivity, specificity, and overall agreement with the full-length PCL-5. Stochastic curtailment reduced the mean test length by up to 88% under the total-score-based method, and up to 84% under the cluster-based method, while always exhibiting at least 92% sensitivity and 99.8% overall agreement, as well as 100% specificity, for the full-length PCL-5.
CONCLUSIONS: Curtailment and stochastic curtailment have potential to enhance the efficiency of the PCL-5 when this assessment is administered by computer. The 2 procedures should be evaluated in future prospective studies. (PsycINFO Database Record
(c) 2016 APA, all rights reserved).
Language: en |