Optimized short-forms of the Cognitive Distortions Questionnaire

Amanda S Morrison,Berk Ustun,Arielle Horenstein,Simona C Kaplan,Irismar Reis de Oliveira,Sedat Batmaz,James J Gross,Ekaterina Sadikova,Curt Hemanny,Pedro P Pires,Philippe R Goldin,Ronald C Kessler,Richard G Heimberg
DOI: https://doi.org/10.1016/j.janxdis.2022.102624
Abstract:Introduction: The Cognitive Distortions Questionnaire (CD-Quest) is a self-report questionnaire that assesses common cognitive distortions. Although the CD-Quest has excellent psychometric properties, its length may limit its use. Methods: We attempted to develop short-forms of the CD-Quest using RiskSLIM - a machine learning method to build short-form scales that can be scored by hand. Each short-form was fit to maximize concordance with the total CD-Quest score for a specified number of items based on an objective function, in this case R2, by selecting an optimal subset of items and an optimal set of small integer weights. The models were trained in a sample of US undergraduate students (N = 906). We then validated each short-form on five independent samples: two samples of undergraduate students in Brazil (Ns = 182, 183); patients with depression in Brazil (N = 62); patients with social anxiety disorder in the US (N = 198); and psychiatric outpatients in Turkey (N = 269). Results: A 9-item short-form with integer scoring was created that reproduced the total 15-item CD-Quest score in all validation samples with excellent accuracy (R2 = 90.4-93.6%). A 5-item ultra-short-form had good accuracy (R2 = 78.2-85.5%). Discussion: A 9-item short-form and a 5-item ultra-short-form of the CD-Quest both reproduced full CD-Quest scores with excellent to good accuracy. These shorter versions of the full CD-Quest could facilitate measurement of cognitive distortions for users with limited time and resources.
What problem does this paper attempt to address?