Ahead of the (ROC) Curve: A Statistical Approach to Utilizing ex-Gaussian Parameters of Reaction Time in Diagnosing ADHD Across Three Developmental Periods

open_access
Open Access
Author:
Galloway-Long, Hilary Susan
Graduate Program:
Psychology
Degree:
Doctor of Philosophy
Document Type:
Dissertation
Date of Defense:
June 12, 2020
Committee Members:
Cynthia L Huang-Pollock, Dissertation Advisor/Co-Advisor
Cynthia L Huang-Pollock, Committee Chair/Co-Chair
Peter Andrew Arnett, Committee Member
Karen Linn Bierman, Committee Member
Barbara Schaefer, Outside Member
Kristin Ann Buss, Program Head/Chair
Keywords:
ADHD
neuropsychology
processing speed
receiver operating characteristic (ROC) curves
Abstract:
Extensive research has examined possible etiological roots of ADHD, with particular attention paid to deficits in processing speed and executive functions. However, because only about 30% to 50% of children with ADHD show deficits on neuropsychological tests relative to normative samples, one suggestion has been that current methods of measuring cognition may fail to accurately capture and describe the deficits seen within this population. The current study aimed to determine if ex-Gaussian parameters of reaction time are more accurately able discriminate between individuals with and without ADHD compared to traditional methods. Cognitive task performance was evaluated in individuals with and without ADHD across three developmental periods: kindergarten-age, middle childhood, and early adulthood. Receiver Operating Characteristic (ROC) curves and the area under the curve (AUC) were used to examine the ability of Go/No-Go and SSRT task parameters to accurately identify individuals meeting diagnostic criteria for ADHD. Possible clinical cut-offs were also explored. Findings were consistent across samples such that the traditionally used variables (SSRT and failed inhibit rate), as well as variable reaction time (indexed by SDRT) successfully discriminated between individuals with and without ADHD. When using ex-Gaussian parameters, this was generally found to be driven by lapses in performance (indexed by tau). Though no single variable was found to consistently provide the greatest combination of sensitivity and specificity when using Youden’s Index, optimal cutoffs generally showed higher specificity than sensitivity. Findings suggest that reaction time can identify individuals with ADHD as well as traditional measures such as SSRT and failed inhibit rate, and that using ex-Gaussian parameters provides a better description of variability in performance during speeded tasks than mean/SD. Therefore, assessment tools would benefit from implementing these methods.