Improved Reliability of Dot Probe Measures with Response-Based Computation: An Application with Young Violence-Exposed Children.

Background: Anxiety and posttraumatic stress symptoms have been associated with threat-related attention bias assessed in the widely-used dot-probe task. A novel method of scoring attention bias using response-based computation has shown improved reliability over standard computation approaches with adult samples. Methods: The current study applies this approach using secondary analysis of dot probe data from a sample of preschool-aged children ages 3-7 (n = 280) with varying levels of violence exposure. Attention bias indices were then examined alongside mother-report and observational measures of child anxiety. The standard approach used average reaction times to create a single measure of vigilant Results: The standard computation approach demonstrated unacceptable levels of internal consistency. In contrast, response-based computation of vigilant and avoidant bias demonstrated good and acceptable levels of internal consistency, respectively. Using the standard computation approach, no significant symptom associations were observed. Using the response-based computation approach greater ratio of vigilant bias relative to avoidant bias (i.e., vigilant bias > avoidant bias) was related to more observed anxious behaviors. Conclusions: Similar to research in adults, response-based attention bias indices demonstrated superior psychometric properties and stronger symptom associations compared to the standard computation approach and may offer advantages over the standard computation approach to study attention bias in young children.