Construct validity and reliability of working memory tasks for people with aphasia

Background: Although several studies have examined working memory in people with aphasia, there is little information about the psychometric properties of tasks used to measure working memory in this population. Aims: The aims of this study were (1) to examine the construct validity, internal consistency, and test-retest reliability of several working memory tasks and (2) to determine which tasks were feasible to use with people with a wide range of aphasia severities. Methods & Procedures: In experiment one, non-brain-damaged adults (n=47) completed a set of working memory tasks that were designed for use with people with aphasia. The tasks included 1-back, 2-back, listening span and forward and backward versions of picture span and square span. Construct validity was assessed by correlating performance on the new tasks with a composite score based on three well-established working memory tasks with known psychometric properties. Ten older adults completed the testing battery twice to measure test-retest reliability. Internal consistency was estimated using split-half reliability. In experiment two, people with aphasia (n=12) completed the new working memory tasks. Outcomes & Results: The picture span tasks demonstrated acceptable test-retest reliability, internal consistency, and construct validity. Controls showed ceiling effects on 1-back and people with aphasia showed floor effects on listening span. Both the square span tasks and n-back showed poor internal consistency for people with aphasia and age-matched controls. Conclusions: A composite score based on the forward and backward versions of picture span provides a measure of verbal working memory with acceptable construct validity, internal consistency and test-retest reliability and can be completed by non-brain-damaged adults and people with aphasia.