A Selective Synthesis of the Experimental Literature on Dynamic Assessment

Abstract

Dynamic assessment is a procedure that determines whether substantive changes occur in examinee behavior if feedback is provided across an array of increasingly complex or challenging tasks. This procedure contrasts with traditional models of assessment in which there is no feedback from the examiner on student performance. This article summarizes a meta-analysis of published studies on dynamic assessment. Thirty studies, which yielded 170 effect sizes (ESs), were reviewed. An analysis of main effects showed that ESs varied significantly as a function of ability group (underachievers yielded higher and children with learning disabilities yielded lower ESs than average, hearing-impaired, and mentally retarded participants did), chronological age (younger students yield higher ESs than older students do), sample size (studies with moderate sample sizes yielded larger ESs than did studies with small or large sample sizes), and type of assessment procedure (scaffolding and general strategies yielded larger ESs than did coaching). Mixed regression modeling indicated that the magnitude of the ES was best predicted by the type of assessment (ESs are higher for general strategies than for scaffolding, and scaffolding yields higher ESs than coaching does) and assessment domain (ESs are higher on visual-spatial than on verbal measures). None of the ability group classifications predicted ES in the complete regression model. These results were discussed within the context of the assumptions related to dynamic assessment and its current limitations.