Design Parameter Values for Planning Mediation Studies with Teacher and Student Mathematics Outcomes

Abstract

Despite the widespread use of mediation analyses to test and unpack program effects across areas of education research, scant empirical literature is available to guide the effective and efficient prospective design of mediation studies. We contribute to this gap by developing an initial collection of empirical estimates of design parameter values for mediation studies drawing on teacher mediators and student outcomes within a three-level structure context: students nested in teachers nested in schools. Our analyses significantly expanded the scope and breadth of prior work by investigating design parameter values across a broad range of contexts (e.g., academic year, summer programs), modes (e.g., surveys, observations, assessments), mediator types (e.g., knowledge, instruction, affect, beliefs, perceptions, attitudes, climate, program implementation), instruments (e.g., MKT, CLASS, MQI, CLE, TIPS), and domains (e.g., algebra, geometry, classroom management). The results demonstrate substantial variability in parameter values across mediators and outcomes and highlight the potential value of collecting pretreatment assessments of mediators (and outcomes) in increasing statistical power. Collectively, the results underscore the significance of drawing on mediator- and outcome-specific design parameter estimates when planning studies.