AERA 2018 Annual Meeting: “The Dreams, Possibilities, and Necessity of Public Education”
April 13-17, 2018
Title: A Systemic Review of Formative Assessment: Efficacy and Characteristics
Authors: Hansol Lee, Huy Quoc Chung, Jamal Abedi, Roxanne Rashedi
Along with previous meta-analytic attempts to examine an overall effect of FA on student learning outcome, there have been continuous attempts to define the characteristics of FA through various perspectives. Nevertheless, little has been studied with regards to how different elements of FA have differential impacts on student learning outcome. To address this limitation, the present study systemically reviewed previous empirical studies that conducted FAs to improve student learning in K-12 classrooms, and we could investigate the efficacy and characteristics of FA. To investigate the impact of FA without any extraneous elements, we only reviewed randomized controlled trials (RCTs) or quasi-experiment studies that controlled for any baseline differences.
First, analysis of the identified 125 effect sizes from the selected 32 studies that had met inclusion criteria (e.g., with control conditions) revealed a small-sized positive effect of FA on student learning (Hedges’ g = 0.28, SE = 0.05, p < 0.001) with the greatest benefits for Mathematics and Arts, corresponding the previous findings from the former meta-analysis research. The current meta-analysis was distinguishable because we computed multiple effect sizes for each study to capture a wide array of characteristic of FA, instead of calculating single effect sizes for studies. In order to avoid the violation of independence assumption across effect sizes coming from the same study, we conducted a multilevel regression analysis as a variance-known model because each effect size—the dependent variable of the equation—has its own standard error estimates (i.e., sampling errors).
In addition to the efficacy of FA, we attempted to identify characteristics of FA and their impact on improving student learning outcome. Based off of the previous meta-analytic efforts in regards to the moderator variables, a total of 15 variables were taken into consideration, such as original target of FA, formality of FA, timing of FA, time points of FA, instruction adjustment after FA, feedback after FA, etc., and our further investigation with the identified moderator variables emerged from the data set led us to find that targeting students and conducting in a formal way are believed to enhance its efficacy, and the adjustment of teacher instruction in light of monitoring via FA and student self-assessment based on feedback from FA was also found to be strong elements in this regard. The research and pedagogical implications about the effective use of FA in K-12 classrooms are further discussed.