Junior and Senior High School EFL Teachers’ Use of Formative Assessment: A Mixed-Methods Study

ABSTRACT

Despite the rising interest in the practice of formative assessment (FA) in language classrooms, research has rarely examined the factors that contribute to differential use of FA. The present study adopts an explanatory sequential mixed-method design to examine differences in FA use among Japanese junior and senior high school EFL (English as a Foreign Language) teachers. Based on Wiliam’s (2010) model of FA strategies, quantitative analysis of a large-scale survey classified 727 participants into three levels (high, middle, and low) of FA use. In the qualitative follow-up phase, four selected survey respondents were interviewed and observed. The results indicated varying degrees of FA use among teachers, suggesting that the four FA variables (intentions, methods, purposes, and feedback) derived from Wiliam’s model succeeded in identifying differential degrees of FA use.

Acknowledgments

The study was partly funded by Grants-in-Aid for Scientific Research (Japan Society for the Promotion of Science) awarded to the first author (Task Numbers 22520550/26381175). Our sincere thanks go to the great editing team of the special issue, Jim Purpura, and all the reviewers for their valuable and insightful comments.

Notes

¹ The feedback items did not consider any distinction between public and personal feedback because adding this dimension increases the number of items greatly and makes the already-long questionnaire even more complex.

² One of the reviewers of this manuscript suggested the possibility of pursuing multidimensional analysis by lumping together all items of the four FA strategies. We believe that separate analyses of each item group (with the Rasch) better serve our purposes because of the theoretical implications derived from the results.

³ Gender and employment are excluded from this analysis because these factors are unlikely to generate any practical implications. Age is also excluded here because of its high correlation with years of experience.

⁴ A few participants did not fit the model. Because additional analyses without misfit participants did not improve the overall fit of the data, all participants remained in the subsequent analyses.

⁵ Tsagari (2011) did not report the types of responses used in her paper, but this was confirmed through personal communication.