Generalizability Theory and the Development of a Performance Assessment for a Clinically Based Mathematics Methods Course

• In Event: Roundtable Session 20
  In Roundtable Session: Designing Performance Assessment Tools in Teacher Education

Abstract

This study used generalizability theory to examine the reliability of a performance assessment used to evaluate preservice teachers’ abilities to engage in the core practice of applying research to mathematics teaching. The system of assessment that included three raters, two tasks, and two scoring occasions is described in detail. The assessment scores were found to be fairly reliable with an absolute generalizability coefficient of .81. It was determined that the facet of rater contributed most to score error variance and that the main effect of the task facet did not contribute to error variance. This study contributes to the field by providing an example of how course instructors can utilize their limited resources to develop reliable performance assessments.

Authors

• Sarah Ann van Ingen, University of South Florida

• Amanda Rose Mohn, University of South Florida

• Elaine Cerrato Fisher, University of South Florida