Validating an Academic Group Tutorial Discussion Speaking Test

This study attempts to validate an academic group tutorial discussion speaking test for undergraduate freshmen students taking initial EAP training at a university in Hong Kong in terms of task, rater and criterion validity. Three quantitative measures (Cronbach's Alpha, Intraclass Correlation Coefficient, and Exploratory Factor Analysis) are used to assess validity of rater scores for the test using a rubric with considerations for assessment of academic stance presentation, inter-candidate interaction, and individual language proficiency. These results are triangulated with post-hoc interview data from the raters regarding the difficulties they face assessing individual proficiency and group interaction over time. The results suggest that current provisions of the rubric in dealing with the assessment of interaction in group settings (namely visual cues such as "active listening" as well as provisions for interruptions in the form of "domination") are problematic, and that raters are unable to separate the grading of academic stance from the grading of language concerns. We also note affective and cognitive difficulties involved with assessing extended periods of interactional discourse including student talking time (or lack of it), the group dynamic, and raters" personal beliefs and practice as threats to validity that the statistical measurements were unable to capture. A new sample rubric and further suggestions for improving the validity of group tutorial assessments are provided.