Q-Matrix Construction: Defining the Link Between Constructs and Test Items in Large-Scale Reading and Listening Comprehension Assessments The Authors Respond

Click to increase image sizeClick to decrease image size

Notes

¹While learner verbal protocol analyses were in fact conducted more recently by other groups of researchers on LanguEdge, results from those studies (e.g., Cohen and Upton, 2006) were not available when we conducted our study between fall 2003 and summer 2004. ).

²Although such an endeavor was beyond the scope of the present project, this issue is being pursued in other related projects. For example, Sawaki and Sabatini (2008) investigated prevalence of score profiles across reading comprehension, vocabulary size, and efficiency of component reading skills among Chinese, Korean, and Spanish L1 speakers in the TOEFL population. The TOEFL iBT Reading section and a series of discrete measures designed to assess vocabulary size and efficiency in different component reading skills were combined in this investigation.

³See Pearlman (2008) for a sample task shell for a speaking task type that was considered in early stages of the TOEFL iBT development but not implemented in the operational test.