Analytic induction, the approach to data analysis adopted in this study, refers to a systematic examination of similarities within broad categories and then using this to develop subcategories (Ragin, 1994). After implementation was completed, all data were compiled by a research assistant. This data was then made available to the authors and, with the aid of the coding software, MaxQDA, the Magnusson, Krajcik, and Borko (1999)
82 P. Osmond and K. Goodnough
Downloaded by [Chiang Mai University] at 00:11 04 December 2015
framework was used as broad categories to begin the qualitative analysis of the transcribed interviews, journal entries, anecdotal notes, survey comments, and student responses. The first and second authors started by reading and re-reading the data set and then recording memos. Next, all qualitative data were initially analyzed and relevant paragraphs were highlighted and coded under one of the five components (orientations to teaching science, knowledge of curriculum, knowledge of student understandings, knowledge of assessment, or knowledge of instructional strategies). Later, these broad themes and supporting data were re-analyzed and coded into smaller sub-categories as common sub-themes emerged. For example, “knowledge of student understandings” was later broken down to include the sub-categories student misconceptions, prior understandings, and conceptual change. The Likert-scale survey questions were compiled using Microsoft Excel and quantitatively analyzed using simple statistics