Abstract
There is a growing national recognition that teachers and teaching are at the heart of successful educational reform. However, few tools exist for measuring classroom instruction. The primary purpose of this article is to describe methods we developed to measure and study teaching, specifically while teachers were using a multimedia intervention for promoting scientific problem solving. Lessons were videotaped, and coding schemes were developed to measure 2 aspects of teaching: (a) the lesson's organization, particularly whole-class instruction used to introduce problems and share students' work; and (b) the nature of tasks and questions given to students. Results showed that the coding schemes were reliable and that they detected differences in instruction across teachers. Qualitative analyses were consistent with the quantitative findings. The codes also captured features of teaching that would have been difficult to detect or verify with qualitative observations alone. Finally, we explored how these measures could be used with student outcome data to examine the relationship between teaching and learning in future studies. We argue that quantitative measures of instruction serve many purposes, not the least of which is allowing researchers to explore the relationship between teaching and student learning at a high degree of granularity.
Notes
1See www.immex.ucla.edu.
2We asked teachers to inform us if their students were allowed to work on the problem set beyond the videotaped period so that we could tape the problem from start to finish. Only one teacher extended work on a problem to a second day. However, the second day of work took place a week after the first, and this teacher merely allowed students to work individually on the problem on which they had previously worked in groups. It was questionable whether this additional period of work constituted a continuation of the first lesson. For clarity, we omitted the second day of work from our analyses.
3We relied on the private work and public work distinctions made during TIMSS for defining some of the segmenting codes (CitationHiebert et al., 2003).