Approaching Classroom Realities: The Use of Mixed Methods and Structural Equation Modeling in Science Education Research

Teaching and learning in schools is influenced by various factors, such as social factors and inter- and intra-individual factors. Social factors comprise, for example, gender, race/ethnicity, social class, school norms, and beliefs/perceptions about schools and their members, while inter- and intra-individual factors include such characteristics as group dynamics, attitudes, interests, motivation, and perceptions about oneself and others (e.g., Arum & Beattie, 2000). In addition, specific curriculum expectations as well as hidden curriculum objectives impact teaching and learning as do other contextual factors, such as subject-specific national standards and national assessment movements, teachers' preferences for specific teaching and assessment strategies that are often deeply rooted in personal beliefs about teaching and learning, and the level of support of school administration. In general, classrooms are complex phenomena, complicating not only efforts to change practice but also the act of research itself. Turner and Meyer (2000) summarized bluntly, “Classroom research is messy” (p. 69). Researchers can either ignore the messiness and complexity of classrooms so as to concentrate on simple interrelations of two variables or the relation to an outcome variable (mostly student achievement) or they can investigate multiple variables from multiple perspectives using a multimethod approach. The latter vision must be embraced as part of any Gold Standard vision, with variables interpreted in relation to an understanding of the whole context.

A systematic review of current research on teaching and learning shows that most educational research has moved far from the old theoretical models that treated the classroom as a “black box” with data collection focusing on quantitative measures of “inputs” and “outputs” (Metz, 2000, p. 65). However, studies in educational research designed as randomized controlled trials with entire groups (e.g., classrooms, schools)—rather than individual students randomly assigned to treatment and control group by lottery—still operate as classic, black-box analyses. Reasons for or mechanisms producing posttest differences between two groups are seldom the focus of such large-scale studies. Instead, the focus is usually on examining programs rather than single treatments. Examples of this approach include studies identifying academic effects of vouchers on African American students who used them to switch from public to private schools (e.g., Howell, Wolf, Campbell, & Peterson, 2002).