Skip to main content
Open Access Publications from the University of California

UC Berkeley

UC Berkeley Electronic Theses and Dissertations bannerUC Berkeley

Making Controlled Experimentation More Informative in Inquiry Investigations

  • Author(s): McElhaney, Kevin
  • Advisor(s): Linn, Marcia C
  • et al.

This dissertation incorporates three studies that examine how the design of inquiry based science instruction, dynamic visualizations, and guidance for experimentation contribute to physics students` understanding of science. I designed a week-long, technology-enhanced inquiry module on car collisions that logs students` interactions with a visualization. Students studied the module and responded to pretests, posttests, and embedded prompts that assessed students` understanding of motion graphs and collisions. In Study 1, students (N=148) made large, significant overall pretest to posttest gains. Regression models showed that the propensity for students to conduct controlled trials was the strongest predictor of learning when controlling for prior knowledge and other experimentation measures. Successful learners employed a goal-directed experimentation approach that connected their experimentation strategy to content knowledge. Study 2 investigated the effect of limiting students` experimentation on their planning, strategies, and learning outcomes. Students (N=58) made large, significant overall pretest to posttest gains. Students constrained to twelve trials isolated variables in their experiments better than the unconstrained students. However, the constrained students significantly underperformed the unconstrained students on the module assessments, indicating that isolating variables during experimentation did not lead to improved learning outcomes. In Study 3, students (N=166) were assigned to conditions that prompted them either to isolate or compare variables. Both groups made moderate, significant pretest to posttest gains. Students in the compare treatment used more diverse experimentation strategies than students in the isolate treatment. Compare students made nuanced interpretations of collision events based on threshold values. Case studies illustrate how comparing rather than isolating variables helped students use wide-ranging strategies to reach complex insights. The findings illustrate how students can benefit from experimentation strategies that do not isolate variables. Spontaneous exploration can help students test new questions that arise from unexpected results, informing the design of controlled tests that better reveal subtle characteristics of the variables. Guidance that encourages students to compare rather than isolate variables may have important benefits, such as prompting students to search for distinctions among the variables. The findings have important implications for the design of inquiry-based science instruction.

Main Content
Current View