Theses Doctoral

How Do We Develop Multivariable Thinkers? An Evaluation of a Middle School Scientific Reasoning Curriculum

Ramsey, Stephanie Holstad

The development of single-variable causal reasoning is well-studied, with children demonstrating an impressive ability to detect causality from an early age. A less studied, and perhaps important, ability is to understand multivariable causality. Individuals who possess a single-variable mental model of causality risk not thinking deeply enough to accurately detect and understand how the world works. Previous work with middle-schoolers (Kuhn, Ramsey and Arvidsson, under review) has shown that students can be supported in developing their mental models of causality through an extended opportunity to deeply engage in self-directed investigations. These investigations used social studies and health content within social studies and science classrooms; the current work evaluates whether similar development can be supported through using science curriculum content. To evaluate the question, an intervention was conducted in which students performed self-directed investigations into databases to uncover relationships in the data. These investigations were carried out by utilizing InspireData, an age-appropriate software that allows students to visually represent data. Ninety-two eighth grade students were assessed after a self-directed investigation of factors affecting precipitation levels in which they used InspireData to interpret data. Approximately 58% of students had previous experience with a self-directed investigation into factors affecting Body Mass Index, also using InspireData. Students either participated in a one-day intervention (the dense condition) or a six sessions within a two-week period (distributed condition). The effectiveness of the intervention was measured through three assessments: 1) The eighth grade research report prepared during the intervention; 2) A graph-reading assessment which used novel InspireData graphs; and 3) The Cancer Task, which provided an assessment of each student's mental model of causality. Intervention students had superior understanding of causality when compared to an out-of-school control group for mental model of causality, but the improvement in scientific reasoning skills was not as dramatic as in previous interventions. Intervention students demonstrated an ability to detect causal relationships during their intervention, as well as on unfamiliar graphs. There were no differences in graph interpretation and research report performance by condition (dense or distribution conditions) or previous experience. These results suggest that the understanding of multivariable causality is a fragile construct which will not always develop under what appear to be similar circumstances. Students in this intervention investigated a database, successfully identifying relationships present in the data, but were not as likely to undergo the cognitive change necessary to improve their multivariable thinking as participants in previous interventions. Beliefs about the nature of science may affect how students participated in the intervention and therefore whether conceptual development regarding causal understanding was possible. Suggestions for further research into the circumstances in which multivariable understanding can develop and implications based on these findings are discussed.


  • thumnail for Ramsey_columbia_0054D_11881.pdf Ramsey_columbia_0054D_11881.pdf application/pdf 4.58 MB Download File

More About This Work

Academic Units
Cognitive Studies in Education
Thesis Advisors
Kuhn, Deanna
Ph.D., Columbia University
Published Here
April 11, 2014