Assessing Strategic Control of Variables in Context

• In Event: Transfer for Learning in the Classroom: Developing Contexts to Examine STEM Learning Skills and Strategies

Abstract

A key aspect of conducting scientific inquiry is controlling for the right variables at the right time. Often termed Control of Variables strategy (CVS), this skill is emphasized in the NGSS core practices, as well as in a wide body of research (cf. Chen and Klahr, 1999). While being essential for productive inquiry, it is also clear that the complex inquiry process cannot be reduced to CVS (cf. Kuhn, Iordanou, Pease, & Wirkala, 2008). Here, we further argue that CVS cannot be reduced to a procedural and pseudo-technical set of actions. While controlling variables may be easy, choosing how and when to do so may be more challenging. Thus, our work on assessing CVS has two goals in mind: to better understand strategic use of control of variables and its relationship to learning, and to evaluate how it can be captured and assessed based on students’ trace data from virtual labs.

We studied strategic use of control of variables using an inquiry environment that supports multiple forms of CVS:
1) An interactive simulation. PhET simulations (Wieman, Adams, & Perkins, 2006) allow students to manipulate properties of the environment and observe the outcomes. We used one simulation on light absorbance (https://phet.colorado.edu/en/simulation/beers-law-lab) and one on capacitors (https://phet.colorado.edu/en/simulation/capacitor-lab-basics). For example, in the Light sim, students can change the width of a beaker and observe how this affects the absorbance of light going through that beaker.
2) A data table. Students could save trials from the sim, arrange them in the table, and restore them. The table included all necessary information to draw inferences.
3) A graphing tool. Students could add points from the table to the graph, choose relevant axes, and change the scale of the axes between linear, log, or exponential.

Each element in the environment affords a different form of CVS:
CVS-Explore: In the sim, most actions are instances of CVS. For example, when students change the width of the beaker, all other elements are being held constant by default. This allows students to get immediate feedback on their CVS actions.

CVS-Table: In the table, students can collect data in which all variables but one are held constant.
CVS-Graph: In the graph, students can plot points that differ along only a single variable and draw inferences from their plots.

Students were tasked with discovering the relationship between variables in these simulations. Analysis of log-files and test performance revealed that only CVS-Graph was associated with learning. Further, CVS-Graph was effective only when used together with other relevant skills, such as linearizing graphs with exponential relationships.
We also found that students transferred their CVS behaviours across the two simulations, showing the generality of the skill and its detection.
These results highlight the strategic nature of CVS - done right, it is a meaningful scientific tool. However, to be effective, it needs to be used together with other relevant skills, applied in context.

Authors

• Ido Roll, Technion Israel Institute of Technology

• Sarah Perez, University of British Columbia

• Jonathan Massey-Allard, University of British Columbia

• Deborah L. Butler, The University of British Columbia

• Joss Ives, University of the Fraser Valley

• Doug Bonn, The University of British Columbia