Investigating Chemistry Learning Using Virtual Lab Activities in Real Classrooms

• In Event: 65.063 - Theory-Driven Design of Online Learning Systems, Data Structures, and Analytic Techniques

Abstract

Objective
Chemists view chemistry as the science of explaining reactions, analyzing substances, and synthesizing new compounds. However, many chemistry students see chemistry as quantitative problem solving. The aims of the ChemVLab+ project are to create authentic, problem-based online activities that help students connect the procedures of chemistry with core concepts of the discipline. In the current presentation, we explore how classroom level variables such as instructional sequencing and engagement can be integrated with fine-grained data from log files to create a better picture of whether and how students learn from these activities.

Theory
The ChemVLab+ activities were designed based on principles from chemistry education, cognitive and learning sciences. As authentic and contextualized practice promotes learning in chemistry classrooms (Cuadros, Leinhardt, & Yaron, 2007), the activities are created around real world scenarios. As active engagement in processing and linking representations supports student learning (e.g., Clark & Mayer, 2011), we created activities that require students to map between multiple representations including molecular views, beakers, and chemical equations. Finally, as research demonstrates the efficacy of that feedback with explanations (e.g., Pashler et al., 2007), we use multiple layers of formative assessment with intelligent tutor based hints and reports for teachers.

Methods
In the current study we explored whether four ChemVlab+ activities on the topics of thermodynamics, equilibrium and acid-base chemistry, improve student outcomes in chemistry in diverse California high school settings. Fourteen teachers with 1473 students participated in the study. In the current analyses, we include data from 1195 students that completed both pre and posttests. Teachers attended a 3-hour professional development workshop about the use of the activities and study requirements. Students completed pretests, used the four activities in the course of normal classroom instruction, and completed posttests.

Data
A range of data was collected through teacher logs, student demographics, classroom observations, and pre and posttests. For comparability, the same assessment was used as a pre and posttest and was made up of multiple choice and open-ended items that were scored for a possible total of 34 points. Items were researcher-generated or selected from released standardized tests, and were validated by IRT analyses in field-testing. In addition, all student actions in the open-ended learning environment were logged.
Results
Results from a repeated measures ANOVA revealed that posttest scores (M = 16.1) were significantly greater than pretest scores (M = 13.8). Though students were not randomly assigned to condition, ongoing analyses using educational data-mining techniques support the claim that the activities improve science learning outcomes. Further, information from teacher logs suggests that students using the activities as out-of-class assignments had larger learning gains than those using the activities in class.
Significance
Many reports of findings from online learning environments fail to account contextual variables that may significantly impact learning. Data from multiple sources must be considered to create a full picture of whether and how students learn from online systems, and what inferences can be made from student interactions with these systems.

Authors

• Jodi Davenport, WestEd

• Anna Rafferty, University of California - Berkeley

• Michael Karabinos, Carnegie Mellon University

• David Yaron, Carnegie Mellon University