Both self-explanation and sketching are known to be effective methods for teaching science and mathematics, but for whom do they work and why? This 3-year project with middle school students includes 1) comparing learning by self-explanation, sketching, and the combination of the two in different orders , 2) comparing these techniques across low-spatial-demand and high-spatial-demand representations, 3) relating knowledge, spatial abilities, and age (individual difference variables) to effects of these treatments, 4) following students longitudinally from 6th through 8th grades, and 5) using student verbalizations and sketching-in-progress to understand why the treatments work the way they do.
The study is being conducted in a medium-sized, medium-achieving middle school in New Jersey with middle school students from science and math classes. This is a small-scale experiment with multiple individual difference covariates and student verbalizations and sketching-in-progress (process data) collected.
The project uses a longitudinal and cross-sectional research design and will generate evidence that is correlational and causal [experimental]. We are collecting original data on students in 6th-8th grades in math and science classes using assessments of learning, observation [personal observation], and survey research [paper & pencil]. Interventions using Read-and-sketch; Read-and-Self-explain; Read, Sketch, and Self-explain; Read, Self-explain, and Sketch are being compared with a Read-only condition.
Instruments or measures being used include how many of 8 problems (4 math and 4 science) are solved correctly. Hidden figures and mental rotations tests will capture dynamic and static spatial abilities, along with a visuospatial working memory test. We will also measure motivations for using diagrams in math and science (valuing, self-efficacy, preferences, and anxiety) and motivations for sketching in math and science. We will use a series of repeated-measures ANOVA and ANCOVA analyses on cross-sectional data, growth curve modeling on longitudinal data, and correlations for process data.
Findings will be posted as they become available.
We will develop scoring rubrics and coding categories for analyzing the think-aloud and sketching process data.