The project uses a cross-sectional and comparative research design and will generate evidence that is both associative [quasi-experimental] and causal [fixed-effect model and nested two-level regression analysis with students at first level and classrooms at second level].
The interventions used will be a genetics project based science (PBS) curriculum + narrative, PBS + narrative + conflict, and PBS + narrative + conflict + community as compared to a genetics PBS curriculum control. To address our research questions, we will measure changes in students’ affective and learning outcomes. Students will complete affect self-report surveys immediately preceding the instructional intervention in their classrooms. Students will do this for all of the specific affective constructs that are relevant to this study- motivation, engagement, attitudes, and plans- for which specific survey instruments have been identified from the literature. In addition, we will collect test results from students using genetics content assessments. Immediately following the completion of the instructional intervention, students will complete the same affective surveys and genetics assessments. We will also collect demographic information from all participating students: gender, race, ethnicity, socio-economic status in the form of free or reduced lunch status, and scores on standardized achievement tests.
We will conduct a nested two-level regression analysis (hierarchical linear modeling or HLM), with students at the first level of the model and classrooms at the second level. We will adjust the standard errors of the mean differences for clustering since students are nested within classes. Random assignment of classrooms will assure that the control and three experimental groups are similar on average in all characteristics. Our first model will address our first hypothesis about the size of shifts in motivation, engagement, attitudes, plans, and genetics achievement. Specifically, we aim to determine the value added by each game-based element on these outcomes of interest, and we will use the model to compare the change in outcomes between any two groups. To address our second hypothesis, determining for whom the game-based design elements work, we will build a second regression model that includes interaction effects for students’ membership in groups underrepresented in science and engineering careers (e.g., African American, Hispanic or Latino, American Indian, or female) to determine if the treatment effect is moderated by specific student characteristics.