Trends in the Salience of Data Collected in a Multi User Virtual Environment: an Exploratory Study
MetadataShow full item record
CitationTutwiler, Michael Shane. 2014. Trends in the Salience of Data Collected in a Multi User Virtual Environment: an Exploratory Study. Doctoral dissertation, Harvard Graduate School of Education.
AbstractIn this study, by exploring patterns in the degree of physical salience of the data the students collected, I investigated the relationship between the level of students’ tendency to frame explanations in terms of complex patterns and evidence of how they attend to and select data in support of their developing understandings of causal relationships. I accomplished this by analyzing longitudinal data collected as part of a larger study of 143 7th grade students (clustered within 36 teams, 5 teachers, and 2 schools in the same Northeastern school district) as they navigated and collected data in an ecosystems-based multi-user virtual environment curriculum known as the EcoMUVE Pond module (Metcalf, Kamarainen, Tutwiler, Grotzer, Dede, 2011) .
Using individual growth modeling (Singer & Willett, 2003) I found no direct link between student pre-intervention tendency to offer explanations containing complex causal components and patterns of physical salience-driven data collection (average physical salience level, number of low physical salience data points collected, and proportion of low physical salience data points collected), though prior science content knowledge did affect the initial status and rate of change of outcomes in the average physical salience level and proportion of low physical salience data collected over time.
The findings of this study suggest two issues for consideration about the use of MUVEs to study student data collection behaviors in complex spaces. Firstly, the structure of the curriculum in which the MUVE is embedded might have a direct effect on what types of data students choose to collect. This undercuts our ability to make inferences about student-driven decisions to collect specific types of data, and suggests that a more open-ended curricular model might be better suited to this type of inquiry. Secondly, differences between teachers’ choices in how to facilitate the units likely contribute to the variance in student data collection behaviors between students with different teachers. This foreshadows external validity issues in studies that use behaviors of students within a single class to develop “detectors” of student latent traits (e.g., Baker, Corbett, Roll, Koedinger, 2008).
Citable link to this pagehttp://nrs.harvard.edu/urn-3:HUL.InstRepos:13383549