Exploring Collaboration Interaction Patterns and Game Play Processes in an Educational Game

• In Event: Game-Based Measures of Learning Processes: Challenges, Methods, and Solutions

Abstract

Collaborative problem solving (CPS) is a competency critical for success in today’s workforce, and therefore important to develop and assess in educational contexts (Graesser et al., 2018). Digital environments such as educational games provide a unique space for students to interact around complex problems and display their CPS capabilities. In particular, the captured discourse and behaviors can support inferences about individuals’ capabilities. In the current study, we used students’ gameplay behaviors and conversations to explore their collaboration interaction patterns, and how those relate to gameplay processes.

Thirty-nine dyads (Mage = 13.6 years, SD = 0.9 years) completed pre-task surveys, an educational game (Physics Playground, Shute et al., 2013), and post-task surveys. Each dyad completed six levels, drawing items so a ball could hit a targeted balloon. Students worked on their own computers and communicated via video conference. The teams completed 186 levels (M = 4.77; SD = 1.18).

Audio recordings of students’ interactions were coded for the presence of one of five CPS interaction patterns (collaborative, cooperative, dominant/passive, dominant/dominant, expert/novice). The CPS interaction pattern usage was cooperative (41.9%), collaborative (32.8%), dominant/passive (19.9%), expert/novice (3.8%), and dominant/dominant (1.6%). Statistical tests compared the dyads’ gameplay interaction patterns on a level.

Results indicated that teams tended to nudge the objects they drew more frequently if their interaction pattern was collaborative relative to dominant/passive (p = .011). While there were no differences where interaction patterns were more likely to obtain a gold trophy, teams were more likely to earn a silver trophy if their interaction pattern was collaborative as opposed to dominant/passive (p = .035). Next steps will be analyzing the process data to explore how gameplay features are associated with content knowledge and self-perceived performance. These results will be reported at AERA.

Authors

• Jessica Andrews Todd, Educational Testing Service

• Yang Jiang, Educational Testing Service

• Laila Naylor

• Sofia Wilson, The University of Texas - San Antonio

• Monica Toscano

• Jonathan Steinberg, Educational Testing Service