Testing an Integrated Team Effectiveness Framework in League of Legends

Abstract

Objective: This paper investigates factors impacting team performance in the Multi-player Online Battle Arena gaming environment, League of Legends™, by testing an integrated Input Mediator-Outcome team effectiveness framework. Background: Secondary data and Naturally Occurring Data Sets (NODS) are data that have been collected from respondents without research interests in mind and can occur naturally in the environment. There are numerous sources of secondary data, including government data, financial databases, industry association groups, and Application Programming Interfaces, which this research utilizes to study the performance of teams. Methods: Path Analysis and Partial Least Squares Discriminant Analysis (PLS-DA) are analytical methods that are well suited for large data sets and sample sizes, confirmatory in nature, and can test a theoretical model. This research utilizes both in order to study factors impacting team performance. Results: A total of 5,927 matches from 742 teams are sampled and analyzed. Six team performance measures are used to discriminate between winning and losing teams, including role familiarity, team familiarity, team effectiveness, team efficiency, and the Kills, Deaths, Assist (KDA) ratio. Using path analysis and supervised PLS-DA, the models led to the successful prediction of 89.4% of the matches. The error rate for the PLS-DA model is 0.106 (Q2 = 0.523; R2 = 0.551). Conclusions: This work shows how objective, detailed data on teamwork may be used to provide insights into questions of the performance of teams. Additionally, the results demonstrate the value of using path analysis and PLS-DA to test an integrated framework. Application: This research highlights the value and feasibility of studying virtual teams for new insights into team performance.