Related Experiment Video
Updated: Aug 14, 2026

Measuring Engagement of Spectators of Social Digital Games
Published on: July 3, 2021
What makes players stay: Using real gameplay data to explore players' game loyalty to E-sports MOBA games
Haojian Li1, Lingji Li2, Erjia Xu3
1Department of Communication, University of California, Davis, 1 Shields Ave, Davis, CA, 95616, USA.
Abstract:
In recent years, esports and Multiplayer Online Battle Arena (MOBA) games have gained substantial global popularity, yet the mechanisms underlying player loyalty in the MOBA context remain insufficiently integrated within a unified theoretical framework. Drawing on the Stimulus-Organism-Response framework, this study proposes a three-stage model in which game cooperativeness and competitiveness serve as stimuli, self-efficacy and immersion as organism variables, and game loyalty as the response. Using DOTA 2 as a representative MOBA game, we conducted two studies to test the proposed model. Study 1 surveyed 519 DOTA 2 players and employed structural equation modeling. The results showed that competitiveness predicted game loyalty through the sequential mediation of self-efficacy and immersion, whereas cooperativeness influenced loyalty through self-efficacy but not immersion. Study 2 examined an independent sample of 152 DOTA 2 players using gameplay data to assess whether a key pathway identified in Study 1 could also be observed through behavioral indicators. In-game metrics, including kills, assists, character difficulty, damage, and matches played, were mapped onto the theoretical constructs examined in Study 1. The results supported one salient pathway: character difficulty, as a behavioral indicator of self-efficacy, mediated the relationship between kills, as an indicator of competitiveness, and matches played, as an indicator of game loyalty. These findings clarify the psychological mechanisms linking cooperative and competitive game features to player loyalty in the MOBA context, offer practical implications for game design, and demonstrate the potential value of gameplay data as behavioral indicators of theoretically relevant constructs.
