Social network indices as performance predictors in a virtual organization

Abstract

Measuring an individual's potential to accomplish a goal has key implications in organizational studies, behavioral analysis, business and management among other areas. This relative potential, which is dependent on a myriad of factors, e.g., experience, resource management, cooperation, etc., could shape the individual's performance, motivation, leadership, and likely success. The increasing prevalence of virtual organizations provides unique challenges and opportunities to study complex research questions via convenient data collection. In this NSF-funded research, we introduce a multi-factor dynamic model to estimate the potential of an individual to accomplish a goal in a virtual organization. We analyze data from a Massive Multi-player Online Role-Playing Game (MMORPG), Travian, collected in a controlled environment for over 3.5 months. The data contains activities of 7,406 players, including, players' profiles, daily snapshots of individual and alliance attributes, amount of gold, military strength, trades and cooperation, diplomacy status as well as a 2.3 million-message graph. We model players' potential to survive as a multi-factor function derived from the data attributes, including SNA-based network measures from the message graph. The model predicts a player's potential to win the game or to achieve a specified goal. The target is to identify the influence of network behavior for the player, and how network statistics for each player influence the accuracy of the prediction.