Time to Play?

Abstract

The online session behavior of users in a multiplayer online role playing game is examined. Three dimensions of online sessions are studied: frequency of sessions, distribution of start and stop times, and duration. These measures are modeled and examined econometrically to understand their relationships with attributes of the users and their avatars. Frequency is analyzed by using a negative binomial regression to uncover relationships with contributing factors, such as the experience level of players and the number of avatars per user account. Start and stop time distributions are examined aggregately with discrete choice models that capture the probability of starting a session in relation to the sociodemographics of users. Duration models are estimated by using data on the characteristics of events that occur within the duration and other user sociodemographics and player attributes. Since sessions are not explicitly indicated within the data set, sessions were stitched together on the basis of an assumed threshold for offline duration. Overall, these models and results provide insight into temporal characteristics of online sessions, including their aggregate traffic characteristics, such as frequency, headways, and duration length. These results also provide information on the nature of online game playing from a user perspective.