Oleg discusses how Bayesian methods can address the small, mixed datasets typical in game user research.
SPEAKER: Oleg Jarma, Ph.D. fellow at ITU.
ABSTRACT: Games User Research (GUR) is a young and eclectic field, drawing from psychology, HCI, and game design, and routinely producing small, mixed datasets that combine behavioral metrics with subjective player reports. This is precisely the kind of data that Bayesian methods were built for, yet they remain far from standard practice in the field.
This talk presents two cases where Bayesian analysis added genuine value to GUR goals: one uses Bayesian mediation models with Dirichlet mixtures to unpack how game difficulty shapes a player’s perceived challenge; another applies a hierarchical Bayesian model with Gaussian Random Walks to trace how each player’s affective state evolves across a session, and how confusion moderates the link between momentary experience and overall enjoyment.