How Do Players Experience Win/Loss Patterns? An In-The-Wild Study of a Casual Mobile Game

This study investigates how the sequencing of win and loss outcomes shapes player experience in games, a question that has received little systematic empirical attention despite widespread design intuition that event order matters. Drawing on prior work in game studies, human-computer interaction, psychology, and sociology, the authors frame the problem around early wins, streaks, and the ordering of events, and argue that existing guidance from flow theory, narrative frameworks, and operant conditioning research remains too broad to inform concrete design decisions about pacing and reward sequencing.

To address this gap, the researchers conducted an in-the-wild, between-subjects experiment using Pizza Time!, a casual single-player mobile game co-designed with a professional studio and released in one national market. Each level contained ten pseudo-random events split evenly between wins and losses, and data from 94 players were analyzed across three sequence categories: prefix/suffix, streaks, and alternations. Measures included play duration, level progression, continuation, and in-game ratings.

The findings suggest that prefix/suffix patterns replicate primacy and peak-end effects, with sequences that begin or end in wins producing higher continuation. Loss streaks, but not win streaks, significantly affected play duration and continuation, indicating that clustered defeats discourage players from persisting. The authors interpret early-win bias as a possible route to an illusion of control and suggest that streak-breaking events may exploit peak-end salience, while recency effects remain untested.

The principal contribution is a tentative taxonomy of binary event sequences that appear to influence player behavior. The authors acknowledge the limitations of this exploratory work and call for experimental manipulation of these categories to establish causality, application in richer skill-based games and diverse cultural contexts, and finer-grained behavioral and self-report measures. They conclude that these insights may inform pacing, dynamic difficulty adjustment, and reward sequencing across interactive experiences.