My Bachelor Thesis
"Decreasing Player Frustration using Peak End Rule on Game Graphics"


Frustration is an important topic when designing video games, if games are too hard, players might get frustrated and stop playing. But if they are too easy, it will be boring, resulting in the same outcome. So it is important to take the so called Flow Theory into account when designing a video game. To lower player frustration, this thesis describes an alternative approach to applying the Peak End Rule on video games. Without changing the difficulty level, I examined the effects of modifying the graphical fidelity in a casual game, corresponding to the Peak End Rule, which means using a higher fidelity, for the middle and the end part of a level and a lower fidelity (a blocky-look) for the rest. For this, I developed a small casual game, called 'MoonLeaper', that is based on the well-known and easy to learn mobile game 'Doodle Jump'.

For my study I tested four different versions of the same level individually in between groups. The four conditions only differ in their graphical fidelity throughout the progress of the level: condition 1 "only high fidelity", condition 2 "only low fidelity", condition 3 "applied Peak End Rule" and condition 4 "applied inverse Peak End Rule". The participants played one random condition and assessed their experience in a questionnaire based on "The Player Experience of Need Satisfaction" (PENS), "Self Assessment Manikin" (SAM) and the "Intrinsic Motivation Inventory" (IMI) afterwards.

The results (state 13.04.2021) show, that there is a significant difference in condition one and two (Z = -2,337, p = 0,0194), regarding the Effort/Importance subscale from IMI. Thus, the participants that played the second condition put more energy into the game and it was more important to them to succeed (median c1 = 3,3, median c2 = 4,2). In condition three and four, there was also a significant difference in the Interest/Enjoyment section of IMI (Z = 2,493, p = 0,0127). Condition three (median = 5,214) is more enjoyable than condition four (median = 4,714). For the time measure, there is a significant difference between condition two and three (Z = 2,319, p = 0,0204). The second condition took 111,1 sec on average, while the third condition took 94,3 sec. So there is proof, that graphical fidelity and usage of the Peak End Rule does affect player experience in some way, but the data analysis is still in process. I will present both quantitative and qualitative results in this thesis.