'Git Gud!' -- Evaluation of Self-Rated Player Skill Compared to Actual Player Performance

This paper evaluates the efficacy of self-rated skill as a method of differentiating player performance by analyzing data gathered in 4 previous user studies. Analysis confirms that self-rated skill can be effective for differentiating actual performance on average, but that it is not necessarily predictive for every game, and that while player performance is comparable across gender, few male participants self-rated at the lowest skill level, and no females self-rated at the highest. Additional findings suggest having participants self-rate on a five point scale, but applying those ratings in three tiers may be effective for differentiating game performance by player skill level across gender.