“Hot Hand” on Strike: Bowling Data Indicates Correlation to Recent Past Results, Not Causality
Panels A–B: The conditional probability for a strike in the 9th (green) and 10th (red) frames given the number of strikes in the first 8 frames is plotted. The 95% confidence intervals shown are calculated using bayesian framework with Jeffreys prior (see ). Data is shown for two individuals with the most identifiable patterns. The numbers at the title of each panel represents the player's code according to the supporting information S1. Dashed lines represent the base probability of success for each player in the corresponding frame and dotted lines show the linear fits results. Panels C and D show the distribution of slopes () and obtained from the fits of all players respectively.