Many people have tossed a coin, but few have stopped to ponder the statistical and physical intricacies of the process. A preregistered study collected 350,757 coin flips to test counterintuitive predictions from a physical model of human coin tossing developed by Diaconis, Holmes, and Montgomery (DHM; 2007). This model claims that when people toss a regular coin, it tends to land on the same side it started with. DHM estimates that the probability of a same-sided result is approximately 51%. Our data strongly supports this accurate prediction. This means that the coins often land on the same side. public relations(same side)=0.50895% confidence interval (CI) (0.506, 0.509), BFipsilateral bias=2364. Furthermore, the data revealed that the degree of this ipsilateral bias varied considerably between people. Our data also support the general prediction that when people toss a regular coin (the first side is randomly determined), there is an equal chance of getting heads or tails. public relations(Head)=0.50095% CI (0.498, 0.502), BFFront and back bias=0.183. Furthermore, this lack of heads-tails bias does not appear to differ between coins. Our data thus provide strong evidence that when some (but not all) people toss a fair coin, it tends to land on the same side as it started. Our data provide convincing statistical support for her DHM physical model of coin tossing.
by František Bartoszothers, i.e. paper From the end of last year.