Unless the subjects were wrestlers or models, why should the choice of fruit v. cake involve self control at all? For a fair number of people in the US, choosi

Unless the subjects were wrestlers or models, why should the choice of fruit v. cake involve self control at all? For a fair number of people in the US, choosing fruit over cake is largely a matter of willpower. (I hope you're not arguing that a significant fraction of the population needs to exercise willpower in order to choose cake over fruit!) You could argue that that percentage isn't significant enough for the experiment's sample size, certainly. But otherwise, all things being equal, there should be roughly the same number of people who need to exercise willpower to choose fruit over cake in either group.