Human actions are caused. But they are not necessary. When faced with hunger, you can eat, or starve like a hermit. So in any human interaction never expect any particular action. There is nothing inevitable about human behavior. All actions are evitable.
The strength of statistical methods lies not in their predictive capacity about a particular behavior, but in maximizing the expected utility over all possible actions. However even such a distribution describing the prevalence of human actions, is not everlasting, like the laws of physics. A mere understanding of well known distributions of human behavior, can render those distributions invalid.
Do we all know that market returns are normally distributed? Too bad. Because we all know it, it is not so anymore.
Much like how you would not go near pool tomorrow if a legitimate psychic (i.e. a scientist) predicted that you would die in a pool tomorrow, if a scientist described the prevalence of death in pools, you wouldn't go near a pool often.
However this has not stopped anyone from looking predictable behavior. Behavioral economists are forever on the lookout for predictable irrational/rational tendencies in individual behavior in a crowd.