Given that a common claim made for fasting is that it will be easier to stick to, that is sort of a valid caveat, though. Sure, anything works better if you do it than if you don't do it. Using people randomly assigned to Longo's style of fasting isn't really generalizable to other forms of fasting. Especially when they're not being controlled to another restricted style of eating, the control there was people eating whatever they bloody well pleased, and still ten percent dropped out.


Some things are difficult to look at by scientific method. Randomization is a good thing--but how then do you study the effect of self-selection of a plan on individual adherence?