One cannot possibly design proper controls with human behavior. Even in pharmaceutics, confounding variables are already common place that led to a considerabl

One cannot possibly design proper controls with human behavior. Even in pharmaceutics, confounding variables are already common place that led to a considerable replication crisis, with human behavior it would be even more impossible to control for every variable, and measure only what one wants to measure. A very simple problem with such a design which would perhaps indicate that human beings that are not allowed to be kind would be less happy is simply that human beings become unhappy when they are told what to do, not necessarily anything related to kindness; it's certainly not implausible. And the fundamental problem remains that a man cannot be compelled to participate in any study, and that alone is a selection bias that cannot be overcome — studies thus select upon the kind of person who have the time and willingness to participate in them.