While I understand your perspective one has to be extremely careful in deriving conclusions from such data or analysis. The only way to be able to approximate

While I understand your perspective one has to be extremely careful in deriving conclusions from such data or analysis. The only way to be able to approximate a supportable conclusion would be a double-blind study with randomly chosen and normally distributed individuals in both the control group and the group being studied for the effect in question. Isolating for the variable under study is one of the cardinal rules of this kind of statistical analysis. The simpler statement is the often-quoted, correlation is not causation. These things respond to dozens, hundred, thousands of variables and are difficult to analyze without adherence to strict statistical methods.