Flaws in group selection Andreas Zourdos, University of Westminster 21 July 2014 As we can see in the sociodemographic characteristics of the groups of the study, the fasting group is also an almost non smoking group (with only 7% of the group smoking), while the non fasting group is clearly smoking a lot more, with 55% of the group being smokers. Sarri and colleagues are giving the fasting group an unfair advantage on the health markers that they want to study. Smoking can and will affect serum total cholesterol, serum triglycerides, LDL-C level and will lower HDL-C when compared to non smokers (Batic-Mujanovic et al, 2008). In order to have a fair comparison of two diets, and/or to see if fasting per se does any good, we must create an "all things being equal / ceteris paribus" experiment. The group design is so unequally created that it's impossible to rule out whether it was fasting that affected the blood markers, or the lack of smoking. Competing interests Author declares that he has no competing interests.