It is virtually impossible to design a convincing study on the effect of FoodstuffA on human health. The ideal study would look something like that: - Enlist pe

It is virtually impossible to design a convincing study on the effect of FoodstuffA on human health. The ideal study would look something like that: - Enlist people into study, divide them randomly into two groups - Design a diet, where everything will be exactly the same, except FoodstuffA for group 1 will be replaced with some FoodstuffB for group 2. - Keep all subjects under supervision 24/7 to eliminate possible cheating - Continue the study long enough to determine long term effects (years? decades?) Even in this case, there will be opportunity for speculation. Maybe the FoodstuffA is not bad, but the FoodstuffB is just beneficial, maybe we should have used FoodstuffC. So, there's two alternatives that are realistic: 1. Observational study. Ask 1000000 people about how much red meat they eat, compare health, do conclusions. Lots of confounding variables - smoking, physical activity, alcohol consumption etc etc. 2. Randomised controlled study. Will be fairly short due to funding and subject to criticism due to design of the diets of participants (no, group 1 had too much fat! no, group 2 had too little carbs!). Both options are not ideal and subject to controversy.