It's far too small of a sample size for something as complex as human diet and exercise. You'd need a much larger sample size with people of varying types of diet and exercise regimes.