That would not be a very informative study. The experimenter could not control for factors that might both have direct health outcomes and increase the likelih

That would not be a very informative study. The experimenter could not control for factors that might both have direct health outcomes and increase the likelihood that someone prefers red wine. Wealth and being Italian are the obvious ones there, but there are sure to be others.