A few days ago, I received a WhatsApp message about the relationship between drinking wine, eating red meat, and coronary disease. At the end of the message was the conclusion that speaking English caused coronary disease because the French and the Italians ate red meat and drank red wine but did not die of the coronary disease. It was a joke of course that easily summoned a chuckle. However, there is some food for thought there especially in the light of studies like taking Tylenol during pregnancy can cause autism in the offspring. Or acid reflux during pregnancy has a positive aspect -- the child will have a head full of gorgeous hair! No, that is not an old wives tale, but an actual study conducted at Johns Hopkins University. The most recent of these correlation studies is about linking atheism to autism. I am neither an atheist nor autistic, but the article made me cringe. I did not have acid reflux during either of my pregnancies, but both my babies had lovely hair; I often sneer when I hear someone talk about this study. A couple of years ago, a study linked less sleep in teens to depression and cognitive imbalance. Is that not true for everyone, and not just adolescents?
The importance of statistics in research does not need any proof. All kinds of researchers use some statistical design to help analyze their data and draw conclusions. The “scientific method” involves making predictions before making observations. Unusual observations entail an explanation, which is tested through experimentation followed by replication to draw a conclusion. This sounds simple, right? The problem arises when statistics becomes a science instead of a scientific method. Personally, I do not like the idea of science, logic, and reasoning, being hijacked by correlation. For example, why did someone even hypothesize that atheism could be linked to a neuro-developmental disorder? If people's ideas do not conform to a particular popular belief, is it because of a neurological, developmental, or metabolic disorder? One can always form a hypothesis, take a random population sample and correlate it to something to draw a conclusion, but is that science? Where is the cause and effect? "Correlation does not imply causation" is a well-known phrase in statistics, which highlights the fact that correlation between two variables does not imply that one causes the other. The assumption that correlation proves causation is questionable even in statistics. Nature magazine maligned the P values (the gold standard in statistics) in February 2014 to confound the “scientific method” further. So, why are all these researchers wasting time and money on correlation studies is anybody’s guess.
So, do articles that question correlation studies, make people angry? I do not know, but your comments will help. Simply write a yes or no in the comments and I shall analyze the data (provided there be enough for a correlation study). What is life without a little whimsy, eh?
