I feel part of my job with this blog is to pass along some basic knowledge on how to interpret studies. It really is not as hard as it sounds and it can really help you make heads or tails of that study that you see published in the morning paper. There is a lot of bad science out there for various reasons. When we talk about observational studies, we are simply looking for patterns. Observational studies do NOT prove causation and can be very misleading! The classic example that I was given in med school is this: You can run a study that compares rates of coffee consumption and rates of lung cancer. If you run a retrospective (looking back in time) observational study like this, you would conclude that the more coffee you drink, the higher your risk for lung cancer. Now we know this is ridiculous but how can this possibly be explained? Well, it just so happens that if you are a coffee drinker, you are more likely to be a smoker. These confounding variables come up in observational studies ALL THE TIME. In fact, when people worry about meat consumption and the health risks purportedly linked to it, we will see that the studies used to make that conclusion are fraught with confounding variables. Particularly a confounder called the “compliance bias” — which we will get into in a later post. For now, here is a wonderful video lecture by Tom Naughton which I think does a great job on explaining some of the problems with these studies. Finally, remember this: the gold standard study in science is the Randomized Controlled Trial. An RCT is a prospective study (a study that looks at data going forward in time). It isolates one variable (a test, treatment, procedure, etc.). Participants in the study are randomly assigned to either the “treatment group” or the “control group”. If you know you are looking at an RCT, you are more than likely looking at an excellent study.