A. “Scientists tend to be quite specific about cause-and-effect relationships or ‘causality,’ which means having strong evidence that ‘X definitively causes Y’,” says Adela Hruby, PhD, MPH, a scientist in the HNRCA Nutritional Epidemiology Program. “Strictly speaking, only randomized controlled trials in humans can demonstrate causality, but other study types, like cohort studies, can provide evidence supporting causality.
“Imagine you want to study whether owning cats causes happiness. In the simplest randomized controlled trial, a researcher randomly assigns volunteers to either a ‘control’ or a test ‘exposure.’ In our study, the control is not having cats and the exposure is having cats. Importantly, volunteers don’t choose whether they get a cat, which could bias the results. If the study is done well, by the end of it, you could reasonably conclude that any difference in happiness between the two groups was due to cat ownership, since that was what distinguished the groups.
“In a study showing associations, participants are not assigned an exposure. Rather, the exposures are ways the participants have chosen to live, including having children, cats or dogs, vacations and so on—all factors that affect happiness.
“Although researchers can statistically account for the non-cat factors, they cannot completely rule out these influences on happiness. This is why association studies can provide evidence supporting relationships between cats and happiness, but cannot prove that cats cause happiness.
“These studies are nevertheless critical in nutrition and many other sciences. For example, they allow us to do things that trials often cannot—like examine effects of things researchers can’t ethically assign to people, like smoking.”