Correlation vs Causation

An MStranslate “Understanding Science” article by Dr Natalie Thomas

“I’m not saying that the number of films Nicholas Cage appeared in caused more people to drown in swimming pools… But I am saying that when Nicholas Cage appeared in more movies, more swimming pool drownings happened…”

Many articles relevant to multiple sclerosis speak to what causes MS, and with good reason. Understanding the causal factors involved with the disease suggests we’re closer to possible treatments and preventative measures. But are many of the studies we refer to actually investigating what is correlated with MS? And does it really matter?

Consider these two factors as examples: the number of people who drowned falling into a swimming pool and the number of films Nicholas Cage appeared in. The first question we could ask is if there is a relationship between the two; are they correlated? A positive correlation means that when one thing goes up, the other goes up too. A negative correlation is when one factor goes up, the other goes down. As demonstrated by the graph, the number of people who drowned by falling into a swimming pool DOES correlate with the number of films Nicholas Cage appeared in. Statistically speaking, it’s a strong correlation too. However, correlation clearly must always be taken into perspective; are there any other possibilities that could result in the similar rate of drownings in swimming pools and Nicholas Cage movies? In order to infer causation, a true experiment must be performed where subjects are randomly assigned to different conditions, whereby the many variables or other possibilities, can be controlled for. For a memory surge on what controls are and why they are important, refer to the link in the comments for our video on controls. A controlled experiment in this regard would require a population living in a location where no Nicholas Cage films are screened, but where both locations have the same number of swimmingly pools and where all other variables are controlled for. So, good luck conducting that experiment- but it does a beautiful job of explaining that whilst these two factors may be related, we can’t get too ahead of ourselves and suggest that Nicolas Cage’s movies cause drownings in swimming pools. Now, obviously in this case it is easy to make this distinction because the link is somewhat nonsensical, but when the two variables seem to ‘fit together’, people are drawn to believe that the relationship is causal.

So why is this mix up so prevalent in the media’s reporting- especially in media headlines? Let’s imagine you’re flicking through the endless articles published online, news outlets, twitter, etc… Which would catch your attention first? “One study shows that when Nicholas Cage appeared in more movies, the number of people drowning in swimming pools increased” or “Nicholas Cage movies cause swimming pool drownings.”

Although researchers are careful to shape their findings appropriately, much like the first headline, the media often transforms it to a much flashier, but incorrect proposition. I don’t think this is always done in a malevolent fashion to merely sell more papers. Often, no doubt, the reporters don’t completely understand the distinction themselves.

The last point on this topic worth commenting on is why scientists spend so much of their time conducting correlational studies if they create so many problems? Put simply, whilst no scientist believes that correlation necessarily means causation, these tests are the main way to evaluate observational studies in big cohorts.

With this all in mind, we don’t want to commit the opposite fallacy – dismissing the importance of significant correlations. This statistical approach provides supporting evidence that improves our overall understanding. We do, however, need to remember the difference between correlation and causation, and remember to read between the lines of any media headline.