big data | April 25, 2013

What Causes What? Charles Wheelan Talks Causation In Stats On NPR

In our daily lives, we are often quick to confuse correlation with causation. As big data speaker Charles Wheelan discusses on Planet Money (a radio segment hosted on NPR), the distinction between the two is an important one to make. Searching for causal relationships has helped us to evolve as a species, and is a very important part of how we live day-to-day. Everything from the decisions parents make for their children and public policy officials make for the public, to determining what has causes illness or pain, is impacted by our understanding of causality. Sometimes, however, we are so quick to seek out what caused something to happen that we mistake correlation for causation. Just because one factor has a relationship with another does not mean that one has caused the other.

It is important to realize this distinction so that we can make better sense of the studies we read and the decisions we make. One example he provides in the interview is an experiment regarding the effects of estrogen replacement therapy on women. The initial study discovered what seemed to be a logical causal relationship between an increase of estrogen and the improved health of the women surveyed. However, it turned out that the positive relationship was correlational and not causational. After doing a secondary study, they actually discovered that the women given estrogen experienced decreased health as opposed to the control group who was not provided with the hormone. After comparing the two studies, it was discovered that the estrogen itself was not the key reason women were fighting illness better—there were other factors at play. This is why it is important to pinpoint whether two variables are simply varying together, or whether one has indeed an impact on the other.

In his book, Naked Statistics, Wheelan uses real-world examples to explain complex statistical theories. Not only do these case studies help us understand the world of stats—they show us why we should want to understand them. By applying these concepts to examples that are relevant to our lives, Wheelan "strips the dread out of data." The author of Yahoo!'s popular "Naked Economics" column and a lecturer on public policy at the University of Chicago, Wheelan is adept at sharing his work both in his writing and on stage. His talks on big data and statistics are not only extremely useful—they're fun too. He shows us that anyone can—and should—understand stats, and provides his audiences with the tools they need to make more informed decisions.

Up Next

design | April 24, 2013