The Numbers Don’t Lie—Do They?

correlation chartDo any of these conclusions sound familiar? They’ve all appeared in the news at one time.

  • Children living near power lines have higher rates of leukemia; therefore, the electric field around the lines causes cancer.
  • Because the number of children diagnosed with autism has climbed at the same rate that the number of children receiving vaccines, we can conclude that vaccines cause autism.
  • The rise in global temperatures at the end of the 20th century is due to the increased use of fossil fuels in that same period.

While it’s true that in all these cases (and I could list many more), the two statistics correlate, that really doesn’t tell you whether or not one is caused by the other. In fact, as every good statistician knows, correlation doesn’t imply causation.

When we read the morning paper, or click on the latest alarmist news story, it’s important to ask ourselves if the conclusion is based on a correlation. Let’s take these three examples and look closer.

It’s true that children living near power lines have higher rates of leukemia. However, when researchers measured the actual strength of the electric fields inside the homes, the correlation disappeared. They concluded that it wasn’t the power lines that were causing the problem. Since homes near power lines usually cost less, perhaps there was something relating to the families’ lower economic status that led to increased rates of cancer. That’s a correlation that appears over and over, and probably has some validity. The children might eat fewer vegetables, for instance. We just don’t know.

Then there’s the autism-vaccine link. It’s been repeatedly debunked, with hundreds of studies. The original paper making the claim has been withdrawn, with the author admitting he made the whole thing up. Yet, we hear someone claim that their child was diagnosed with autism right after he received his vaccines, so the shot caused his autism.

In actuality, there has been no rise in autism. The definition of what comprised autism changed, and what was considered a single condition is now described as a “spectrum.” As a result, children who would have once been labeled “difficult” are now called autistic. It’s the number of diagnoses that has increased.

Then there’s the whole issue of climate change. According to some measurements (and this topic alone could be an entire book), global temperatures have risen. So has our use of fossil fuels. But again, correlation doesn’t always imply causation. It might be the CO2 levels. It might be because we’re still coming out of a little ice age (the world was significantly cooler during the Roman Empire). It might be due to sunspots (and there’s a strong correlation here as well). Or, it might be something we haven’t thought to investigate.

Whenever I see a claim that science has “proven” something, I tend to be skeptical. Science rarely proves anything. Rather, it suggests. It proposes. It tests. And it’s willing to be wrong.

There’s a website (now also a book) that you have to see. It’s a list of graphs showing a very strong correlation between all sorts of interesting trends. The graph at the top of this page is one example. Go ahead, take a look, laugh. Then keep those graphs in mind the next time you encounter a news story insisting that x causes y.

Spurious Correlations

Advertisements

Whatcha think about this?

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s