Ice Cream and The Zodiac: How to Read Articles Like a Researcher

As a researcher myself, I tend to be skeptical of online/media voices that cite surprising statistical findings. Despite even the best intentions, sometimes people who lack training in research methods and statistics can accidentally misrepresent findings. One misleading headline later, the general public is terrified that everything in their fridge will give them cancer. Talk about lost in translation. 

How we read and interpret what we find online can really change the way we lives our lives. It can impact what foods we choose to put in our bodies, ways we try to support our physical/mental health, etc. So, it’s immensely important for us to discern whether the facts are being misrepresented. 

To illustrate my point, I’ll use a study which found that elevated ice cream sales were linked to higher incidences of homicide. Let’s pretend this article caught the attention of an eager journalist who then published the following: “Ice-Cold Killers: Dessert Consumption Causes Homicidal Rampage!” This definitely puts a different spin on the “I scream, you scream, we all scream for ice cream” song. Yikes. 

Photo by Sittisak C on Pexels.com

Ice cream and murder may very well coincide with each other, but this does not mean that a few scoops of Vanilla Bean will turn you in to the Zodiac. How is this the case? Well, because when two things are correlated with one another, this means that as one variable changes (goes up or down), the other variable changes as well. It does NOT mean that one directly causes or leads to the other. 

Think of it this way: If all we know is that two things are related to one another, we still need to answer the following questions.

  • Does eating ice cream make you want to murder someone? 
  • Or, does committing murder give you a particular craving for ice cream? 
  • Or, is there no causal relationship between these two?
  • Are these two instead both caused by something else?

In the case of our example, neither variable causes the other. Instead, they are both a result of a third (also called a confounding) variable: HEAT. When it gets hot outside, people start to crave that cold and refreshing taste of sweet cream. Separately, hot weather also tends to make people a little more angry: getting all sweaty and uncomfortable apparently makes people more likely to act on their ugliest impulses. 

My preposterous ice cream headline illustrates the importance of distinguishing between correlation and causation. So, if you read an article that claims that depression is related to higher yoga class attendance, should you swear off your weekly Vinyasa flow for the sake of your mental health? Or, is it possible that some people with depression simply tend to find solace in the compassionate yoga community? In this case, you are very safe to nama-stay in your yoga class.

The examples I used in this post make the difference between correlation and causation pretty obvious, but sometimes the language that articles use is more muddy. My point is, the next time you come across research findings that make you want to move to Fiji, put on your researcher hat and ask the following questions: Is this relationship correlation or causation? What else might be happening here? Is there any other explanation?

You’ll read a lot of things on the internet. Think critically before deciding to swear off ice cream.