Here’s How the Media Turns Research Into Misleading Clickbait

Photo by Good Good Good on Unsplash

Here’s a fact that probably won’t surprise you:

Exposure to social media has a negative effect on teen depression.

Now, here’s a fact that probably will surprise you:

According to a leading researcher, “eating potatoes has the exact same negative effect.”

I’m not kidding!

When you look at the actual research, it turns out that the effect of social media on teen depression is extremely small. Specifically, social media can explain 0.36 percent of teen depression symptoms. And even then, the study only found covariance among girls, there was none for boys.

That means 99.64 percent of teen girls’ depressive symptoms aren’t related to social media. Similarly, 99.64 percent of depressive symptoms aren’t related to potatoes. Not only that, but listening to music — any music — has 13 times the negative effect on teen depression that social media does.

Weirdly enough, I haven’t heard an outcry about kids’ access to music, or to french fries. Have you?

The disconnect here is that people tend to care more about whether there’s “an effect” than how big that effect is.

In statistics, we call this effect versus effect size.

Whenever we learn that something has an effect, it’s good to ask the follow-up question, “How much of an effect does it have?”

In the case of social media and teen depression, the answer is, “Not very much.” Many other causes of teenage depression are more significant than social media. For example, getting fewer than 8 hours of sleep has at least three times the effect that social media has.

Of course, talking about that isn’t much fun. And it’s even less fun to sift through thousands of different variables with a small effect size and find the real culprits. That’s nuanced work that requires patience and a fine-toothed comb. It’s more convenient to say there’s “an effect,” blame social media for teens’ depression, and call the case closed — even though that’s not the whole story.

Here’s another situation in which effect size is routinely ignored: the impact of video games on violent behavior. It would be…