In an earlier blog post, I reported on an analysis of 34,000+ Americans’ health interviews with the Centers for Disease Control and Prevention (CDC). To my astonishment, Megan Traffanstedt, Sheila Mehta, and Steven LoBello found no evidence that depression rises in wintertime, or that wintertime depression is greater in higher latitudes, in cloudy rather than sunny communities, or on cloudy days. Moreover, they reported, even the wintertime “dark period” in northern Norway and Iceland is unaccompanied by increased depression.
Given the effectiveness of light therapy and the acceptance of major depressive disorder “with seasonal pattern” (DSM-5), I suspected that we have not heard the last word on this. Indeed, criticism (here and here) and rebuttals (here and here) have already appeared.
Reading Seth Stephens-Davidowitz’s wonderful new book on big data mining inspired me to wonder if Google depression-related searches increase during wintertime. (To replicate the CDC result, I focused on the United States, though further replications with Canada and the UK yielded the same results.)
First, I needed to confirm that Google Trends does reveal seasonally-related interests. Would searches for “basketball” surge in winter and peak during March Madness? Indeed, they do:
We know that Google searches also reveal seasonal trends in physical illnesses. And sure enough, “flu” searches increase during the winter months.
So, do searches for “depression” (mood-related) and “sad” similarly surge during wintertime? Nope, after a summer dip, they remain steady from mid-September through May:
Ditto for Google entries for “I am depressed” and “I am sad.”
My surprise at the disconfirmation of what I have taught—that wintertime depression is widespread—is like that experienced by my favorite detective, Agatha Christie’s Miss Marple: “It wasn’t what I expected. But facts are facts, and if one is proved to be wrong, one must just be humble about it and start again.”
[Note to teachers: you can generate these data in class, in real time, via Trends.Google.com.]