In a recent article for the New York Times,
Seth Stephens-Davidowitz published a fascinating story (Google Searches Can Help Us Find Emerging Covid-19 Outbreaks) about using Google Trends to find coronavirus outbreaks in other countries, as well as finding unexpected side-effects of the virus.
A few people then asked me if Google knows this, why doesn't it tell us? The short answer is "you have to go looking for it." Here's why:
The article by Stephens-Davidowitz is the story about his use of Google Trends to discover what people are searching for about coronavirus. That is, he had the clever idea to use the Trends data to DISCOVER these effects in the data. The Google Trends system has all kinds of phenomena in it as revealed through the lens of what people are searching for.
As an example, look at this Trend graph for "kayaking" in the web search dataset for the United States.
That's not a surprise. Kayaking searches surge in the summertime. Makes sense. There are thousands of such observations that can be made by looking at the Google Trends data.
But in the NY Times article, Stephens-Davidowitz points out that a similarly surging query in Trends is [I can't smell] and an even more obviously surging query is [loss of smell].
Here's the Trend graph for this query [loss of smell] (US only, past 30 days):
As interesting as this is, it's even more dramatic when you take into account more of the temporal context. Here's that same query trending over the past 5 years:
That uptick at the end is an amazingly large response and puts the current search query volume in context. This year is NOT like other years.
Likewise for other common COVID symptoms:
But notice this chart (below) about fever--see the annual cycle in the wintertime? That's to be expected. Yes, there's an uptick this year (far right side), but seeing the context of the previous five winters' worth of fevers makes the severity of THIS year much more apparent.
This is true in other countries/languages as well. If you ask about fevers in Italy (using the Italian word febbre), you see something similar.
Same thing happens with another COVID symptom--fatigue. There's an annual drop in fatigue queries right at Christmas (I've marked Dec 25 of each year with arrows). It's not a huge signal, but interesting nonetheless.
|Searches for [fatigue] over the past 5 years. The arrows point out the lowest volume of searches across the year, which happens to be at Christmas. I'll let you figure out why this might be true.|
Even so, there's a distinct uptick in [fatigue] queries as well this spring, even though it's not quite as obvious as the other ones.
There's a great of information within Google Trends--but you have to go looking for it.. it won't magically raise its hand and say "here I am!"
A really interesting post to read on this topic is Coronavirus Search Trends (written by the Google Trends team) that shows the latest breakout queries at both a national and worldwide level. These are queries that have suddenly spiked in volume--a real indicator of what people are interested in finding out more about!
So while Google Trends is a great resource for researchers, you still have to go digging a little bit into the content in order to see what's happening.
Enjoy exploring with Trends. Let us know what you find!