Using Google trends to examine food-related behaviors and concerns under COVID-19

A screenshot of a social media post Description automatically generated

Ok, what is Google trends? Google trends is a free analytical service tool that shows what people search for across the world. It can also show searching trends by sub-regions and cities within a country. This is going to be impacted by access to technology and communication, so a lot more “pull” of this data will occur in urban areas in Low- and Middle-income context.

In the book Everybody lies, Author Seth Stephens-Davidowitz talks about how Google trends in the United States can be used to inform the prevalence of gay Americans, half of whom “are in the closet” in certain states. Oh, my favorite Google trends data story is what women around the world google during pregnancy. In the US, searches are related to whether women can have coffee or wine or Tylenol or “how to prevent stretch marks” or “how to stay fit”. In India, Ghana, and Nigeria, women are more concerned about how to have sex or sleep during pregnancy, or if they can eat moringa (an edible plant).

Anyways, this is such an AMAZING book, a page-turner for a topic that is so focused on data and statistics. The book also highlights the common fallacies in assuming causality (explained very nicely). Highly recommend it and an e-version is available in most public/university libraries in the US (I borrowed it from my local public library).

So, I have seen Google trends use for a variety of health-related outcomes. Recently, I came across this study conducted by Agriculture Economics, Sociology, and Economics department from Penn State, who used Google trends in the US to look at food sourcing patterns, and how worries changed since the start of the pandemic in the US. At the beginning of pandemic (February 17), people were worried about hoarding, then they were worried about how to procure foods directly or locally, and the following week, people googled about food delivery.

Similarly in Spain,  Dr.Laura Laguna from Instituto de Agroquímica y tecnología de los Alimentos, presents relevant research doing comparative analysis of social media, Google trends, and consumer survey (this was the article that inspired this entire post!). They find trends in Spain go from restaurants to recipes to delivery as the pandemic progresses. Google trends of banana and eggs show really nice signals as well.

I played around a bit on Google trends, I used the word “chicken” to see trends to see what people are googling in India and East Africa. Here are some cool trends I found:

  1. “Chicken” in India has weekly cycles and search spikes on Sunday (see data below that shows the trend in the last 90 days). This makes total sense to me because growing up in Madras, we had chicken only one day of the week and it was always Sunday. Examining by city and sub-region within India shows that this is happening in the south. Cool, right? There’s also some interesting downward trend happening (estimated using eyeball statistics).

A screenshot of a social media post

Description automatically generated

2.‘Chicken’ search trends in the last 12 months in Tanzania, Kenya, and South Africa show a large spike happening in April 2020, more pronounced in Kenya and South Africa.

A screenshot of a social media post

Description automatically generated

A screenshot of a social media post

Description automatically generated

A screenshot of a social media post

Description automatically generated

How do people analyze google trends data? Here’s a simple guide on how to use google trends and systematic review paper summarizing methodologies and evidence gaps in using Google trends. No surprise here, people use time-series analysis to evaluate these cycles and trends, and one nice thing about time-series analysis is that it generally gets at the omitted variable bias – because when you control for a present variable using the past value of the same variable, you do generally account for the whole host of factors that affect the ‘past value of the same variable’.

One thing that this review notes is that not a lot of folks use this for forecasting – there is a real value in that especially to carefully predict any disruptions in food systems in this COVID-19 world.

Enjoyed this article? – then Tweet or share your favorite Google trend!