Eat. Tweet. The two-part cycle has become ubiquitous, flooding Twitter feeds with #foodporn photos of latte art and tacos from the fusion restaurant of the moment.
And while many bemoan the trend as another product of millennial tech addiction, sharing what is being eaten is nothing new — people have used food as a way to connect with others and express themselves long before the Internet revolution. What has changed is how food is shared — and consumed — with digital communities.
Like it or not, the notorious “food tweet” is here to stay, and is impacting nutrition across the Twittersphere — and beyond. Scientists from the University of Utah surveyed 4 million food tweets from mid-2014 to mid-2015 to create HashtagHealth, a study that explores the foods people tweet about and the impacts these messages can have on physical communities.
Of the 4 million tweets — all of which were publicly available and geotagged — the three most mentioned foods were coffee, beer and pizza.
To analyze tweets, scientists came up with a long list of terms that could be related to food and drinks and performed a search.
“We wanted our food capture to be as comprehensive as possible and to really understand what people were saying about food, so our dictionary of food was actually close to 1,500 terms [originally],” assistant professor at the University of Utah College of Health and lead Hashtag Health author Quynh Nguyen, Ph.D., told Food Dive.
Ultimately, Nguyen’s team found that 25 search terms were included in half of the tweets analyzed. Other top tweeted terms include Starbucks, IPA, wine, chicken, bbq, ice cream and tacos.
“About 5% of the time people are tweeting they are tweeting about food, and in those tweets people will basically say the same thing over and over again,” Ngyuen said.
Beyond being well-liked, Ngyuen said the repetition of certain terms could be because users want to cash in on their social capital. In other words, users may be tweeting about foods in the hopes that digital consumers will give them a like, retweet or follow, even if the tweets don't reflect their everyday eating habits.
“Some people might tweet about celery sticks to show that they're healthy, but people also might lean towards tweeting about more celebratory items like cupcakes, so [tweets] don’t perfectly align with behavior,” Nguyen said.
Happy #healthy tweets
Despite the gap between curated social media posts and reality, Nyguen argues that food tweets can make a substantial impact on digital communities.
The impact is twofold. First, despite coffee, beer and pizza’s status as the top three most-tweeted-about foods, there were more tweets about fruits, vegetables, nuts and lean meats than about unhealthy food. About 16% of tweets were about healthy food, while only about 9% were about fast food.
This digital trend reflects of consumers turning away from traditional comfort foods in exchange for better-for-you alternatives.
The HashtagHealth team found that healthy food tweets are strongly associated with positive emotions that can be shared throughout social media.
For example, someone sees a tweet about healthy eating with a photo of a homemade hummus and veggie spread, and feels compelled to try the recipe herself. After some time in the kitchen, she takes a photo of her own creation, tweeting it with her own message that expresses her healthy habits and spreads positive sentiments to her Twitter feed.
But how can happiness be measured on Twitter?
“Out of all the indicators we presented in this paper, that was probably the hardest to measure. ...The final way that we present in the paper was to use a[n] ... algorithm that takes a tweet and dissects it to words, then uses the words in the tweet to predict happiness levels,” Nguyen said.
Tweets using terms such as laughter, joy, and love were classified as "happy." Language was the most powerful indicator of tweet sentiment, but researchers also considered emojis and words typed in all caps.
“About 20% of [general] tweets are 'happy,' and sentiment around fast food is actually lower than that; only 14% of fast food tweets are 'happy,'” Nguyen said. “But about 28% of healthy food tweets are 'happy', and that was interesting.”
Ngyuen believes that the strong positive sentiment in tweets about healthy food can change nutrition for the better, encouraging people to think about what they eat and introducing them to new ways to maintain healthy habits.
“We may have more potential to influence each other now than when conversations were just face-to-face,” Ngyuen said. “Even if tweets don't align 100% with what they're eating, tweets seem, at least preliminarily, to relate to area outcomes.”
Healthy neighborhoods are happy neighborhoods
Nguyen isn’t simply referring to emotional outcomes, either.
“We think that these indicators of the social environment derived from social media may have potential impact on health,” Nguyen wrote to Food Dive. “So far, we are finding that they do predict area-level health outcomes — at various levels: zip code, census tract, county and state.”
Researchers found that geotagged tweets that mentioned fast food and junk food were more likely to come from neighborhoods that were considered food deserts. Tweets about healthy foods were geotagged to healthier, more affluent communities with many food resources.
Researchers also found that healthier neighborhoods are happier neighborhoods. More positive tweets about healthy foods all came from similar locations.
“There are people who look at the physical structures of neighborhoods, like grocery stores or transit systems to determine healthy behavior, but social media also has the potential... to inform health in certain areas,” Nguyen said.
Right now, Nguyen says it’s unclear what the final conclusions of the research will be. She said tweets are biased as a data source.
“Current research is saying that only about 23% of adults are on Twitter,” Nguyen said. “And people ages 18-49 are more likely to be on Twitter than other age groups.”
This limited age range skews outcomes. The automated algorithm can also impact the results, categorizing tweets with only 85% accuracy. For example, it mislabeled tweets about NBA athlete Stephen Curry as food tweets.
Regardless, Nguyen and her fellow researchers are optimistic about this method for studying social influences on nutrition habits and overall community health.