Paging Dr. Twitter
The sentiments we express online offer a surprisingly reliable gauge of well-being.
By July 8, 2015 - last reviewed on June 10, 2016published
Crunching the words of many can paint a colorful—and accurate—picture of public health. A team of researchers collected tweets from Twitter users in 1,347 U.S. counties and analyzed the language for its relevance to one important indicator: local rates of death from heart disease. Their paper, published in Psychological Science, reports that the kinds of words people use on Twitter can in fact predict such outcomes at the county level.
Psychological factors like chronic stress and anger are known to increase heart disease risk. The researchers found that Twitter language related to anger (including words like hate), negative relationship experiences (jealous), negative emotions (crying), and disengagement (bored) corresponded with greater risk of death from atherosclerotic heart disease (AHD). They also discovered a reverse association between AHD mortality rates and social-media language reflecting positive emotions (wonderful) and engagement (interesting). The Twitter-based model proved to be a marginally better predictor than one derived from traditional risk factors like income, obesity, and smoking.
The most avid Twitter users are not the demographic currently dying of heart disease, yet an abundance of negative words from an area seems to indicate something about the lives of its inhabitants, says Margaret Kern, a psychologist at the University of Melbourne and co-author of the paper.
“It’s getting at the community’s psychological profile.”
A Text For Help
Since the summer of 2013, counselors at Crisis Text Line have received and sent more than 6 million text messages, answering notes from young people who struggle with suicidal thoughts, bullying, depression, and other tough issues and offering referrals to additional resources. The service is free, 24/7, and confidential (the number to text is 741-741). On its Crisis Trends page, the organization also breaks down anonymized data from texts to convey some of the thoughts and concerns that senders have in common.