Researchers Can Tell People's Incomes From Their Tweets

Lower-salaried Twitter users swear more; those who make more money are more likely to talk politics.

Christian Hartmann / Reuters
The words in tweets can reveal a lot. Twitter data has previously been used to determine which countries were more likely to suffer from heart disease and used as a tool to measure depression. Now, computer scientists have linked people’s tweets to their income.
A group of researchers, led by Daniel Preotiuc-Pietro of the University of Pennsylvania, first looked at Twitter users’ self-described occupations. They used a British job-coding system to sort the occupations into nine classes and determined an average income for each class. After finding a representative sample for each job class, researchers analyzed 5,191 Twitter users and more than 10 million tweets. They looked for words that each job class used distinctively.
The overall picture showed that lower-income users were more likely to swear and use Twitter to talk to their friends. Higher-income users used Twitter more professionally, particularly as a tool to disseminate news. Those with higher incomes spoke more frequently about politics, and were more likely to express fear and anger on Twitter.
Some of the findings were unsurprising—the words a person uses can be tied to gender, and the study supported the fact that men earned more money than women. Older users, as well as those with more education, tended to have higher incomes.
But, there were some interesting revelations in the study published in the journal Plos One. People whose tweets were relatively optimistic had a lower mean income, while higher-income users sent out tweets that were less anxious and religiously unaffiliated.
Researchers weren’t able to find any significant relationship between political orientation and income, nor where they able to find any strong correlation between life satisfaction and income. Researchers note in the study that the findings were a “necessary first step” and suggest there are several future applications of the research, ranging from health to marketing.