Whether it is possible to determine the approximate level of their income from the messages of users on Twitter, a group of researchers from different countries decided to find out.
The study was led by Daniel Preotiuc-Pietro of the Center for Positive Psychology at the School of Science and the Arts at the University of Pennsylvania. The group also included: Svetlana Volkova from Johns Hopkins University, Vasileios Lampos and Nikolaos Aletras from King’s College London and Yoram Bachrach from the Microsoft Research Center.
The researchers took an unconventional approach: instead of asking study participants direct questions, they studied their social media posts, in which they were not shy about sharing many details from their personal lives. The scientists were interested in whether it would be possible in the future to replace the conduct of expensive and often biased surveys with the analysis of information from social networks.
Read more:
- Sherry Turkle: “I’m not anti-tech, I’m pro-talk”
To begin with, scientists were interested in the user’s occupation, which was determined based on their records. In the UK, there is a classification of professions that divides them into nine categories. The researchers determined the average worker income in each of the nine categories and then collected a representative sample of Twitter users for each of them. As a result, more than 10 million messages from 5191 users were analyzed. Daniel Preotuk-Pietro claims this is the largest amount of data ever collected for this kind of research.
The scientists then created a statistical algorithm to collect natural language data, which helped determine which words are most commonly used by people in each of the nine occupations.
Read more:
- Rules of electronic etiquette: what is dangerous to talk and write about
After analyzing the data, the researchers noted the following trends.
People with high incomes:
- Have more followers.
- Post less frequently.
- They rarely give positive or negative assessments, write in a neutral tone.
- Other people’s messages are retweeted more often, their own messages are also “retweeted” more often.
- More often they discuss topics related to politics, business and non-governmental organizations.
- They express anger and fear more often in their messages.
People with low incomes:
- They have fewer subscribers, but they write messages more often, because they use the service primarily for communication.
- More often openly declare their religiosity.
- Emotions such as sadness, surprise, disgust, and anxiety are more commonly shown. At the same time, optimistic messages are also mainly posted by users with lower incomes.
- More often place links to news, pictures and videos.
- Use more swear words.
Pass the tests
- What are you afraid of in communication?
“People with relatively low socioeconomic status mostly use Twitter to communicate with their peers, which is why they write much more often. Those with higher incomes are more likely to use the service to spread news and generally tend to use it for professional rather than personal purposes,” says Nikolaos Aletras.
Подробнее см. D. Preoţiuc-Pietro et al. «Studying user income through language, behaviour and affect in social media», PLOS One, September 2015.