Few people advertise their income online, but it turns out tweets can people help decipher the sender's salary.
Researchers from the University of Pennsylvania analyzed the tweets of over 5,000 U.K. users to see if their social media posts could link them with their respective income brackets.
The results, published in the September edition of academic journal PLOS ONE, showed that higher earners, for example, tend to swear less but express more fear and anger on Twitter.
The study took users' self-described job titles which were then cross-referenced through a U.K. job code system to determine an average income for each group. The highest income bracket was made up of conservation and environment professionals, followed close behind by production managers and directors. On the other end of the spectrum were hairdressers, sales supervisors and low level factory jobs.
From there, researchers used an language analyzing algorithm to identify words, emotions and content used differently, or exclusively by each income bracket.
A portion of the results verified well-known income links — analysis was able to determine age and gender, both of which were linked to earnings.
However, the team found some surprises.
Those who earned less tended to post more optimistic tweets but used more swear words. Their more affluent counterparts posted more neutral content, exhibited more fear and anger, but less surprise, sadness and disgust.
Results also showed how a different economic classes used the social media site in their everyday lives.
"Lower-income users or those of a lower socioeconomic status use Twitter more as a communication means among themselves," Daniel Preotiuc-Pietro, the post-doctoral researcher who led the study, said in a press release. This group used more personal language and alternative spelling in their tweets.
"High-income people use it more to disseminate news, and they use it more professionally than personally," Preotiuc-Pietro explained.
Higher earners also tended to discuss politics, nonprofits and corporations than their lower-earning peers, with words like "lobbyists," "advocacy," "initiative" and "benefits," more common in this group.
It's likely this information could be applied in both marketing and politics, the paper explains.
"It's the largest dataset of its kind for this type of research," Preotiuc-Pietro said. "The dataset enabled us to do something no one has really done before."