When we read tweets, what assumptions do we make about the strangers behind the 140 characters?
A group of researchers based at the University of Pennsylvania has carried out a study – “Real Men Don’t Say ‘Cute’” – to identify when people’s preconceived stereotypes about Twitter users are taken too far.
“People might not be aware of [these biases], or they might not present themselves as having them,” co-author Daniel Preotiuc-Pietro, a postdoctoral researcher in the university’s School of Arts and Sciences, told Times Higher Education.
The study, published in Social Psychological and Personality Science, tested the hypothesis that language is full of social cues that run deeper than obvious differences in dialect; we signal our membership of social groups through subtle use of prepositions, articles and pronouns. Women, for instance, tend to use more positive and personal language.
To identify biases, the researchers showed participants a collection of tweets from a single user, removing any contextual clues (such as a profile picture). They then asked the participants to guess an attribute of the Twitter user.
The researchers looked at assumptions made about age, gender, political orientation and level of education. They found that most of the time, the participants were correct in their inferences – although women were often mistaken for men when they discussed technology or the news.
The category in which assumptions failed, however, was level of education. Participants performed particularly badly at identifying tweets written by people with advanced degrees.
“People don’t use stereotypes in the wrong way,” said Dr Preotiuc-Pietro. “They just over-rely on them. For example, they may see a swear word and they say, ‘OK, this person is lower educated.’ Almost everything is exaggerated rather than incorrect.”
The study revealed that people with the lowest levels of education were assumed to frequently swear, talk about themselves, and use conversational abbreviations such as “LOL”. People with advanced degrees were assumed to never curse, and instead talk about technology and conferences.
While there is an element of truth to these assumptions, the participants vastly underestimated levels of education. They guessed that only one in 20 of the selected sample had an advanced degree, while the true figure was one in three. Their findings suggest that people with PhDs and other high-level qualifications are far more profane, informal and self-obsessed than we commonly assume.
Dr Preotiuc-Pietro hopes that their research will challenge our reliance on stereotypes. “The first step in making people correct their stereotypes is to be aware of them,” he said.