2017 news psychometrics oppositesdontattract 883x432 1

Opposites really DON’T attract

15 February 2017

The article at a glance

Study shows that ‘birds of a feather do flock together’ – finally providing scientific backing for the idea that people seek romantic …

Study shows that ‘birds of a feather do flock together’ – finally providing scientific backing for the idea that people seek romantic partners and friends with similar personality traits.Opposites don't attract

Classic Hollywood movies are full of ‘opposites attract’ romantic storylines – ranging from pampered Rose and penniless Jack in Titanic, to prostitute meets millionaire businessman in Pretty Woman, to snobbish professor finds cockney flower girl in My Fair Lady.

This has always seemed to fly in the face of research and conventional wisdom, which holds that friends and romantic partners tend to share certain characteristics such as age, values, education and intelligence – but scientific research has consistently failed to establish such a link for personality.

Until now.

A new study published in the journal Psychological Science, co-authored by University of Cambridge researchers and based on digital behaviour, shows that opposites really don’t attract. Instead, as the title of the journal article spells out, “Birds of a feather do flock together” – people are attracted to those who are similar.

Dr David Stillwell
Dr David Stillwell

“When we looked at actual digital behaviour – which registers what people like and don’t like, and how they express themselves – people really do seek out friends and romantic partners who are similar to them,” says study co-author Dr David Stillwell, University Lecturer in Big Data Analytics & Quantitative Social Science at Cambridge Judge Business School, University of Cambridge.

“While ‘opposites attract’ might make for a sentimental movie plot, the evidence in real life indicates otherwise,” says the first author, Dr Youyou Wu of University of Cambridge and Northwestern University.

The study’s breakthrough is in identifying why decades of previous studies had established little evidence that a similar personality was a key factor in friendship and romantic involvement: that prior research usually asked people to assess their own personality traits by questionnaire – which results in people comparing themselves to others around them, known as the “reference group effect”. For example, an objectively introverted engineer might regard himself as relatively extroverted if surrounded by even more introverted engineer friends.

Instead, the researchers used a Facebook app called MyPersonality to collect Facebook data from 295,320 consenting participants based on their Facebook “likes” and language used in their status updates – which provided a more accurate assessment of their true personality traits compared to a large population, circumventing the reference group effect. For example, people who are extroverted tend to like “dancing” and “partying”, while they’re more likely to use terms such as “great” or “amazing” or “happy” in their status updates.

“Self and peer reports are inappropriate methods for studying personality similarity, because they amplify the differences in actual personality and obscure the similarity among partners and friends, who likely unconsciously treat one another as reference groups,” the study says.

The study is based around five commonly used factors of personality:

  • Openness to experience
  • conscientiousness
  • extroversion
  • agreeableness
  • neuroticism.

The study concludes:

Our findings provide evidence that romantic partners as well as friends are characterised by similar personalities. Relatively strong similarity was detected between romantic partners and between friends when we used Likes-based and language-based measures. By contrast, self-reports yielded only weak to negligible similarity.

The study is co-authored by Dr Youyou Wu of the Psychometrics Centre, University of Cambridge, and Kellogg School of Management, Northwestern University; Dr H. Andrew Schwartz of the Department of Computer Science, Stony Brook University; Dr David Stillwell, Cambridge Judge Business School, University of Cambridge; and Dr Michal Kosinski, Graduate School of Business, Stanford University.