Researchers have warned that big data trends discerned from social networks, like Twitter and Facebook, misrepresent the real world because they use biased information.
Computer scientists at McGill University, Montreal and Carnegie Mellon University in Pittsburgh, have discovered that social media is often guilty of “population bias,” as large sections of society are not represented by the sites.
Only five per cent of over 65s use Twitter, for example, while Pinterest’s dominant audience is females aged 25 to 34. Even Facebook, which possess a relatively broad demographic cannot provide entirely reliable data, as there is no “dislike” button.
Dr Derek Ruths, assistant professor at McGill’s School of Computer Science, believes this misrepresentation could have a significant effect on the real world, with a huge number of research papers based on this erroneous data.
“A common assumption underlying many large-scale social media-based studies of human behaviour is that a large-enough sample of users will drown out noise introduced by peculiarities of the platform’s population,” he said. “These sampling biases are rarely corrected for, if even acknowledged.”
A website’s design can also influence people’s decisions by leading them to various links, for example, meaning social networks are also often guilty of “interest bias.”
Previous studies have been used to predict fluctuations in the stock market, the performance of Hollywood films or to justify government spending, but the latest research suggests that just because something is labelled as “big data,” doesn’t mean it should be followed blindly.