• by Erik Sass , Staff Writer @eriksass1, July 22, 2014

The proliferation of brand and product recommendations online is causing information overload, prompting people to turn to their friends for word of mouth recommendations via social media, according to a new study by researchers at Oxford University.

The study, titled “A simple generative model of collective social behavior” and published in the Journal of the National Academy of Sciences, used mathematical models to analyze download patterns for apps on Facebook, covering a total of around 100 million downloads. Before they downloaded the apps, users were presented with a list showing the most popular app downloads overall, and were also notified about their friends’ recent app downloads.

The study found that Facebook users were significantly more likely to download the apps their friends had downloaded than apps from the list of most popular downloads. Comparing the two sources of information, the bestseller list had a “mild” influence but friends’ recent downloads was much more dominant, the authors concluded.

Lead researcher Dr. Felix Reed-Tsochas stated: “We found that copying in the social network environment is driving behavior more than bestseller lists are. Obviously global popularity is important to some extent, but what people in your social network have been doing seems to play a more significant role in online behavior. This might be because users need to make quick decisions in information-rich environments, but other research has identified similar imitative behavior in the off-line world.”

According to another recent study, the tendency to “copycat” behavior in the online world extends to entire cognitive processes. In a study published in the Journal of the Royal Society in February, titled “Analytical Reasoning Task Reveals Limits of Social Learning in Networks,” researchers at the University of Oregon grouped 100 subjects into five “social networks” with 20 members each. The social networks were structured to have different levels of “connectivity,” controlling the extent and frequency of interaction between subjects. The subjects were then given a series of challenging “cognitive reflection tests,” requiring the respondents to use their analytical skills to avoid arriving at the seemingly obvious (but incorrect) conclusion. Subjects first had to solve the problem themselves; then, as the experiment went on, they were allowed to see answers from other members of their social network before they gave their own.

The researchers found that subjects in networks with high connectivity were able to give the correct answer more often when they were allowed to see their neighbors’ answers first. The more connections an individual had, the more likely they were to give correct answers. But when the subjects were then forced to solve three more problems without the benefit of their social networks, the error rate rose to the previous levels, when they were solving problems by themselves. In short, while the social networks had helped them get the right answer, they hadn’t improved their ability to think through the problems independently.