The Opinion Matching Effect (OME): A subtle but powerful new form of influence that is apparently being used on the internet

Abstract

In recent years, powerful new forms of influence have been discovered that the internet has made possible. The Search Engine Manipulation Effect (SEME) was discovered in 2013, and a comprehensive report about its effectiveness was published in 2015, with multiple replications published since then. SEME research shows that bias in search results can produce large shifts in the opinions and voting preferences of undecided voters without their awareness – upwards of 80% shifts in some demographic groups. We now introduce another new form of influence: the Opinion Matching Effect (OME). Many websites now “help” people form opinions about products, political candidates, or political parties by first administering a short quiz and then informing people how closely their answers match product characteristics or the views of a candidate or party. But what if the matching algorithm is biased? We first present data from real opinion matching websites, showing that responding at random to their online quizzes can produce significantly biased recommendations. We then describe a randomized, controlled, counterbalanced, double-blind experiment that measures the possible impact of this type of matching. A total of 773 politically diverse, eligible US voters participated in the experiment. They were asked to form opinions about the two candidates in the 2019 election for Prime Minister of Australia (thus assuring that our subjects were initially “undecided”). After reading basic information about the candidates, they were asked questions about their opinions and voting preferences. Then they were given a short quiz about various political issues, after which they were told how closely their views matched those of each candidate. Then they were asked those questions again, and we measured changes in opinions and voting preferences. Participants were randomly assigned to one of three groups: people who were told their answers closely matched the views of Candidate A, Candidate B, or both candidates equally. This produced shifts in voting preferences between 51% and 95% in the bias groups, with no participants showing any awareness of having been manipulated. In summary, we show not only that OME is a large effect; we also show that biased online questionnaires exist that might be shifting people’s opinions without their knowledge.

Keywords: Opinion Matching Effect, OME, online manipulation, recommender systems, online quizzes, voting advice applications, election manipulation