Bandwagon Effect

What it is

A bandwagon was a wagon that carried a band, and in American election campaigns to jump on it was to join the procession. The research phrase keeps that picture of people falling in behind something that is already moving.

The research sense starts with Leibenstein's 1950 paper in the Quarterly Journal of Economics, Bandwagon, Snob, and Veblen Effects in the Theory of Consumers' Demand. He defines the bandwagon effect as the extent to which demand for a commodity rises because others are also consuming it. He sets it beside the snob effect, in which demand falls as others consume, reflecting a wish to be exclusive, and the Veblen effect, in which demand rises with price. The paper is demand theory. It describes how market demand would behave if individual demand depended on what others did, and it contains no experiment and no account of why a buyer would respond that way.

Later work kept the name and moved the question. In opinion research the claim becomes: does telling a person which way public opinion is running move that person toward the leader? In cultural markets and online ratings it becomes: does showing earlier choices change what later people choose, and does that change the final ranking? These are neighbours of Leibenstein's claim, not the same claim, and a sentence that says the bandwagon effect explains a result should be read against the setting in which the result was found.

In effect

Richard Nadeau and two co-authors (1993, International Political Science Review) used a panel design and an experiment in which respondents were told the state and direction of public opinion on abortion and on the constitutional future of Quebec. They built the test around three methodological criteria, including allowance for underdogging and for opinion shifts with other causes, and report a bandwagon effect of roughly 5 to 7 per cent on both issues. The authors say the reasons for the effect remain unclear.

An earlier panel study from the same team (Cloutier, Nadeau and Guay, 1989, International Journal of Public Opinion Research) followed student groups' opinions on North American free trade. Exposure to a majority trend produced both bandwagon and underdog effects, which the authors attribute partly to the stimulus and partly to people's efforts to keep their opinions consistent. See Underdog Effect.

David Rothschild and Neil Malhotra (2014, Research & Politics) randomly assigned a diverse national sample to receive information about different levels of support for three public policies. They find that opinion as expressed through polls affects individual-level attitudes, although the size of the effect depends on issue characteristics. They also describe the existing empirical literature on the bandwagon effects of polls as surprisingly limited.

In cultural markets, Matthew Salganik, Peter Dodds and Duncan Watts (2006, Science) had 14,341 participants listen to and download previously unknown songs in an artificial music market, some able to see how often earlier participants had downloaded each song. Stronger social influence increased both the inequality and the unpredictability of success. Quality only partly determined outcomes: the best songs rarely did poorly and the worst rarely did well, but other outcomes were possible.

On a social news aggregator, Lev Muchnik, Sinan Aral and Sean Taylor (2013, Science) manipulated the first rating a comment received. A positive first rating increased positive ratings by 32 per cent and raised final ratings by 25 per cent on average. A negative first rating was corrected by later readers, so the bandwagon ran in one direction only. The authors report that the positive herding depended on the topic and on whether users saw opinions from friends or from enemies.

What it does not say

It does not say the people on the bandwagon are making a mistake. The economist Michelle Baddeley argues in Copycats and Contrarians (2018) that it is not necessarily irrational to follow others, in her example a restaurant queue. Her objection to herds is that even when the herd is on the right track, valuable private information is lost when people disregard it, a cost carried by the group.

It does not say that every poll moves every voter. The poll experiments report effects of limited size that vary with the issue, and two of them count movement away from the leader as well as toward it.

It is not identical to social proof. Robert Cialdini's term describes people looking at what others do and copying it when unsure. The bandwagon effect names the growth in support that follows, and the studies that use it measure the shift in support, not the reasons behind it.

Finally, the strongest experiments are not about elections. The music market was artificial and the rating experiment was run on one website, so neither shows how a real chart or a real campaign behaved.

---