A mental contagion that afflicts our politics more than ever.
Originally published March 12, 2008
Decades ago, at the dawn of the television age, Marshall McLuhan described what was happening as Western society overhauled its 500-year-old nervous system, replacing print with electronic media. “The new electronic interdependence,” he wrote in The Gutenberg Galaxy, “recreates the world in the image of a global village.” McLuhan did not see this as progress; indeed he foresaw Western culture entering “a phase of panic terrors, exactly befitting a small world of tribal drums, total interdependence, and superimposed co-existence.”
But the fact that our brains are practically hardwired to electronic media these days makes us susceptible to all kinds of social contagion, not just “panic terrors.” One of these contagion phenomena, known to social scientists as “the bandwagon effect,” is particularly relevant in primary-election season.
The bandwagon effect refers to the tendency of voters to be supportive of a candidate purely because he or she is perceived as popular. Thus a small or local success for a candidate – a good outcome in an early primary, or a jump in a local poll – can lead infectiously to a broader popularity, however undeserved by other criteria. After his wins in Iowa and New Hampshire in early 2004, for example, John Kerry moved up sharply in the polls and became the front-runner.
The effect is far from being the only influence on voter decision-making. But its existence has been demonstrated in several well-known social science experiments, and the importance that candidates, journalists and voters now deliberately place on early primary contests.
Perhaps because of the unflattering light it sheds upon the electoral process, the bandwagon effect is seldom called by its right name. George H.W. Bush celebrated it as momentum—“the big Mo”—after unexpectedly winning the Iowa Republican caucuses in the 1980 presidential campaign. Others have called it “the Iowa bounce.” Hillary Clinton’s surging poll numbers a year ago had commentators talking about her newly-perceived “electability.”
If the bandwagon effect is merely about electability, then in a sense it is less worrisome. To the extent that a voter is decidedly partisan, and merely wants to choose the most electable candidate to lead his party to victory in the nationwide contest, his sensitivity to other voter choices during the primaries would be perfectly rational.
But the experimental data tend to contradict the notion of the bandwagon effect as a rational partisan strategy. A study in 1994, by researchers from the University of Southwestern Louisiana and the University of Kentucky, found the effect to be particularly strong among “independent” voters not affiliated with any party. A study in 1996 blamed “particularly unsophisticated independent voters” for the effect in a German election. A UCLA study published in the Journal of Applied Social Psychology in 1998 hinted that the effect was rooted in traits similar to suggestibility:
Bandwagon effects were stronger for women compared with men, and for 2 of 3 PAD (pleasantness, arousability, dominance) basic temperament factors; that is, for individuals with more arousable and less dominant temperaments.
All this ought to be unsettling. But should it really surprise us, given that political candidates are sold using some of the same irrational, mood-altering techniques – wavy flags, smiling children, stirring music, flannel shirts – that advertisers use to sell consumer products?
A few weeks ago, reading about Obama’s surge in the polls, I sensed my own inner readjustment: Before, when he’d had only half Hillary Clinton’s poll numbers, Obama had seemed little more than a smooth-talking political queue-jumper. After his post-Oprah surge he seemed … less objectionable somehow. Even Mike Huckabee, as he exploded in the polls, earned a twinge of acceptance. Maybe he’s not as ridiculous as I’d always thought. I should mention that I am not at all partisan, and have not registered as a Republican or Democrat. Neither Obama nor Huckabee fits into any political category that I find appealing. In other words, “electability” is not their only problem, in my eyes. Nevertheless I felt better about them, knowing that others did. This shift in thinking, by the way, felt perfectly normal and natural. But it felt better to recognize it as an instinct and reject it.
Assuming that the bandwagon effect is a basic instinct, we might ask why it exists. Did it have some adaptive function during our early cultural development? Was the tendency to “pick a winner” a necessary survival trait when politics was more brutal than it is today? Or was it merely part of a general mental interdependence that held early human society together? And could it date back further still, to the very basic mammalian instinct to stay safely within the herd?
Whatever the answer turns out to be, there is no shortage of evidence that in modern societies human opinions remain deeply interdependent. Social psychologists have found this even in the romantic market, where “romantic popularity” turns out to be a key factor in people’s mating preferences: In other words, to a person in the mood for love, a prospective mate will seem more attractive than otherwise if it is perceived that the crowd finds her attractive too. (Small wonder that our culture places such high value on celebrity.)
Conceivably, such a trait could have evolved as a way of leveraging one’s individual decision-making with the summed “wisdom of the crowd.” But the wisdom-of-the-crowd effect, described by James Surowiecki in his book of the same name, requires that opinions be relatively independent of each other. The more tightly interwired and interdependent a crowd becomes, the more its opinions amount not to wisdom but to herd-like madness.
One can view this process as a contagion, but arguably it is better seen as a runaway autofeedback phenomenon, in which our collective judgement comes to depend more and more upon itself, thereby becoming less stable, more subject to sudden movements, and less connected to reality — in other words, more “volatile,” as pollsters and pundits now routinely observe of public opinion.
Periodically, it is recognized that Iowa’s caucuses and New Hampshire’s primaries have a disproportionate effect on the presidential race. Typically it is suggested that all the state primaries be held on or near the same day. But Iowa and New Hampshire are only parts of the problem. Literally every revelation of public opinion prior to the national election feeds back into that opinion and distorts it. And because that feedback effect can cause strange, sudden movements, upward or downward, in a candidate’s popularity, the management of political campaigns is now largely about the management of those irrational swings – damping the slumps, boosting the surges, and getting the candidate’s poll numbers to peak at election time.
The logical way to prevent these herdlike movements is not only to condense elections to the fewest possible dates, but to hide poll results from the public, at least until all votes are cast. In the world of commerce, information is routinely kept secret to prevent unwanted feedback effects. But in a society with a bedrock faith in free speech, the censorship of poll results seems unlikely at best.
Even an acknowledgement of this issue is culturally problematic. We humans have a deep need to see not chaos but order, to find patterns and to navigate by them. We are inclined to rationalize the swings of public opinion as if they were driven entirely by real-world events. Thus the opinion bubble that lifted Huckabee from single-digits to the top GOP spot in the space of a few slow-news weeks has been explained as the product of “Huck’s” aw-shucks style, his Christian principles, the weakness of Giuliani and so forth. In a like manner we explain the to-ings and fro-ings of the stock market, sports teams, fads and fashions, the fortunes of celebrities and more or less everything else in life: We are instinctively, irrationally bent on rationalizing what is never fully rational.
***