Just enjoy the opinion polls

Hillary Clinton had reason to rejoice over the opinion polls during the last US presidential election, while Donald Trump was apparently a bit scared — he repeatedly accused the media for being biased against him. Similarly, the Conservatives repeatedly complained before the 2012 election, that the media showing the opinion polls would help Obama. Many experts and politicians believe that opinion polls are instrumental in creating or mobilising public opinion. “There is no such thing as public opinion. There is only published opinion.” This is what Churchill had said, in his own style. However, is an opinion poll a self-fulfilling prophecy? Does it have enough potential to control the future?

First, consider the “bandwagon” effect of the opinion polls. Some electorates certainly have the tendency to support the apparent leader, and moreover the natural tendency of people is to stay with the winner. Donald Trump in 2016 or Mitt Romney in 2012 was afraid of such bandwagon effect of the opinion polls. There were several small studies in the US regarding this at different time points. In a 1994 article in the Journal of Politics, a study on a group of students from the Kentucky University was reported, where the support towards the winner of the opinion poll was double among the students who knew the opinion poll results than those who were ignorant about it. Again, two researchers from Stanford University and Dartmouth College have described a study on about 900 people around the 2012 presidential election, where 9.1 per cent people opined they would have changed their votes had they known the opinion poll results. The bandwagon effect in this case is “7.2 per cent to 11 per cent with 95 per cent probability”. 

In practice, the amount of bandwagon effect is utterly unknown, and very likely to change from situation to situation. The bandwagon effect might be more for young voters. According to a report, before the Delhi assembly election in 2013, an Aam Aadmi Party (AAP) internal study showed that 77 per cent of those who wanted to vote for the Bharatiya Janata Party (BJP) believed that the BJP would win. These percentages were 78 and 91 for the Congress and AAP respectively. Thus, the bandwagon effect is quite clear. A Centre for the Study of Developing Societies (CSDS) study in the backdrop of the 2014 Lok Sabha election shows that about 43 per cent of the voters float in hawa. However, there are many sources of hawa. Opinion polls, according to that CSDS study, provide only 3 per cent of the hawa. However, in many cases, that may be more than sufficient to change the result of an election.

However, there is no reason to think that winner in opinion polls would always get public support. Hillary Clinton, Ed Miliband or Atal Bihari Vajpayee could vouch for that. Actually there might be an opposite effect of opinion polls as well, called the “underdog” or “boomerang” effect, which is sort of compassion towards the trailing candidate. Democrat Harry Truman was trailing behind Republican Thomas Dewey by more than 5 per cent in Gallup’s opinion poll in the 1948 US election. Many Republicans were so confident of a Dewey victory that they didn’t even turn up to vote.

The rest is history of monumental blunder of opinion polls — Truman won by a 6 per cent margin. Experts attributed a part of this to a sympathy-wave towards trailing Truman. In fact, in 2008, both Barack Obama and John McCain tried to portray themselves as underdogs to gain such sympathy. It is again impossible to exactly quantify this underdog effect, although some researchers believe that its effect might be less than that of the bandwagon. However, an analysis of the data from 1950-97 in Britain showed that the underdog effect was significant in that country.

The third important effect of opinion polls is the possibility of “strategic vote”. Let’s take a simple illustrative example. Suppose there are three political parties, say A, B, C, contesting in an election. From the opinion polls, Party C understands it has absolutely no chance of winning. Instead, Party A is going to win, whom Party C dislikes most. So the supporters of Party C might vote for Party B only to defeat Party A. Such a strategic vote played a key role in the 2014 election in Holland, and also in the huge victory of Justin Trudeau in Canada in 2015. Such strategic polls are inevitable in multi-party democracies, and these are bound to be deeply affected by the opinion polls.

In reality, all these effects are mixed — within many other issues of the election, and hence the quantum of their individual effects are absolutely unknown. More detailed studies are certainly needed in the larger interest of democracy.

Opinion polls might have another serious effect on democracy in terms of voter turnout, particularly if they exhibit big differences between the potential winner and the loser. In the US, in 1996, Bill Clinton was far ahead of Bob Dole in the opinion poll. As a result, the polling percentage was only 49, the lowest in 72 years.

If, however, the opinion polls from different organisations indicate different directions, they can pacify the effects. Interestingly, often, most opinion polls speak in the same tune, even when they are proved wrong later.

However, personally I don’t advocate against opinion polls. In fact, without opinion polls, the thrill of elections will be halved, if not even less. The undesirable effects of opinion polls would be reduced if people consider such survey results more of fun, rather than take them seriously.

The author is professor of statistics at the Indian Statistical Institute, Kolkata