National - Elections 2004

The case against banning exit and opinion polls

There is a clear case for trying to improve the quality of opinion polls and forecasting, which have not seen much innovation since the 1980s. But, a legal ban, unfortunately, does not achieve that, writes Yogendra Yadav.

The current debate on banning opinion polls and exit polls could do with a dose of informed opinion. It is crucial, for we are headed for yet another major policy decision on reforming our polity that has not had the benefit of a dispassionate reflection or careful scrutiny of available evidence.

The debate has focussed on either the very narrow and technical questions of law or the very fundamental and ethical questions of democracy. The question of law, which came up in the Supreme Court in 1999, is about whether the Election Commission has the power to impose such a ban, and under which law. The Commission had cut a sorry figure last time when confronted with this question. The Supreme Court had rudely reminded the Commission that a consensus at an all-party meeting did not provide the required legal sanction. The Commission is unlikely to go ahead with a ban this time without any legal protection. Hence the Government's reported move to enact an ordinance empowering the Commission. The fundamental question of the Constitution is whether such a law, if enacted, would not violate the fundamental right to freedom of expression. There is indeed a strong argument here and this may well be a clincher in the legal battle that is bound to follow any attempt to ban opinion polls.

Valid and powerful as these arguments are, these do not convince an ordinary citizen. The debate on this question must address the simple and non-legal question raised by the political class: are opinion polls good for the functioning of democracy? Do opinion polls influence how the people vote in a way that is illegitimate or unfair? Won't a ban on opinion polls produce a better and more equitable climate for political competition? These questions need to be addressed openly and with the help of all the evidence at our command. Let us ignore the ill-informed accusations and hostility expressed by the political class and examine these points at their logical best.

The position that opinion and exit polls are bad for our democracy rests on the following chain of arguments, of which each step appears plausible. One, media-driven opinion polls and forecasts based on these are widely noted by the public at large. Two, following from the first, these polls therefore influence the process by which people make up their mind about who to vote for. Three, this influence is either illegitimate, for most of the forecasts are not correct, or undesirable anyway as it adversely affects the level-playing ground in politics. Not everyone makes all the three arguments. And not every critic of the impact of opinion polls questions the professional integrity of the pollsters. Yet all these are arguments that are widely deployed and need to be looked at one by one.

Fortunately, some of these admit of a straightforward factual answer. And we have fairly reliable evidence to do so. The National Election Study (NES), a series of large and social scientifically reliable surveys of the Indian elections conducted by the Centre for the Study of Developing Societies (SDS) has addressed this question in two of its recent waves. The nationwide surveys conducted as part of the NES 1996 and the NES 1999 asked a representative and fairly large sample of respondents all over the country if they had read or heard about opinion polls and survey related forecasts made by newspapers and magazines in the elections concerned. To those who had hear about it, the survey asked further questions about whether this knowledge had influenced their voting choice and how. Another large survey carried out by the CSDS during the recent Assembly elections again asked these questions to see if the pattern was different in a metropolitan setting with high media exposure. The evidence gathered by these surveys (paradoxically, you need a survey to gather reliable information about the impact of surveys) is quite robust and reduces considerably the need for speculation on this count.

The first question about the reach of the surveys can be answered fairly precisely. In 1996, about one fourth or 26 per cent of the country's electorate had known about the opinion and exit poll based forecasts. By the 1999 Lok Sabha elections, the numbers went up to 29 per cent. We do not know the exact number for this time, but it would not be unreasonable to infer that with the expansion of print and electronic media the number would be close to one-third now. These are humbling figures for the pollsters and the media. About two-thirds of our citizens live beyond the pale of the media and the world of news created by it. Particularly so, if we remember that only one-third of those who had some knowledge of the opinion polls had actually read these. The majority had only `heard about' it and were therefore less susceptible to its influence. Unsurprisingly, the proportion of those aware was higher in urban areas, upwards of 40 per cent, and much lower in rural areas, still around a quarter in 1999. Looked at from another vantage point, it could be argued that one-third or one-quarter of the electorate are not small numbers. We could be talking of 15 crore to 20 crore electors here.

Sure, but it is important to take a look at the next set of evidence, before making up our mind on this question. The crucial question is: what proportion of those who know about polls get influenced by it in their voting decision? The evidence from both the national polls suggest that the numbers drop drastically here. Of course, a lot of those who know about the polls say they are influenced by these, but a close scrutiny of the evidence shows that most of this influence is by way of reinforcing the decisions taken already. A little more than one-tenth of those who knew about the polls actually changed their mind about who to vote for on the basis of this information. That works out to 3.7 per cent of the total electorate in the 1996 elections and 3.3 per cent in the 1999 elections. In other words, only a minuscule proportion of the universe of citizens change their voting decision on the basis of opinion polls. Once again, the proportion may be small, but the sheer numbers may still be big, big enough to make a difference in a close contest.

We must therefore look at the final set of evidence, about what exactly happens when people get influenced by opinion polls. The popular belief is that knowledge about opinion polls can produce only one consequence: people tend to go with the hawa (wind). Research all over the world suggests that while polls do create a `bandwagon effect,' they could also create an `underdog effect,' or produce a sympathy factor for the reported loser. The CSDS study in the recent Delhi election tested this question in an ideal condition. The setting was metropolitan and the media exposure quite high: 41 per cent of the respondents knew about surveys and forecasts made by the media. One-fourth of them, about 10 per cent of the total electorate, said they were influenced by the forecasts. And this was a wave election: all the surveys were forecasting a sweep for the Congress. It was thus an ideal case to test if the Congress benefited from a `bandwagon effect' created by media forecasts. The evidence negates this hypothesis. The 10 per cent influenced by the polls were almost equally split: half of them switched from the BJP to the perceived winner, the Congress; but the remaining half switched in favour of the perceived loser, the BJP. The net effect was negligible. Now, this is not decisive evidence and more research is necessary before we reach firm conclusions. But prima facie, there seems to be little evidence to support the largely held view that the party that is shown to be the winner in the polls derived huge advantage from this.

To be fair, this does not exhaust or meet all reasonable suspicions and objections to opinion polls. It could be said that what look like small percentages nationally could be quite a large number of persons in specific areas and constituencies. It could also be argued that while the effect of the polls is quite limited on the public at large, its effect on the morale of the party workers is quite strong and must be considered. Someone else could argue that while all this applies to opinion polls done before the elections, exit polls are in a class by themselves, specially in our country where they are carried out when some parts of the country are still to go to the polls. These are not irrelevant or unreasonable objections. And these cannot be met only by the evidence supplied so far.

These questions invite us to look at larger issues that go beyond numbers. First, if one grants that opinion polls and exit polls can have an effect on some crucial areas and sections during the elections, it still remains to be seen if the effect is illegitimate or unfair. After all, every election result is always influenced by the voters' sense of who is going to win. And they rely on all kind of evidence to sense this: news reports, gossip, personal information and guesswork. What is wrong if they get more systematic information, something that is available to, and is used by, all the political parties? To my mind one can object to this only if there is some reason to believe that the forecasts made by the media are systematically biased and manipulated.

That brings me to the final point about the accuracy and professionalism of opinion polls in the country. This is not the place to get into a detailed assessment of the accuracy or otherwise of opinion polls in India, but let me just note that contrary to popular impressions the overall record of Indian pollsters is not bad by international standards. Of late we have seen some really bad forecasts, but these have been cases of poor professionalism and not of political manipulation. By and large, most of the agencies and media houses involved in this exercise have

© Copyright 2000 - 2009 The Hindu