Rediff Logo News Find/Feedback/Site Index
September 14, 1999


E-Mail this report to a friend

Indian voters tend to back winning horse say pollsters

A five-member Constitution bench of the Supreme Court today dismissed the petition filed by the Election Commission seeking a ban on publishing opinion and exit polls till the election ends on October 3.

The EC had sought a ban on the grounds that results from the earlier phases of the poll would unduly influence voters in the later phases of the general election. However, the petitioner, Jain TV, contended that any such restriction would be a curb on the fundamental right of expression, as guaranteed by the Constitution, besides a restriction on press freedom.

Chief Election Commissioner Dr Manohar Singh Gill had declared earlier that such opinion polls will influence poor, illiterate voters in the latter stage of the election, thereby affecting its fairness.

However, opinion pollsters do not agree with Dr Gill.

"On the contrary, it is more likely that the middle-class urban voter, who alone has access to the results of such opinion and exit polls in the first place, is likely to be affected by the results. The illiterate voter in the village is most likely not even aware of such polls," said Dr N Bhaskara Rao, chairman, Centre for Media Studies. The CMS is among India's oldest pollsters.

Yet, the pollsters seem to agree that opinion and exit polls do have an effect, but only a marginal one. Dr Rao said a few voters were influenced to support the winning party. "Our studies have shown that opinion polls do affect the voters. If they did not, why would the EC be so worried" he asked.

Centre for the Study of Developing Societies director Yogendra Yadav, who is also a well-known psephologist, echoed Dr Rao. "My institution carried out a survey on whether our opinion polls effect electoral chances and found the answer in the positive," he said. "Indian voters tended to back the winning horse!"

However, pollster Dorab Sopariwala is not too sure about such studies. "While surveys tend to indicate bandwagon effects, it could be that most people surveyed declare their support to the winning party, but may not actually vote for it," he pointed out.

Even Yadav agreed that not enough studies have been carried out on the influence of opinion and exit polls to reach a definite conclusion. "Our study found that only 16 to 17 per cent had even heard of opinion polls, only three per cent felt that opinion poll results affected their own decision, but finally, less than one per cent said they actually changed their decision," he said.

According to Yadav, while the swing caused by opinion and exit polls is less than one per cent, in India's first-past-the-post system that could translate into as many as 20 seats! "The swing can affect the fence sitters to cast their vote for the party most likely to win," he said.

Unlike the West, where the fence sitter may vote for the underdog or the losing party, in India the voter is more likely to join the bandwagon or vote for the winning party.

"Given that today, political stability is one of the key issues, it is even more likely that the voters will coalesce around the winning party to give it a good majority and thereby stop the annual elections, which seem to have become a part of our system," said Yadav.

Dr Rao believes the Election Commission does have a case, but is not in favour of legal measures. "In today’s world of information and communications, to seek a legal ban is frivolous. Legal redress is like using old tactics for new technologies. How will the EC ever enforce it?" he asked.

He felt the EC should have called a meeting of all political parties and leading editors to hammer out a broad agreement on poll code behaviour. "The media can agree on some sort of code of conduct, like the one that political parties have accepted," the veteran pollster said.

Yadav too favours a "model code of conduct for opinion polls" to ensure that opinion polls just before election day and exit polls while an election was on should not be published.

One major complaint against opinion polls is that they are not truly representative or fair. How, ask politicians, can a sample survey of just a few thousand people do justice to an electorate that today numbers over 605 million?

That India is the world's largest electorate needs no repeating. The second largest electorate, the United States, is only 180 million, less than a third India's size. Moreover, there is the much talked-about heterogeneity of India, an anthropologist's delight and a random survey sampler's nightmare.

Politicians and others are quick to state that given the size of and differences within India, any survey is only a limited exercise that must not be allowed to influence others. Most opinion polls claim a random sample of as little as 2,000 to a maximum of 20,000.

Sopariwala admits that conducting opinion polls in India is truly a feat with few parallels. "There are other problems in terms of regional differences, language, literacy, but all of which can be surmounted," he points out.

"Yet," he asserts forcefully, "one of the fundamentals of statistical surveys is that a sample size is not really an indicator of accuracy. For instance, it is like a large cake, and a small size or a big size will both truthfully give you a taste of the cake."

The other pollsters are in complete agreement. "We have found after being in the business for over 20 years that a good sample of 6,000 people across the nation can give a fairly accurate picture," says Dr Rao, and cites his organisation's record of the same.

"Even in the US, sample surveys cover just 1,000 persons or less. So given that our electorate is about thrice the size, a survey of 2,000 to 3,000 is fairly good to have an assessment," added Sopariwala. "And an exit poll or survey of 20,000 is more than sufficient."

The key is not just numbers but intelligently picking up the correct samples. "No doubt years ago this caused problems but today we have enough data to correctly pick up representative samples throughout," claimed Dr Rao.

"India should be seen in regions because we have found that regions vote along the same lines. Second, certain caste groups vote similarly but since they are spread out, their influence can be limited in India’s first-past-the-post system. For instance, the last time the Bahujan Samaj Party won 20 per cent of the votes in Uttar Pradesh but only about four seats. Thus there are certain thumb rules," said Sopariwala.

Yadav has the final word. "Despite the odd mistake by pollsters (such as the Tamil Nadu outcome in 1998), and a few suspect groups, the truth of the matter is that over the past few elections, opinion polls and exit polls have been fairly accurate. All of which only proves our professionalism," he declared.

Tell us what you think of this report