Opinion polls, not the final word

Whether Bharatiya Janata Party-led National Democratic Alliance will win or lose this general election, will be known only 16 May but according to the media coverage and the results of various opinion polls Narendra Modi’s ascent to power in Delhi is almost certain. This has put the credibility of the opinion poll agencies on stake. Opinion polls, at times, are equated to the toss of coins. Political parties who are on the wrong side of the figures generally allege that agencies are influencing the electorate in India. Almost all political parties except the BJP have been demanding ban on exit and opinion poll.

Election commission, on the other hand, has partially restricted the media. Results of opinion and exit polls cannot be aired after the poll process has started. This is not a blanket ban on agencies to conduct the poll. This is also a matter of studies whether opinion polls influence the voters or not because if it is true then why Atal Bihar Vajpayee-led NDA government lost election in 2004 even after all agencies forecasted clear sweep for the NDA. Projection of seats, behaviour of share market, eventuality of sports and rating of TV channels’ (TRPs) are unpredictable but projection and aspiration still take place. Opinion polls work much better in bi-polar elections rather than multi-cornered contest. In case of our polity almost half of states face multi-cornered contest. It is easier to forecast election outcome in a state with two main political parties, if the survey gets their vote shares correctly. However in states where there are three corner contests, it may particularly go haywire where more than four parties are in fray. India’s population is not homogenous based on regional, caste, community, linguistic, religious identity; issues of secularism and communalism too make comparison with western countries difficult.

Generally opinion polls are conducted in India before the name of the candidate is announced and manifesto of parties are released. Thus they fail in capturing the mood of nation. It is also said that opinion polls are fallible in India but exit polls and post-poll surveys could be relatively more accurate because almost all voters make up their mind about whom to vote only towards the end. It is also being said that the voters may change their mood at the last moment thus proving the opinion polls wrong. History of opinion polls during the last six general elections has been a mixed bag. Seat predictions, based on election surveys, done by various media houses in 1991 and 1999 were fairly accurate. It gave a big boost to opinion polling in India but the question is, why do all agencies run in one direction? The reason behind it is that psephology is not a complete science like rocket or nuclear science but it is a combination of both science and art. The seat predictions were made on opinion/exit poll conducted during the elections. Left is not included in others in 2009. Pre-poll and exit poll show that 2004 and 2009 went wrong but four surveys from 1991 to 1999 more or less were correct. The poll predictions made by different media houses and poll pundits during the 2009 general elections once again failed to predict the victory of the Congress-led UPA as well as 2004 election. Poll pundits predicted a near neck and neck fight between the Congress-led UPA and the BJP-led NDA.

According to the polls, the UPA was supposed to get 185-205 seats while the NDA’s numbers was between 165 and 199. Final tally raised the credibility and accuracy of different agencies as the UPA finally scored 206 seats while the NDA was confined to just to 159 seats.

Not only 2009 election but pollsters made the biggest mistake in opinion poll history of India in 2004. Agencies had not only gone wrong in seat projection but even lost direction. Initial opinion polls suggested that the NDA was going to going to sweep the elections. Results shocked the political parties as well as pollsters and political analysts as the NDA was reduced to 189 seats while the UPA won 222 seats.

Questions were raised about the failure to predict the NDA’s defeat during the 2004 Lok Sabha election despite the ‘India Shining’ slogan. Over confident former prime minister Atal Bihari Vajpayee, who was expected to come back the third time, admitted that the defeat was due to Gujarat riots. Like Vajpayee the pollsters and journalists failed to understand the real reasons of defeat. The NDA had lost not because of the  Gujarat riots but because of the shrinking numbers of alliance partners. The UPA expanded its alliance partners with 18 parties and the NDA reduced its partners from 24 parties to 13 parties in 2004 election.

The Congress was only seven seats ahead of the BJP’s 138 seats. It happened due to the grand alliance of Congress-RJD-LJP in Bihar, Congress-JMM alliance in Jharkhand, anti incumbency factor against 10 years of TDP’s rule and Congress-TRS alliance in Andhra Pradesh.The BJP finally opted AIADMK instead DMK-MDMK-PMK’s old alliance in Tamil Nadu. DMK-Congress alliance bagged all 39 seats while Jayalalithaa-led AIADMK failed to open account in the state. Opinion and exit poll agencies failed to understand the art of alliance rather than playing with numbers. Seat predictions based on election surveys done by various media houses during general elections held in 1991, 1996, 1998 and 1999 were fairly and reasonably around actual results. DRS, one of the survey agency predicted almost actual number of BJP’s 182 seats in 1998 and 1999. If poll agencies predicted fairly in four general elections than why did they go wrong in 2004 and 2009 Lok Sabha elections.

Accuracy of sample surveys depends on many factors. The sample should be large enough to yield the desired level of precision and everyone in the population should have an equal chance of being selected in the sample but only few agencies follow it. Agencies should be more professional and must keep a close eye on the survey because surveyors prefer to fill the questionnaires in the shadow of tree rather than visiting sampled houses. Two, an election survey needs the training of field investigators so that personal inclination should not influence survey. Thirdly, except for a few academic institutions, most of the market research organisations do not have financial resources to conduct regular surveys. Generally agencies conduct opinion polls to capture the mood of nation before candidates’ announcement and release of manifesto but fail to conduct surveys when campaign peaks. It is also very difficult to be accurate because of the fluctuating mood of Indian voters.

Voters may change their mood even on polling day due to various factors. Therefore pollsters must be extra cautious on following the voting pattern on the polling day. Fourth, pollsters play with numbers but forget arts while journalists use their sixth sense but forget about the numbers. Fifth, macro level studies are required over migration from one state to another; who is migrating and who would get affected by this migration process.

A recent sting operation revealed that the seat prediction figures are manipulated even in favour of the clients. Projection is not an easy task but it is also not tough. If poll agencies fail to capture direction then people will not take it seriously. People who are part of this profession must establish credibility and impartiality.