Media Resources For Readers For Journalists For Journalism Students Media Statistics Statistics Legal
Development Reporting Media Activism Media and Disability Media and Gender Media Diversity Privacy
Opportunism of pre-poll surveys
Survey methodology is good at explaining correlations between past and existing attitudes, but is poor at predicting future behaviour. And surveys can end up making the elections sound like a horse race, says ANUP KUMAR
Posted/Updated Monday, Nov 25 11:11:12, 2013

A controversy has been brewing over banning pre-poll surveys in India. The issue is worthy of a serious discussion – especially when it comes to transparency in methodology and the relevance of pre-poll surveys in democratic elections which require an informed, deliberative electorate.

The growing use of opinion polling in Indian elections reminds one of Jürgen Habermas prescient criticism of the opportunism of public opinion polls during German elections, which takes place at the cost of deliberation over political projects and differing visions.

A compelling reason behind the argument to ban pre-poll surveys in India is that such surveys can sway the voters to choose a party or an individual candidate supposedly leading in opinion polls, that is, have a bandwagon effect. There is also a potential drawback here for the party or individual who is leading in the poll – the supporters may become complacent and not show up at the voting booths.

The last round of pre-poll surveys, among others, included the IBN-Week Pre-Poll Survey, which was reported and discussed over four nights by ever politically passionate Rajdeep Sardesai on TV. The IBN-Week survey forecasted that this was the end of the road for the Congress Party. It would lose in the upcoming four out of five state elections – with only Mizoram likely to stick with the Congress Party -- although, in Mizoram too, there seemed to be an anti-incumbency factor at work.

The IBN-Week pre-poll survey was conducted by CSDS and Lokniti two organisations that are known to use sound scientific methodology in their sampling that is representative of the entire population. The survey showed the BJP could win in Chhattisgarh, Madhya Pradesh and Rajasthan. The fate of Delhi was a little uncertain because Aam Admi Party (AAP) had emerged as a real contender. Not surprisingly this has bolstered the confidence of AAP, which had earlier had released the results of its internal survey, a couple days after ABP-Nielson survey showed that the BJP could emerge as the leading party in the Delhi Assembly elections.

The analysis and the data by AAP on its website showed that the polling was conducted not only to learn where AAP stands vis-à-vis Congress and BJP, but it was also done to see among which demographics and constituencies APP needed to do more work in order to improve its chances in December. The surveys showed that AAP has emerged as a major electoral force in the Delhi Assembly elections. It looks like even if AAP does not emerge as the leading party, it will play an influential role in the formation of the next state government of Delhi. This will put the party in a good position to play a bigger role in the 2014 general election.

The Congress party described internal polls and pre-poll surveys as malicious and fake. Some would say it was not surprising that the Congress party has called for a ban on pre-poll surveys, especially after the nominations have been filed and the Election Commission has notified the campaigning period. Indeed, some years ago the commission with the support of all the major political parties had successfully placed a ban on media coverage of exit polls, during the voting period that these days is conducted in phases and is spread over many weeks.

But what often gets overlooked is that for short-term political gain pre-poll opinion surveys end up costing the larger political project democratic deliberation, making the elections a horse race and a media spectacle. There are compelling reasons why some social scientists argue that opinion polls not only measure opinion, but also foster false consciousness. Survey methodology is good at explaining correlations between past and existing attitudes, but is poor at predicting future behaviour.

In the West, polling has evolved into a multimillion-dollar industry since the first opinion polls was conducted in 1824 that successfully forecasted the victory of Andrew Jackson against Quincy Adams in the US presidential elections. In the 2012 presidential election, in the US, the controversy over opinion polls was big news. Republican pollsters and even the venerable Gallup, an independent polling company, forecast a Romney victory -- whereas, Democratic pollsters and Nate Silver of FiveThirtyEight Blog on The New York Times forecast an Obama victory. ( was recently launched as a news website in partnership with ESPN). However, the nasty public confrontation and name calling among pollsters in the U.S. in 2012 was not because they were debating the science of polling, but the subtext was mostly about how results of the polls have a “push” and “drag” effect on public opinion.

Yogendra Yadav, a pollsters and political scientist and now a AAP candidate in the Delhi Assembly elections, defended pre-poll surveys not only for development of internal party strategy during the campaigning period, but also to provide some idea to the public about the shifting political climate. For example, recently at women’s forum in Delhi, Yogendra Yadav called upon the AAP cadre to focus their attention on mothers and grandmothers – a segment of electorate among which AAP was lagging behind.

He did argue however that transparency in methodology should be a prerequisite and did not discount the possibility that pre-poll surveys could have some bandwagon effect on the political behaviour on the voting day.  

Conducting a scientific poll in India is very challenging because of complex, interlocking, demographic independent variables such as religion, ethnicity, caste and class which make randomised sampling excruciatingly demanding and can be a pollster’s nightmare. Not surprisingly, in India the reliability and the internal validity of the samples have made opinion polls quite unreliable. There are a few exceptions when it comes to local and state elections. To the puzzlement of many scientific pollsters, psephologists who rely on heuristic analytical tools based on historic trend, census data and social context have been more successful in forecasting election results.

To sum up the assumption behind advocating a ban on media coverage of pre-poll surveys, by some political parties, seems to be that such surveys can have undue influence on individual voter’s choice, and this can even happen when these polls are not conducted scientifically.


Finally whether pre-poll surveys have predictive power in the context of an election or not, a more pertinent question here is the value of conducting these surveys in a democracy other than their utility for development of political strategy inside a party and research. Critical scholarship in social science has especially questioned the value of media coverage of opinion polling during elections because it distracts the public and the media away from engaging in deliberative manner with the real issues and making an election a horse race -- who is up and who is down. 

Anup Kumar is assistant professor at the School of Communication, Cleveland State University

Such articles are only possible because of your support. Help the Hoot. The Hoot is an independent initiative of the Media Foundation and requires funds for independent media monitoring. Please support us. Every rupee helps.

Reuters is looking for an experienced editor with strong news judgment and rewriting skills for its Asia Desk in Bangalore. This desk complements the Reuters ed
The Asian Age in Mumbai and Delhi is looking for sub-editors and senior sub-editors for the news desk. Excellent language skills and an interest in news are man
Does India need an independent and autonomous public broadcaster?
Cannot say
Copyright2011. THE HOOT. All Rights Reserved.