p r i m e c o n c e r n

Sifting fact from opinion polls
By KV Prasad

HE Congress crying foul over opinion polls is seen as a ‘loser’s argument’. But the fact is unreliable opinion polls ahead of elections can sway the voter unfairly. 
The Tribune solicits views across the board, and finds favour for a check on quality of polls and a ban after election notification.
The decision of the Congress party in agreeing to a proposal of the Election Commission of India (ECI) to prohibit publication or dissemination of results of poll surveys between the period of election notification and actual voting has triggered a debate.







Sifting fact from opinion polls
By KV Prasad

What the parties say


THE Congress crying foul over opinion polls is seen as a ‘loser’s argument’. But the fact is unreliable opinion polls ahead of elections can sway the voter unfairly. 
The Tribune solicits views across the board, and finds favour for a check on quality of polls and a ban after election notification.

The decision of the Congress party in agreeing to a proposal of the Election Commission of India (ECI) to prohibit publication or dissemination of results of poll surveys between the period of election notification and actual voting has triggered a debate.

The proposal, on which the ECI sought political parties’ views, was based on suggestions emanating at meetings with various parties held at periodic intervals by the poll body to bring about reforms in the electoral process. The idea was to prevent influence such surveys could have on the voter’s mind.

That the move came close to elections in the five states of Chhattisgarh, Delhi, Madhya Pradesh, Mizoram and Rajasthan added spice to the endless rounds of discussions across television news studios.

The timing also gave political opponents like the BJP to expand their context, viewing it against the cornerstone of freedom of expression and speech guaranteed by the Constitution.

There are strong arguments on either side. Do political parties still consider that the Indian democracy has not come of age? Do these parties not have confidence in the maturity of the voter? Does liberal democracy disallow opinion and the voter to analyse what is best suited for her or him and the country?

There are many arguments against surveys based on opinion polls close to voting, but the foremost is that it can influence the ‘fence-sitters’, or the undecided voter, which American pollsters refer to as those occupying the middle ground.

While it is common to refer to the Gallup or other such polls that take place in the US on issues of import, it is not correct to extend the analogy directly to India because in the US the voters are largely registered either as a Democrat or Republican, with only a small portion remaining undecided. Then the presidential race is largely restricted to these two parties while others in the fray barely get notice as they are unable to match the big two in raising resources for campaigning.

The experience thus far in India also shows that the majority of opinion polls have not got it very right. While there have been exceptions in which certain pre-poll surveys — in Assembly as well as Lok Sabha elections — came close to the actual results, there is no denying that psephology (the science of predicting how a voter is going to exercise his franchise) is still in infancy. 

What goes wrong

Human behaviour: It is not the same every time. Unlike pure science, in which in a given condition the result would be the same, here making exact prediction can be difficult, as freebies by candidates to woo voters can affect the result.

Social composition: Irrespective of the sample size, representation by social and gender composition is very thin. If a constituency has 20,000 voters from varied social profiles, and from this one picks a sample of 20 voters, which is less than 1 per cent, the sample can at best be termed indicative but not representative. Adding to it, if the sample selected is based on social parameters in a constituency, it may throw up results that are skewed.

Questionnaire: The manner in which a question is asked, value laden or neutral, can determine a response. The response to a straight question like ‘which party do you prefer’ can be different from offering a list with options of parties.

Workers: Investigators are a key component. Their training is important and rigorous quality parameters have to be maintained for those collecting the data. A certain percentage of those interviewed need to be cross-checked. The geographical spread of the data and the time available can affect it. Sometimes an investigator may use own judgement in recording a response to responses like ‘cannot say’ or ‘do not know’.

Lack of transparency: There is little data on the methodology adopted, or how many investigators were trained, number of interviews conducted in a day across the terrain that includes rural, urban and semi-urban areas.

How opinion polls are conducted

Sample size

The number of people whose view is actually sought in a survey to get a sense of the larger opinion is called the sample size. By convention, 33 per cent of the total is considered a statistically significant sample. But on ground the size is largely defined by cost of the survey. Given the time and costs involved, smaller samples in the range of 2,000-6,000 for Assembly elections and 8,000- 30,000 for parliamentary elections are taken up. Mostly the polls are conducted in constituencies with more voters.

Bigger the sample, betters the results, i.e., less margin of error.

Drawing samples

It involves multiple levels. In opinion polls for Assembly elections there are five levels:

* Number of constituencies is taken into account to be covered (selection depends on the voters in that constituency)

* Within each constituency there are 4-5 blocks.

* Within each block there are 30-40 gram panchayats.

* A gram panchayat comprises villages and hamlets.

* Within a village social composition, caste and religious factors play a role.

Representing all

The sample is drawn at level 5, stratified based on social and gender composition. This also increases the cost (more segregation and representation, higher the cost).

Random picking

In house-to-house surveys, respondents are randomly picked from the last election voting list. A polling station generally serves as primary sampling unit, the houses attached to a particular station.

Large chunks

To save cost and time, large polling stations are picked. If it is a village, then villages with population of 5,000-6,000 are taken up.


‘Don’t ban, but transparency must’

Free and fair polls are as important as free press and free speech for a vibrant democracy. These are two sides of the same coin of democracy. Banning poll surveys is no answer, nor a solution for the current concerns. But manipulation of poll surveys could be worse. Both should be avoided, if not voluntarily, then through public discourse and consensus in such a way that we have some understanding of what should be avoided and what should be ensured for a transparent electoral process.

This requires some understanding of the “effects” of poll surveys reported in the media, instead of arguing that they do not make a difference to the poll process or the voting behaviour. If that be so, then why so much ‘hungama’ about the surveys? That has been an industry viewpoint, and as an interested party, the media has been perpetuating those views as well.

I had written more than 25 years ago and also did research on the effects. These effects are not uniform or similar every time. These can trigger various types of reactions, including ‘bandwagon’, ‘underdog’ and ‘tactical’ voting or complacency.

Against this context, I support the Election Commission’s efforts to carry out its mandate — free and fair elections. Its initiative is to elicit and involve political parties and in the process sensitise the larger public and remind other stakeholders of opportunities that they have in availing poll surveys in public media.

Referring to experience of other countries in this regard is irrelevant, given the kind of peculiarities, differences and inequalities we have across the country. Arguing that many countries have no regulation on this is not true and even misleading.

I oppose complete ban of poll surveys. But if and when they are put out for the larger public through mass media, they need to be transparent and stand up to electoral codes of the country. The best bet in that course is for the stakeholders to take precautions on their own so that the poll process is not vitiated any way, even unwillingly.

This view is even more relevant today as public opinions are far more dynamic and vulnerable now than ever before.

N Bhaskara Rao, a pioneer in poll surveys and chairman of the Centre for Media Studies

‘Scientific polls, not party sponsored’

Instead of thinking on the lines of clamping a total ban on opinion and exit polls, the government should tighten the existing regulation that prevents telecast/publication of such polls for a specific period — from 48 hours before the commencement of polling in the first phase of election till the end of the last phase. The prohibition could kick in a fortnight or month ahead of the poll. This would give the voter sufficient time to ponder over the parties and candidates and make a choice uninfluenced by the opinion poll results. In fact, the EC had taken the initiative to ban such polls but dropped the idea after the Supreme Court observed that it would be against people’s fundamental right to freedom of speech and expression guaranteed under Article 19 (1) (a) of the Constitution. The regulation should also ensure that opinion polls are conducted scientifically and not sponsored by political parties or targeted against individuals. Opinion polls should be on the basis of parties’ policies and performance.

TS Krishnamurthy, former Chief Election Commissioner

‘Freedom of speech above all,

Inherent in the right to freedom of speech and expression enshrined under Article 19 (1) (a) of the Constitution is the right to conduct and publish opinion polls. The Supreme Court has held that the right encompasses the right to inform others and the right to be informed. Any ban on opinion polls is not legally permissible because it will take away this right. The rationale for the proposed ban is the alleged rigging of opinion polls. The mischief can be addressed by regulations to ensure transparency. Merely because an actual election can also be rigged is never sufficient ground for banning an election. The remedy lies in ensuring the election is fair. The same is true of opinion polls. The remedy lies in rectifying the alleged malpractice and not a ban.

The apex court has placed freedom of speech and expression on a higher pedestal because it is one of the most important rights in Part III of the Constitution. Imposing a ban would be worse than the alleged malice of fixing opinion polls. Because paid news is common, would we ban news altogether? Since the ruling party does not seem to be doing well in the opinion polls, the timing of the ban would also be questioned on the ground of mala fides. A ban would be completely out of line with the development and expansion of fundamental rights as enunciated by the Supreme Court.

Nidhesh Gupta, Senior Advocate

‘EC only eliciting views’

It is for the government to take a call on the issue. The Election Commission sought the views of about 50 national and regional parties. So far, 11 parties have sent in their suggestions. We are like a post office. We will forward the political parties’ views to the government. It is the government’s duty to take a decision. If we are asked to hold further consultations with political parties, we will do it.

HS Brahma, Chief Election Commissioner



HOME PAGE | Punjab | Haryana | Jammu & Kashmir | Himachal Pradesh | Regional Briefs | Nation | Opinions |
| Business | Sports | World | Letters | Chandigarh | Ludhiana | Delhi |
| Calendar | Weather | Archive | Subscribe | Suggestion | E-mail |