The effect of opinion polls on voters
While the jury is still out on banning opinion polls, pollsters need to fill a credibility gap on methodology, samples
In the opinion poll conducted by the Aam Aadmi Party (AAP) and Cicero Associates in August, there was a question on the voter’s opinion on who is likely to form the next government in Delhi. It can help us understand the issue of the bandwagon effect, which has been much discussed in the past few days in relation to the proposed ban on opinion polls.
In this survey, 818 respondents out of some 3,200 stated that they would be voting for the Bharatiya Janata Party (BJP). Of these, 632 (77%) believed that BJP would come to power in Delhi. Among the 698 respondents who said they would vote for the Congress, 541 (78%) expected it to come back to power. As many as 599 of the 659 (91%) people who said they would vote for the AAP said it would come to power.
It can be easily noticed that each of these three proportions (77%, 78% and 91%) are incredibly high. Considering that only one of the three parties can come to power (assuming the assembly isn’t hung), it can mean one of only two things—either these respondents are such ardent loyalists of the party they support that they irrationally believe that their party will form the government, or they have decided to vote for a particular party because they believe that party is likely to form the government.
It is this latter effect that is being used as an argument in support of the proposed ban on opinion polls.
The theory goes that people like to be on the winning side. If a voter doesn’t have a strong opinion, he is likely to lean in favour of the side that he thinks is winning (unless he is Don Quixote).
Hence, the undecided voter is likely to lean in favour of the party that is seen to be leading the race, thus bolstering its chances at the expense of the party seen as a loser.
There is another factor that accentuates this effect in Indian elections.
Over the last 20 or so years, voters have seen a number of hung legislatures (both at the centre and in various states), and the experience with such legislatures hasn’t been great. Voters want to give a clear mandate to a particular party or coalition that they believe will give them a strong and stable government. This will again result in voters voting for parties that they believe have a good chance of getting a clear majority.
However, to counter this bandwagon effect, there is the underdog effect. By this effect, the party shown to be losing in the opinion polls redouble their campaign efforts, and its supporters come out to vote in larger numbers so that their party does better than projected.
There has been plenty of academic research so far in studying the bandwagon and underdog effects, and the results have been mixed, though there has been greater evidence in favour of the bandwagon effect.
In March, Karnataka saw an interesting natural experiment. Barring Bangalore, all urban local bodies in the state went to the polls two months before the assembly elections. While the results in some of these local bodies were hung, the overall verdict was clear—aggregating across all local bodies, the Congress won more seats than the BJP and the Janata Dal (Secular) put together.
In the subsequent assembly elections, the Congress won a clear majority, winning 122 out of 224 seats.
Most opinion and exit polls, while acknowledging the Congress as the single-largest party, had predicted a hung assembly. It is possible that the local body elections showed voters which way the wind blew, and this marginal vote gain by the Congress helped push it to a comfortable majority.
There is one necessary condition for either the bandwagon or the underdog effect to take place—opinion polls need to be credible, and the situation in India is currently far from that.
Apart from CSDS-Lokniti, no opinion pollster in India deems it necessary to publish the methodology. We are forced to accept the numbers without any information on how many people were surveyed, how those people were chosen and what is the margin of error in prediction. Without this information, it is impossible to take any opinion poll seriously.
In the US, trade bodies such as the American Association for Public Opinion Research or the National Council on Public Polls (NCPP) enforce stringent standards on their member organizations.
According to the NCPP website, for example, any opinion poll whose results are released to the public needs to disclose information on the sponsor of the poll, sample size, methodology and margin of error, among other things, and “endeavor to have print and broadcast media include the above items in their news stories”.
A similar effort in India is necessary for polling agencies to gain credibility.
In the context of the recent proposed ban on opinion polls, it would be interesting to see how other mature democracies deal with the issue.
In the US, Australia and South Africa, among other countries, there is no regulation on the conduct and publishing of opinion or exit polls. In the UK, while there are no restrictions on publishing the results of opinion polls, results of exit polls cannot be published until voting is finished. In Canada, results of opinion polls cannot be published during the last 72 hours before polling. This blackout window lasts 24 hours before polling in France and five days in Russia.
While the length of this blackout period on opinion polls varies by country, nowhere is there a blanket ban for extended periods of time as has been proposed in India recently. A seven-day blackout window imposed by France in 1977 was overturned following a court order that deemed it to be a violation of the freedom of expression.
Editor's Picks »
- Why Indian paint makers are shifting to water-based paints
- 2019 elections still some time away but defence stocks get the jitters
- Complan and Horlicks sale signals low energy in health drinks market
- With fall of the last dove, MPC minutes portend more than one RBI rate hike
- RITES IPO ticks the valuations box, but not the growth one