The failure of exit polls hold behavioural lessons for us

What did you have for breakfast two days ago? Having breakfast is a routine affair. Not much thought goes into such a decision. So most of us would have forgotten the details of what we consumed as the first meal of that day. On the other hand, how many of us have forgotten who we voted for in the general election of 2024? Voting in a national election is not routine behaviour. 

For most people, voting is a much thought-through decision. It is unlikely that one would forget which political party one voted for, and this memory will probably stay intact for years to come. If this is so, why did most exit polls of the recent election fail to elicit this data from the immediate memory of voters with sufficient accuracy to forecast the outcome correctly?

The typical explanation that is provided for the unreliability of consumer research is that its survey methodology was flawed. Not selecting the right samples that accurately represent the larger voter universe of each constituency, for example, could be a significant source of error. 

No doubt, the Indian voter universe, with an electorate whose voting behaviour is influenced by multiple factors, is far more complex than in any other democracy. Once the field work for such a survey is done, using the responses of voters after they left polling booths to calculate the vote share of rival parties within every constituency is quite a task. 

Also read: TMC MP Saket Gokhale writes to Sebi for investigation into alleged stock market manipulation through exit polls 2024

Even with the best statistical methods, accurately evaluating the electorate’s verdict is far from easy. But the market research industry in India is several decades old and various polling agencies have conducted plenty of consumer research across the years. It is difficult to believe that they have not yet figured out the statistical methods needed to make sense of a complex country like India.

There could be other reasons for why India’s exit poll results were wrong. In his book Private Truths, Public Lies: The Social Consequences of Preference Falsification, economist Timur Kuran speaks of ‘ preference falsification.’ This, according to Kuran, is the act of misrepresenting one’s actual views under real or perceived social pressures. 

It happens frequently in everyday life, such as when we tell the host of a dinner party that we are enjoying the food served, although we may actually find it bland. Given the prevalence in society of such behaviour, could it be that some voters cast their vote for a particular political party, but when asked by exit pollsters, cited the name of another party? If so, why should voters do something in private but say something else in public?

As social beings, humans worry a lot about what others think of their views and actions. Many a time, they express falsely held views just to fit in with what the majority consider appropriate. This was best demonstrated by a famous experiment by legendary psychologist Solomon Asch.

Also read: Election Results 2024: Satta Bazaar trumps ‘smart’ pollsters, says Harsh Goenka, ‘believe in people who…’

In the experiment, an original line and three other lines were shown. Of the three lines, the first is clearly shorter than the original line, second clearly longer than the original, and the third is of the same length as the original. This experiment has one real subject and four actors whose identity the real subject is not aware of. 

As part of the experiment, the four actors are first asked which of the three lines is the same as the original. As part of the plan, they all point to the first line. Then the actual subject is asked the same question.

What is interesting is that when the subject is asked this question alone in a room, s/he will always get the answer right—which is the third line. But once the subject has heard what four others have already said, 35% of the time s/he will repeat the answer of the other four—just to avoid being in a minority of one.

Inside a polling booth, voters are confident that their behaviour is anonymous. So they tend to behave according to their personal beliefs. But outside the polling booth, there is pressure to fit in with the views of the majority. So it is likely that many respondents of this year’s exit polls falsified their answers to suit what they viewed as the majority view.

What we witnessed in these exit polls was a clear gap between what people say and what they do. This failure of most research agencies to really understand what people do—and why they do what they do—has larger implications.

Also read: Economic theory was at play all through the Lok Sabha election

An obvious question that the humongous failure rate of recent exit polls raises is whether consumer research is able to effectively understand human behaviour in other strategic areas of application. How good is a policy decision if we get most factors right but get the behavioural aspect of it wrong? 

What is the point in getting the engineering of new roads right, for example, if we can’t get drivers to drive safely on those roads? What is the point of inventing effective vaccines to prevent diseases if we can’t motivate people at large to walk to a nearby healthcare centre to get themselves vaccinated?

This failure of the vast majority of pollsters to correctly predict the mood of the nation is a reminder that making sense of human behaviour is a very tough job. There are far too many complex factors involved and there is no reason to believe that status quo techniques are adequate to effectively unravel the complexities of human behaviour. This is a field in which we need to think anew and act anew.