A survey is like taking the first sip of coffee and decide whether it’s strong enough for you Image Credit: Shutterstock

Until a few days ago the “Khan Market consensus” was that India’s ruling Bhartiya Janta Party was likely sweeping the February 2022 elections in Uttar Pradesh, an election as important as the midterms in the United States.

Then, last week, Prime Minister Narendra Modi made a shock move, announcing that the upcoming session of the national parliament shall repeal three controversial laws that sought to enable privatisation of agriculture.

The pundits started the post-facto explanations immediately: the BJP was clearly losing enough votes in western Uttar Pradesh, one of the hotbeds of agitating farmers, that they had to eat a rare humble pie. Of course the main centre of the agitation is Punjab, but the BJP has no chance of winning the election there. And then there’s Haryana, where elections are a while away.

As always, everyone says they knew there was enough trouble in west UP. Everyone’s wiser after the fact.

Call the voter

This disjunct is a good example of why the Indian media would do well to have more phone “tracker” surveys that give a weekly sense of public opinion, just like in the US or UK.

We regularly read about the rise and fall in the “ratings” of US presidents, not just in campaign season but in regular course too. We have some such surveys in India too, but they are occasional and election-centric.

There are two kinds of surveys: one conducted on the field, and one through phone. The phone surveys are faster to execute and give a running sense of changing public mood. The news channels which commission surveys are still too hung up on a big, unwieldy field survey just before elections begin, and then an exit poll. That’s that.

It is argued that field surveys are of better quality, because people are less likely to make a joke of them. But someone who has overseen phone surveys tells me that people are often honoured to be asked their opinion. It is only a few elite folks like us who are likely to give wrong answers just for fun.

Field surveys also have their problems and biases — the field surveyor is prone to fudging data to reduce the leg work, or survey the skewed sample set around markets and shopfronts. These problems have their solutions, but ultimately the field survey takes too long to conduct to be useful to gauge the change in public sentiment. India Today puts out a “Mood of the Nation” survey just twice a year.

Political parties and consulting firms are ahead of the media in this regard. They are directly or indirectly carrying out phone surveys that ask people the same question again and again over months.

Trends and variations

The only phone tracker survey available in public domain is by Cvoter, though they don’t yet put it out on a weekly basis. They have so far put out three surveys regarding Uttar Pradesh. Whether their “numbers” are accurate or not is not the point.

It is more interesting to see the trend and the variation — if you look at the three surveys, there is a clear trend of the Samajwadi Party picking up steam and the BJP’s lead reducing. Looking at this trend, one can say the Samajwadi Party is picking up because they started campaigning at long last.

Early this year, Cvoter was kind enough to give me access to their daily tracker dashboard for assembly elections in West Bengal, Assam and Kerala. Seeing the daily trend and weekly averages, one was able to make sense of why political parties were doing what they were doing.

In Kerala, I saw the Congress line go up after it handed over their campaigning to Oomen Chandy. This made sense because the survey also showed Oomen Chandy being the most popular Congress leader in Kerala by a long distance. However, the Congress was too late in making the move, since Pinarayi Vijayan and the CPM were consistently ahead.

In other words, tracker surveys in the media can help political discourse go beyond “who got the prediction right”. In West Bengal, Cvoter showed how the Trinamool Congress picked up a percentage point or two after Mamata Banerjee’s accident — clearly, the mass leader got the extra sympathy it needed.

The survey showed the Trinamool Congress consistently ahead, but the BJP was ahead in “perception gap”. This meant that many who were not voting for the BJP thought the BJP was winning anyway.

As polling day came closer, the Trinamool exceeded the BJP in the perception gap too. If you look at the final result — the Trinamool Congress’ sweeping victory — the survey made sense.

From predictions to understanding

The issue is not cost, because it doesn’t cost much to set up a call centre. It costs a lot less than hiring field researchers and making them travel.

The issue is one of a mindset: news channels need to get out of the prediction mindset and see surveys instead as a continuous process to help understand public opinion and its continuum, catching even small shifts.

The news media wants big shifts in every survey so that it can have eyeball-grabbing headlines. However, public opinion shifts gradually and it is important to have a close eye on these gradual shifts if we have to understand and reflect public opinion in real time.

Is inflation growing as a concern for the people week upon week? Or, for example, what has been the immediate aftermath of the BJP’s farm law retreat in its popularity, and on the approval ratings of PM Modi? Is there any significant increase in the percentage of women wanting to vote for the Congress party in Uttar Pradesh after its promise to give 40% tickets to women?

The right demographic mix

Another mindset shift needed is about sample size. It is a myth that the bigger the sample size, the more trustworthy the survey. A survey is like taking the first sip of coffee and decide whether it’s strong enough for you. You don’t need to gulp down half the coffee to make up your mind.

It is more important to get the demographic distribution right, reaching voters you can’t easily find on the roadside, making sure all castes, communities, age groups and geographical regions are proportionately represented in the final analysis of the survey.

This is why “online” polls in India are by definition nonsense. They don’t measure demographics and exclude the digital have-nots to begin with.

Like all opinion polling, many daily tracker surveys will also be off the mark, because it’s a difficult science, and some will be fudged for propaganda purposes, but if we have at least 3-4 regular weekly surveys we can match them with each other, and our own sense of the ground.