Why media ground reports, opinion & exit polls don’t agree with each other on elections

News reports are vital to understand political strategy & local issues, but these factors do not affect the seats tally in a big way.

Representational image | Dhiraj Singh/Bloomberg

With just two phases of Lok Sabha elections left, all eyes are on the counting day, 23 May. So far, we have heard two kinds of accounts.

Post-Pulwama and Balakot, we had a number of opinion polls and forecasts predict that the BJP would be clearly ahead, and well placed to form the next government.

Now, with the beginning of polling and the ban on forecasting, we have seen a vast range of ground reports in the media. Most of these reports suggest a mixed picture: a ‘setback’ for the BJP in the first two phases in Uttar Pradesh, vigorous campaign by the opposition in Bihar, the AIADMK-led alliance queering the pitch for the DMK in Tamil Nadu, and so on.

Now, there are two possibilities. Either the political trends have changed since the final set of pre-election forecasts, thanks perhaps to the cooling down of the Balakot effect. Or, the ground reports are using a different lens and end up with a different picture of the same reality. I suspect it is the latter. Ground reports and opinion poll-based analysis always offer very different snapshots of an election. Both these are valuable, as long as we know how to use these.

Worms-eye view vs birds-eye view

Permit me an autobiographical interlude. With this election, I complete three decades in professional election-watching. I made my first amateurish election forecast in 1989 at an obscure departmental seminar in Panjab University, Chandigarh. This was followed by the first exit poll I conducted for Chandigarh Lok Sabha constituency in 1991.

My joining CSDS in 1993 was followed by a series of election-related surveys, more than I cared to count, till I formally announced my retirement from election forecasting in March 2012. Since then, with a few exceptions when I was tempted or compelled, I have kept away from the business of election forecasting.

Looking back at the hundreds of elections over these three decades, I see a pattern. There are two ways of looking at and reporting about elections: ground reporting and survey-based analysis. Let us call them “worms-eye view” and “birds-eye view” respectively.

Ground reporting is preferred by political reporters, fieldwork-based political scientists and most political workers and leaders. They use direct personal observation, extensive travel, in-depth knowledge of local politics and conversations with a few ordinary voters and political workers. They watch the body language of leaders, the attendance and the mood at political election rallies, hoardings, flags, and other visible marks of political loyalty. Based on all these indicators, they arrive at a qualitative assessment of the ‘hawa’. This was the only way to anticipate electoral outcomes in the first three decades of Indian elections.

Ever since Prannoy Roy’s path-breaking forecasts in India Today for the 1980 and 1984 elections, the sample survey based election analysis has come to stay. Although the word ‘psephologist’ covers any and every student of elections, in India this label is used mainly for this newfangled science of forecasting election outcomes.

Typically, a psephologist carries out a large-scale sample survey, processes it to estimate vote shares for different political parties and then uses mathematical modeling to convert these vote estimates into seat forecasts. Instead of selective pick-and-choose method used by journalists, a psephologist uses statistical techniques for picking a representative sample. Instead of using indirect indicators to infer voting preferences, s/he directly asks the voters who they intend to or have voted for.

Mixed picture vs clear wave

In Europe and North America, the job of election forecasting has been handed over entirely to psephologists. There is a consensus that election forecasting requires a level of precision that only scientific surveys can bring. Journalists, political commentators and political leaders don’t compete with polls in the prediction game. They supplement sample surveys by doing in-depth stories, analysing political strategies and political outcomes,besides providing feel and colour to the election coverage.

In India, we have not yet arrived at this division of labour. After a dream debut in the 1980s and a decent record in the next two decades, pollsters have smudged their reputation in the last decade or so. Too many polls, over-hyped claims, large proportion of dubious or fake polls, and absence of professional code regarding transparency have tarnished the image of psephology. As a result, political reporters and commentators are still in the business of election forecasting, although this is not their forte.

Phillip Oldenberg, an American political scientist, wrote an insightful article ‘Pollsters, Pundits and a Mandate to Rule: Interpreting India’s 1984 Parliamentary Elections’ in The Journal of Commonwealth and Comparative Politics on why all political commentators missed the 1984 Rajiv Gandhi wave. His explanation was simple: there is a triad of journalists, “knowledgeable persons” and political leaders who form a commonsense about election outcome. They keep speaking to one another and reinforcing each other’s reading. This commonsense local wisdom is then stitched together by political analysts and commentators and transmitted through the media. A pollster bypasses this triad and speaks directly to the voter and thus has a better chance of getting to the truth. This is particularly so in a “wave” election: “worms-eye view” tends to see a mixed picture where “birds-eye view” shows a clear wave.

Election reports & exit polls

Now cut to 2019. The news reports we are now reading are vital to understand the strategies of political parties, local issues, impact of candidates and shifts of various caste-community groups. But ultimately, these factors tend to cancel one another and do not affect the seats tally in a national polls in a big way.

“Worm-eye view” is not the best way to anticipate what voters do on the polling day. It is particularly unhelpful if the elections turn out to be a “wave election”: electoral waves are invariably recognised in retrospect. Despite occasional failures, psephologists have done much better in estimating vote shares of different political parties, including in “wave elections”. Good polls also provide a robust picture of which age group, gender or caste-community voted for whom.

So, do read election reports as they come. But wait for the exit polls on 19 May to form a broad sense of what the outcome might be. Also, depend on the polls to see who voted for whom. And then, listen carefully to political commentators and analysts to answer why, how and so-what.

The author is National President of Swaraj India. Views are personal.