In Prannoy Roy's book on India's electoral history, a look at why 'landslides' can be tricky for pollsters to correctly forecast

  • The Verdict uses psephology, original research and as-yet-undisclosed facts to talk about the entire span of India's electoral history.

  • Written by Prannoy Roy and Dorab R Sopariwala, the book provides pointers to look out for, to see if the incumbent government will win or lose.

  • The Verdict is published in India by Penguin Random House.

Published on the eve of India's next general elections, The Verdict: Decoding India’s Elections uses psephology, original research and as-yet-undisclosed facts to talk about the entire span of India's electoral history from the first elections in 1952.

Written by journalist and NDTV co-founder, Prannoy Roy, and channel’s editorial advisor, Dorab R Sopariwala, the book provides pointers to look out for, to see if the incumbent government will win or lose.

The following is an excerpt from the book discussing "why Indian elections should be a dream to forecast". The Verdict is published in India by Penguin Random House.


India has more landslide elections perhaps than any other country. This simple rule makes forecasting elections in India the ‘safest’ country to make predictions—and so we dubbed ourselves ‘landslide forecast experts’. We are now becoming a little worried about this nickname. Landslides are theoretically easier to forecast than close fights. Even with a sampling error, a forecast in a landslide election will still at least name the correct winner—even if the number of seats is wildly off.

One reason for the large number of landslides is a combination of statistics and passion. Statistically, a first-past-the-post electoral system tends to generate large and stable majorities. When it is combined with a fragmented Opposition, landslides are the norm. In addition, we have encountered a deep passion and belief in the power of their vote among people we have interviewed. The most common sentiment we come across sitting at chai shops in the small towns of India is, ‘The MP (or MLA) hasn’t even come to see our town once in five years, and now he comes begging forgiveness. Just watch us kick the scoundrel out of the seat in this election.’

 In Prannoy Roys book on Indias electoral history, a look at why landslides can be tricky for pollsters to correctly forecast

The Verdict, by Prannoy Roy and Dorab R Sopariwala

The passion is also often positive, especially when it is combined with non-economic issues. In the recent elections in Telangana where a mood of sub-nationalist pride invigorated the electorate, the cry of ‘Jai Telangana’ led to a huge landslide for the leader who many said to us ‘had struggled for thirty years to give us our own state’. The passion of nationalism and sub-nationalism is, and has always been, one of the most effective mobilizers of public support and votes.

What is a ‘landslide’: A clear victory is often defined as a ‘sweep’ or a ‘landslide’, for example, the largest party wins a least twice as many seats as the runner-up.

In India the magnitude of landslide victories is even greater. An analysis of all the sixteen Indian Lok Sabha elections, from 1952 to 2014, reveals that the largest party/alliance on average wins nearly 60 per cent of the seats. This, more importantly, is on average three times the seats won by the runner-up party/alliance. By any global standards a 3–1 victory is winning by an enormous landslide.

The same analysis for votes rather than seats, shows that the percentage of votes that the largest party/alliance gets is twice as many as the runner-up party/alliance.

In a landslide this wide margin of victory between the winner and the runner-up clearly makes poll forecasting simpler. The 20 per cent margin of victory of winning parties in India allows for high error rates in polls. Even a large sampling error of about 3 per cent will still forecast the right winner even if it gets the seats and votes wrong.

Lucky Indian pollsters? Not quite, as we shall see later.

In India, where a fragmented Opposition is the norm, when the largest party wins 60 per cent of the seats it usually also wins twice the number of votes won by the runner-up. Both definitions generally apply to landslides in India. Consequently, under either definition, Indian electoral history points to a story of almost the same number of landslides.

From 1952 to 2019 a remarkably high 77 per cent of all elections in India have been landslides, when computed for each state separately in Lok Sabha elections.

Over the years the percentage of landslides, even at the level of each state, during Lok Sabha elections has not changed. Indian landslides have remained at around 75 per cent and above and stayed consistently high over the years. The high ‘landslide rate’ is even more remarkable given that Lok Sabha elections are now more analogous to a combination of a series of simultaneous state elections for the Lok Sabha than a single unified all-India national Lok Sabha election. For instance, the issues and party combinations in the Lok Sabha elections in Tamil Nadu are completely different from those in West Bengal or Maharashtra.

If Lok Sabha elections are landslide-ridden, what about the history of State Assembly elections?

State Assembly elections have a lower rate of landslides. There are a couple of major differences between elections to State Assemblies and Lok Sabha elections. First, let’s focus on the recent period 2002–2018. State Assembly elections now have a lower rate of landslides of around 53 per cent (compared with 74 per cent for the Lok Sabha). A 53 per cent landslide rate is still high by global standards, but it makes these State Assembly elections a little trickier to forecast than Lok Sabha elections.

The second difference is that unlike Lok Sabha elections, the landslide rate in State Assembly elections has changed significantly as state politics have become intensely competitive over time. As a result, there has been a clear drop in the proportion of landslides recently.

In big and medium-sized states, the high landslide rate of 83 per cent in the early post-Independence ‘honeymoon’ phase has now dropped to only 53 per cent. A similar trend of a declining landslides is true of the small states of India.Another exercise when looking at Lok Sabha elections is to compare the results of simultaneous (Lok Sabha and State Assembly elections held together) with non-simultaneous elections.

In non-simultaneous elections, the winning party does almost as well in the Assembly segments that make up the Lok Sabha seats. While in the Lok Sabha seats the largest party wins 72 per cent of the seats on average, this drops by only 5 per cent, in the corresponding Assembly segments.

However, in elections held simultaneously for both the Lok Sabha and the State Assembly, local issues appear to impact the result. The success rate of the winning party in Lok Sabha seats drops by a significant 9 per cent in the corresponding Assembly segments. And even more significantly, the largest party wins only 55 per cent of the seats in the actual State Assembly election, a major drop of 15 per cent from the 70 per cent success rate in the Lok Sabha elections.

Earlier we had indicated that a higher turnout in State Assembly elections could mean a greater interest in local issues. This is further buttressed by the significant impact that simultaneous State Assembly elections have on the success rate of Assembly segments in Lok Sabha elections. Yet another sign of the strong influence and importance of state assemblies in comparison with the Lok Sabha for the voter. This of course is a preliminary finding and hopefully will generate more research in this area.

Returning to the analysis of landslides, one reason for the drop in the number of landslides in State Assembly elections in the last twenty years is increasing competitiveness and the end of the dominance of the major national parties. The growth of regional parties in India and their increasing importance in State Assembly elections where local issues are more important is significant in the history of India’s democracy. A factor that has led to the rise of regional parties is the first-past-the-post electoral system which rewards parties with many more seats per percentage vote if, and only if, their vote is more concentrated in a region rather than spread out over the country.

While the lower rate of landslides makes forecasting a little more difficult, it is compensated by the greater homogeneity of the voter population in states, which makes sampling for State Assembly elections a little simpler.

So for pollsters, getting their forecast correct calls for a balance between the negative influence of greater heterogeneity of the all-India sample during Lok Sabha elections, which complicates the sample design, and the positive effect of the higher proportion of Lok Sabha landslides that is tolerant of a higher margin of error.

While all these landslides suggest that it should be a dream for Indian election forecasters to make the correct ‘call’ about who will get a landslide, in reality this is not wholly true. For one thing, it takes guts to forecast a landslide! Every time pollsters predict one, they really stick out their necks. To get it wrong is to be discredited and forever reminded of it, while getting a ‘close fight’ wrong is almost forgiven—if not forgotten. This is one key reason why poll forecasts tend to underestimate the winner, as pollsters (like voters!) ‘play safe’.

Summing up, Indian Lok Sabha elections are easier to forecast because of the unusually high proportion of landslides averaging around 77 per cent which allows for a larger room for error. However, while they once used to be in high-landslide territory, the incidence of landslides in State Assembly elections has come down dramatically dropping from over 80 per cent, to a landslide rate of just over 50 per cent now. This means pollsters find forecasting Lok Sabha landslides a little less testing than forecasting State Assembly elections. Though the relative homogeneity of the voter population in a state compared with the enormous diversity at the all-India level makes sample design a little easier at the state level.

The Great Diwali Discount!
Unlock 75% more savings this festive season. Get Moneycontrol Pro for a year for Rs 289 only.
Coupon code: DIWALI. Offer valid till 10th November, 2019 .

Updated Date: Apr 10, 2019 20:45:01 IST