No, Analysts Did Not Get Haryana 'Entirely' Wrong

The problem is that we start predicting seats based on vote share. A seat can be decided with a margin of one vote.

Leader of Opposition in Lok Sabha and Congress leader Rahul Gandhi with former Haryana CM Bhupinder Singh Hooda during a public meeting in Hisar.

Photo Credit: PTI

Whereas a majority of experts predicted a comfortable majority for the Bharatiya Janata Party (BJP) during the Lok Sabha elections, including some with outlandish claims of the BJP crossing 350 on its own, almost everyone who claims to be an election analyst gave a decisive advantage to the Congress party in Haryana in the recently concluded assembly elections.

Of course, we ended up with eggs on our faces. While such treatment may seem fair given the differences in the number of seats predicted and the final result, the predictions about Haryana were not as big a blunder compared to those of some states during the Lok Sabha elections (Uttar Pradesh and West Bengal, for instance).

The problem lay not in what we observed but how we ended up presenting it in the media.

First-Past-the-Post

In December 2018, the Congress party came to power in Madhya Pradesh after 15 long years. In that election, it secured 40.9 percent of the popular vote and won 114 out of the state’s 230 assembly seats. The BJP came a close second, winning 109 seats.

However, the BJP’s vote share was 41 percent, i.e., higher than the Congress'. A simple reason for this counterintuitive result, where the more popular party won fewer seats, is that wherever the BJP won a seat, on average, they did so with much higher margins than the average margin on the seats secured by the Congress.

A similar situation had played out earlier that year in Karnataka, where, despite trailing the Congress by around two percentage points, the BJP secured 104 seats, and the Congress merely 80 seats.

Put simply, the conversion of state-wide trends into seats in our electoral system (first-past-the-post) is fraught with risks.

This is unlike the system of proportional representation, where legislatures are formed by parties based on their respective vote shares. As a result, in our system, predicting seats with a rough idea of state-wide trends can happen in two ways — the usage of algorithms that rely on certain assumptions or educated guesswork.

What We Can Observe and What We Can’t Predict

Most analysts I spoke with who went to Haryana in the run-up to the election returned with three broad observations (as did I).

First, the Congress party seemed to be gaining big across large parts of the state, especially within that stretch of the state popularly called the Jatland which is home to roughly half the seats in the state’s legislative assembly.

Second, we also witnessed that, like other states in the Hindi Heartland, Haryana was turning towards bipolarity with little to no space for the regional parties like the Indian National Lok Dal or the Jannayak Janta Party, which, in their combined avatar, were a formidable force in the state’s politics.

Finally, some of us who actually spoke to more representative samples in places like South Haryana and along the GT Road belt did realise that BJP voters might be dissatisfied with their candidate and/or the former chief minister but were largely sticking to the party. From my own experience, the last observation was much truer of South Haryana than the GT Road belt.

Were any of these observations wrong? Let’s look at the data.

The Congress party had secured a 28 percent vote share in 2019. In 2024, it won over 39 percent of the total votes. Is that a trivial gain? Absolutely not. This is what we were observing on the field.

At the same time, the former deputy chief minister of Haryana, and the co-founder of the JJP, Dushyant Chautala, ended up in fifth position in his own seat of Uchana Kalan. Indeed, the JJP was decimated with a vote share of less than one percent.

It had secured almost 15 percent of the total votes cast in 2019. The sum total of the Congress and the BJP’s vote shares swelled up to almost 80 percent from around 65 percent in 2019. Again, this is what we were observing on the ground.

Finally, not only did the BJP hold on to its vote share, it actually increased it by over three percentage points. I admit that, like many others, while I did see the BJP holding on to its core vote bank (non-Jat communities) in both South Haryana and along the GT Road belt, especially the latter, I had concluded that anti-incumbency might cause enough disenchantment among enough of them to dent the BJP’s seat tally.

I was wrong, and it seems like many of us were.

Broadly speaking though, analysts and credible pollsters that conduct surveys of the state or specific regions within the state (and not individual seats) did catch all these trends correctly.

CVoter, for instance, captured the BJP’s vote share at 37.2 percent, which is within the error margin (+/-3 percentage points) of the final tally, i.e. 39.9 percent. They predicted the Congress' vote share at 43.7 percent, with a lower bound of 40.7 percent. The Congress, along with its ally, the CPI (M), polled 39.4 percent, which, admittedly, is an error of 1.3 percentage points, but is in line with the monumental gain that the Congress secured when compared to 2019.

Unfortunately, though, even the news coverage that compares exit polls with the final results is obsessed with seat tallies, which is understandable but unfair to the science of polling, which is more concerned with vote share and not seats.

The problem begins when we start predicting seats based on these vote shares with either imperfect prediction methods, as is the case with the category of polling agencies, or on educated guesswork, which is usually the case with experts who travel on the field.

A seat can be won or lost with a margin of merely one vote. Both these methods are, therefore, vulnerable and prone to errors. So, is there no way that we can call seats precisely?

It would seem that running seat-by-seat surveys may be a solution. Although much more resource-intensive than running surveys based on a representative sample at more aggregate levels, such an exercise should ideally yield accurate seat tallies.

But the fact is that even those polling agencies that conducted seat-by-seat surveys failed to predict the BJP’s win.

We Were Not Entirely Wrong About Haryana

Immediately after Congress' loss, commentators predicting a sweep were on the back foot, and several of them have turned to conventional theories behind the grand old party's loss — over-centralisation in the hands of the Hooda family, Jat v non-Jat polarisation, alienation of Kumari Selja, etc.

However, these factors were always at play, and counterintuitively enough, some of these reasons are actually responsible for the Congress' surge. Therefore, the way I see it, there is only one reason behind its loss, which is its inability to manage the election with respect to individual seats.

Analysts need not be defensive about their predictions, for most of us did see the above three trends holding. Eventually, the sum total of margins by which the Congress lost the nine seats which would have ensured a majority, is less than 23,000 votes.

The number for the next 13 seats is less than 47,000 votes. Compare these numbers against the Congress' gain of roughly 20 lakh votes between 2019 and 2024. Of course, there was a Congress wave, which we all picked up. What we should not have done is predict seats, for with state-level trends, there is no reliable way of doing that correctly.

While many theories can be floated behind the Congress huffing and puffing right next to the finish line, none is clearer than the role of independents, some of whom were backed by the BJP and others backed by leaders within the Congress. Anybody who followed this election knows that seats like Pundri and Ambala Cantonment were gifted to the BJP.

Similarly, more than the loss in the state, the Hooda family’s loss of face in Sonipat district, one of its bastions, where it could win only one out of six seats, can be attributed to the Congress' inability to fight against the BJP’s micromanagement.

While one hopes that the Congress party will introspect, even as the BJP revels in its glory, as analysts who deal with state-level trends, we should certainly introspect on this habit of predicting seats, notwithstanding the pulls and pressures exerted by the viewers and other consumers of such numbers. I certainly will.

(Dr. Kartikeya Batra is a Visiting Fellow at the CVoter Foundation, and an incoming faculty member at Azim Premji University. He holds a PhD in Economics from the University of Maryland. This is an opinion piece and the views expressed above are the author’s own. The Quint neither endorses nor is responsible for the same.)