New Delhi: When Anura K. Dissanayake was elected as Sri Lanka’s first left-leaning president, it was a victory propelled by the electorate’s “profound” sense of rejection of the entire system and the establishment, said Ravi Rannan-Eliya, who has been running Sri Lanka’s top systematic opinion poll.
For the first time since the economic crisis and the ‘aragalaya’ protest that ousted the Gotabaya Rajapaksa presidency, Sri Lankans went to the polls last Saturday. In the results declared on Sunday, the Janatha Vimukthi Peramuna (JVP)-National People’s Power (NPP)’s Dissanayake won 42.31% of the vote, with the Samagi Jana Balawegaya (SJB)’s Sajith Premadasa coming a distant third at 32.76%. The incumbent Ranil Wickremesinghe, who ran as an independent, obtained 17.3%.
The economy was, unsurprisingly, the focal point of the elections, with the population battered by the difficulties stemming from the conditions of the IMF bailout.
A trained public health specialist, who has also been trained in polling from Harvard University, Rannan-Eliya has run the Sri Lanka Opinion Tracker Survey (SLOTS) since 2021 as executive director of the Colombo-based think tank Institute of Health Policy.
It is the only opinion poll done in Sri Lanka using the multilevel regression and poststratification (MRP) model, which was once again proven successful in the recent UK general elections.
In an interview to The Wire, he discussed the 2024 presidential election, noting that the outcome largely met expectations. He pointed to a significant voter shift, where the electorate rejected the political establishment in favour of Dissanayake, known as AKD, as a candidate for change.
“People have finally lost patience with an establishment that doesn’t hear them. I think disgust is the correct word,” said Rannan-Eliya.
While AKD attracted support from urban, educated professionals, Premadasa struggled to gain traction despite being the leader of the main opposition at a time when disillusionment with the establishment was widespread across all classes.
Here are edited excerpts from the interview:
Were there any surprises?
It was mostly within our expectations, both geographically and in terms of the vote count. This is the first time we’ve conducted polling for a national election, so we anticipated some errors. There were a few issues, which we may have mentioned on Twitter, but overall it went smoothly. I believe the results fell within our margin of error, so I’m quite satisfied.
We expected Sajith to perform well in the North and East, ahead of AKD, which is what happened. We also expected AKD to do better across the South, especially in the central districts of Nuwara Eliya. In Badulla, we anticipated Sajith would fare better due to the CWC [Ceylon Workers Congress] block vote, as they were unlikely to support AKD. This too aligned with our expectations.
What about the votes AKD received in the Northern and Eastern provinces, despite finishing third or fourth?
Even in our internal district-level estimates, we anticipated he would pick up some support. I expected him to do slightly better in the Eastern Province and potentially win, not with a majority, but with a plurality. That didn’t quite happen, but he still performed respectably. Even Gotabaya Rajapaksa used to receive votes in these areas, so it’s not surprising.
One factor I hadn’t fully considered was the lower voter turnout in the North and East. We had assumed turnout would be more uniform, but it’s not surprising in hindsight. Historically, turnout has been lower in these regions in recent elections, so this was expected.
The only other notable detail is that the independent Tamil candidate garnered some support. Obviously, there is diversity of opinion, even in the North.
What was the main reason they were planning to vote for AKD?
I’ve been expecting this for almost a year now, but especially in the past three or four weeks, based on our recent interviews. I was expecting him to win, and I was expecting him to win on the second count.
There’s a whole set of things going on.
For a while, over the past two weeks, I did think there were some signs he might do even better and maybe win on the first count, but that didn’t happen. Still, I definitely expected him to win on the second count.
I’ve noticed for months that a lot of people who were planning to vote for AKD didn’t actually expect him to win.
I had a conversation on polling day in the morning – I got in a taxi, and I just chatted with the guy. He was listening to everything, and everyone was waiting. Then he said, ‘Oh, you know, I didn’t expect this result.’ He was surprised. Even the people who voted for the JVP were surprised by how it turned out.
It wasn’t really about the specifics of the policies. Most people don’t know them, to be honest. This was a change election, if there ever was one. People voted against the establishment, across all parties. It was a protest vote – not just against the establishment, but against an establishment that’s completely out of touch.
And that’s why I think the SJB might be surprised. I know some of their people were expecting they might win. But they’re out of touch. It’s not just the SJB; it’s the whole political elite. They’re out of touch, and people know it. This vote was against the establishment that, in the voters’ opinion, messed the country up. The issue for the SJB and Sajith was that they didn’t position themselves as the change candidate.
So the mandate doesn’t mean voters are unhappy with the IMF program?
No, because the NPP was really consistent in saying they wouldn’t break the agreement. There was a lot of scaremongering, like, ‘Oh, if you vote for them, they’ll break the agreement.’ But they were very clear, especially on TV in the past two or three weeks.
AKD came on TV and said they wouldn’t break it. All they were saying is they’d go to the IMF and propose some alternative adjustments. It wasn’t like, ‘We’re going to throw it out.’ It was more like, ‘We’re going to go and talk about it.’
And the SJB’s position was the same. I think most people, on some level, accept that this is necessary. I don’t think this vote was a rejection of the IMF program.
A lot of people who planned to vote for Dissanayake did not expect him to become president, Rannan-Eliya said. Photo: X/@anuradisanayake.
In India, the recent general elections resembled a series of cluster elections held across various states. Was the situation similar in Sri Lanka?
Sri Lanka’s electorate is pretty unified. I mean, there’s a bit of regional differentiation. Like I mentioned before, in some districts, you can expect the CWC bloc, but generally speaking, there are just two geographical areas. You’ve got the North and the northernmost two districts in the Eastern Province, which vote similarly. But the southern district of the Eastern Province is more like the rest of the country.
If you divide the country into those two areas, it’s fairly homogeneous – electorally speaking. And that’s been the case historically.
Actually, in my view, that’s one reason why provincial devolution never made much sense. We don’t have, like India, big regional differences. Even during provincial council elections, when we used to have them, they were basically nationalised.
The vote isn’t really about local issues. There aren’t any strong local champions. People vote as they would in a national election.
Historically, there’s a close correlation between votes in one constituency and those in another part of the country.
So, no, we don’t get big regional differences, except for the North-East and North-South.
And this voting pattern has never changed in any election?
No, not really. Even in our modeling, where we’re working with small samples, we treat it as two separate regions. We have one model for the North and another for the South because that’s what fits the data.
Can you explain your modelling a little?
Our modelling uses something called MRP, which is pretty unusual. It’s really only gained popularity in the UK. In their last general election, six companies used it, and it was originally introduced by YouGov two or three elections ago. Now, in the UK, if you’re a polling company and not doing MRP, no one really cares about your results. So it’s become mainstream there, but not in the rest of Europe.
In India, there’s been only one MRP study, which was done as academic research. The findings were published in a journal probably a couple of years after the last election. They found that MRP results were comparable to the best polls in India, which validated the method.
So, what we do – and this should be common to all good polling – is control for age, sex, ethnicity, etc.
If we have too many women, we adjust for that. If we have too many Tamils, we adjust for that. Same with people from one province. The idea is to make the sample representative.
But the one thing that’s most predictive of your vote in Sri Lanka – and everywhere else – is who you voted for last time.
Standard polls, particularly in the US and Europe, control for past voting behaviour, and we did that too. That worked fine.
But what messed up our model in the past two or three months was something I didn’t notice right away. See, the JVP-NPP got 2-3% of the vote last time. And for two years, the percentage of respondents in our polls who said they voted for the NPP last time was around 1.5-3%. So, we assumed people were telling the truth.
Then, all of a sudden, in the past two or three months, that percentage started going up. It hit 5%, then 6%, then 10% and finally 15% – which is impossible.
At first, I thought it might just be a bad sample for that day, but no, it was a real trend. This wasn’t random noise. It was a shift in people’s behaviour.
That created a huge problem because our model was designed from the start with controls for this. So, when it saw 15% of people saying they voted for the NPP last time, it shrank that down to 3%. It treated those 15 people as if they were only three, which is what the model was supposed to do for consistency. It did this for all candidates.
But the issue was, many of those people who said they voted NPP last time were actually voting for the NPP now. So, in effect, it took away 3-4%, maybe even 5% from the NPP.
It kept getting worse, and at the last minute, we had to do a quick fix. We did the best we could in 24 to 48 hours.
I had to get it out by midnight because of the media reporting guidelines. I remember we only managed to send it out at 10 pm, and we didn’t even have time to translate it into Sinhala or Tamil.
Sajith Premadasa, pictured here, failed to capitalise on his leading the opposition in a country whose economy completely tanked and the government was literally chased out of office, Rannan-Eliya said. Photo: X/@sajithpremadasa.
Your polls have also shown an increase in support for Wickremesinghe over the last few months.
Yeah, that was real. So whichever model we used – old or new – we were definitely picking up on that. He went from around 3-5% to 7%. He definitely saw an increase once he started campaigning.
But the chance of him winning was always, like, zero, to be honest. He was on a bit of a sticky wicket, and he had a very bad hand. Some of those cards he dealt himself, though. So, you know, you can’t really blame anyone else for that.
Given his personal baggage and where he started from three months ago, I think he ran a decent campaign. But he was never going to win. Still, he ran a good enough campaign to get 15% and draw some votes away.
And he was targeting two completely different, totally contradictory groups of people. That was kind of amazing, actually. He was pulling votes from the pro-incumbent government voters – the earlier SLPP [Sri Lanka Podujana Peramuna] voters – we could see that in our data. At the same time, he managed to appeal to the pro-aragalaya voters too.
So, in fact, he was getting votes from the people who voted for the ones that created the mess, and also from those who were against the mess and wanted to throw the government out. It was very interesting targeting. Some of his ads were aimed at that second group.
Tomorrow or next week, you’ll probably hear the SJB saying, ‘Ranil destroyed our chances. Without him, we would’ve got 50%.’ I don’t think that’s true.
Who did the former SLPP voters go for this time?
Ranil or AKD. It was split. I’ve been saying this for two years now. Almost a year and a half, basically.
Basically, what happened was Sajith has been stuck at 30-40% for three years. When the economic crisis hit and the government collapsed, his popularity just went down the drain. That was early 2022, and he didn’t really pick up any votes.
So, as the leader of the opposition in a country where the economy completely tanked and the government was literally chased out of office, he was still unable to capitalise on that.
What happened – and you can see it in our data, because we asked people who they voted for previously – was that all those former Rajapaksa voters ended up going to either Ranil or AKD.
And, as I said, that’s not a recent thing; it’s been happening for at least 18 months. So, Sajith just wasn’t able to make significant headway into that 60% of the electorate.
He could’ve legitimately gotten away with claiming to be an outsider. And so, he had the potential to say, ‘Look, I’m not really establishment’, because his father [Ranasinghe Premadasa] was literally an outsider.
And he [Ranasinghe Premadasa] fought an election when the JVP was literally trying to bring the country down, killing people. So it wasn’t just a campaign against the democratic opposition of the day. It had also been a campaign against the JVP, who were killing people and weren’t even in the election, right? And he was the only guy who could pull that off.
He survived the election, and then he crushed them militarily. So, round one went to Premadasa.
He was the prime minister, but he wasn’t part of the traditional, Westernised, Anglophile Colombo elite. He was pro-market, but he also believed there had to be something for the poor. He had a lot of social programs. He introduced an income transfer for all people, and it’s still there long after, just under a different name. And poor people liked him.
The problem with Sajith and the UNP [United National Party] is they went back to that neoliberal track they’ve been on for 50 years, which doesn’t recognise the need for that balance.
Sajith could’ve revived that legacy – he had the brand, and he occasionally talks about it, but I don’t think he fully understood it.
I don’t think they [the SJB] fully grasped the depth of public rejection of the system this time. It’s not enough to just compete against Ranil as ‘Ranil-lite’.
What do you think drove voters toward the NPP this time?
This isn’t about the IMF – it’s about people’s rejection of the establishment. People have finally lost patience with an establishment that doesn’t hear them. I think disgust is the correct word.
But whatever, people were fed up with the whole political establishment. A rejection of the status quo and a desire for change. A lot of people just gambled on the NPP. Who knows how it’ll turn out?
The Westernised social elite runs across all parties. Honestly, there’s even a bit of it in the NPP now, because a lot of people in the NPP aren’t Marxists – they’re just urban, upper-middle-class people.
I know lots of people who voted for the NPP – lots of young doctors. It’s been happening over the past year, but they’re not Marxist.
The voters [who voted for the NPP this time] are not loyalists. They could easily vote for someone else next time. Because 99% of voters hadn’t voted for them last time.
I’d say the NPP might be better understood as one of those modern Western centre-left parties, like the Democrats, Labour or the SPD in Germany [Social Democratic Party of Germany], because if you look at their constituency, it’s not poor people anymore. In our polling data, Sajith actually did better among poor people. So these are middle-income, upper-middle-income, educated, often professional, urban Sinhalese.
AKD has rural support – he’d need that to win – but he did best among this urban middle-income Sinhala voter.
