In April 2018, ahead of the Karnataka assembly elections, Bharatiya Janata Party national president Amit Shah improvised on a line that he has used variations of for the last four years. “The Modi wave has now turned into a tsunami,” he said. Just one week later, as his party fell short of a majority and was then out-manoeuvred at alliance-building by the Congress, it was tempting to get into wordplay with the wave metaphor: was it now a trickle? Or was it just the low tide before the high tide?

The “Modi wave” has been a mainstay of political writing since 2013. A wave election is a loose phrase for an election in which one party wins big. But a wave associated with a particular leader is an election, or a phase of electioneering, in which a leader’s popularity outstrips even that of his or her party. It becomes the force behind the party being propelled into the lead in areas where it was otherwise not expected to do as well, said psephologist and social scientist Sanjay Kumar, director of the Centre for the Study of Developing Societies.

Official election statististics don’t tell us about the popularity of a leader – they only tell us how many people voted for each party. For voter preferences about leaders, and for demographic details about voters, researchers turn to surveys. The most respected of these, with a 50-year history of polling voters before and after elections, is the National Election Study conducted by the Lokniti programme at Centre for the Study of Developing Societies on a nationally representative sample.

In both the 2009 and 2014 rounds of the National Election Study conducted right after voting had ended, more than half of the respondents said that the party mattered more than the candidate in their voting choice. But of all possible considerations in favour of the party of choice, the top reason was “good leadership”, cited by 18%-20% of respondents. This shows Indian voters might vote for the party rather than the person – but the person heading it matters the most to them.

In the 2014 election, the evidence for a presidential streak in India’s Westminsterian system was even more direct: the prime ministerial candidate mattered more than the local candidate, voters told Lokniti surveyors.

Twenty eight percent of people said they had voted for the BJP, but 36% said they wanted Narendra Modi as prime minister. On the other hand, 18% said they had voted for the Congress, but only 14% said they wanted to see Rahul Gandhi as prime minister, with an additional 5% choosing Sonia Gandhi and Manmohan Singh.

On key issues in 2014 – tackling corruption, price rise, unemployment, terrorism – the Congress trailed the BJP by 10 percentage points to 12 percentage points in voter confidence, but on the same issues, Gandhi trailed Modi by 15 percentage points to 17 percentage points. If Modi was not the BJP’s prime ministerial candidate, a full 20% of respondents said they would have voted for another party.

In short, there was a Modi wave in 2014.

What happened after 2014?

All evidence indicates the wave of support for Modi continued through the subsequent assembly elections. His name began to figure as the preferred chief minister in unlikely states. In Jharkhand in December 2014, a post-poll survey done by Lokniti found he was the fourth most popular choice for chief minister – more popular than Jharkhand Mukti Morcha leader Shibu Soren. In Bihar’s 2015 election, over a third of respondents in a post-poll survey voted Modi as better for Bihar’s development than Nitish Kumar. On tackling unemployment and uniting castes, Modi came even closer to Kumar’s numbers.

In Uttar Pradesh’s 2017 election, Lokniti found in a pre-poll survey that Modi ranked third in a list of chief ministerial preferences – higher than the eventual choice, Adityanath. Respondents were asked what mattered to them most while choosing whom to vote for, and after the usual answers (party, chief ministerial candidate, local candidate), another 10% offered a fourth option – Modi.

Even in the southern states where the BJP’s organisation structure has failed to take root, 95% of a sample polled for the Pew Global Attitudes Survey in 2017 viewed Modi favourably.

So is the Modi wave still surging in 2018?

The answer is yes and no.

A pre-poll survey conducted by Lokniti in May this year showed that anti-incumbency is creeping in. There are now more people dissatisfied than satisfied with the ruling party’s performance on all fronts, with the lack of jobs being identified as the biggest problem.

But to tackle most problems – except the GST, women’s issues and farmers’ issues – voters are still leaning towards the BJP. They seem to be saying that they are not happy with the job the ruling party is doing, but it is not clear someone else could do it better. Thirty two percent of the respondents said they would vote for the BJP as against 25% for the Congress.

These are not bad numbers for the ruling party. Support for the BJP is nearly the same as what Lokniti’s last pre-poll survey before the 2014 election found. Much of this could be because preference for Modi, after rising post-2014, is down by just two percentage points to a still strong 34%.

But the difference is that Gandhi’s numbers are now up to 24%. Modi and Gandhi are rated as being equally likeable with Gandhi having slightly higher net likability. For both, around a quarter of supporters are recent converts (higher for Gandhi) and both have some recent detractors (more for Modi).

What does this mean for the 2019 election?

Lokniti has interpreted the May 2018 numbers as “indicative of a declining trend, one that the BJP has been unable to stem”. When a party loses favour with people, the leader’s personal popularity alone cannot pull it through. In 2004, the BJP’s then PM candidate Atal Bihari Vajpayee was the leader of choice for 38% of respondents, even more than for Modi in 2014, Lokniti data shows. In comparison, the preference for the BJP was a good 15 percentage points lower. The BJP lost that election.

However, BJP MP and national spokesperson GVL Narasimha Rao, a former psephologist himself, is sceptical about the Lokniti data, pointing out, correctly, that the organisation got several elections including the Uttar Pradesh outcome wrong. “We do not agree that there is anti-incumbency. It is just Congress propaganda,” he said dismissively. “We believe the Prime Minister’s personal charisma has carried the BJP into areas that it didn’t have a presence for decades.”

Much will depend on how much Modi’s personal popularity is able to override voter unhappiness with his party. In the three BJP-held states that will soon vote in assembly elections, dissatisfaction with the party is high, said Yashwant Deshmukh, a psephologist whose firm CVoter has tracked these numbers every week.

Do you want to change state government immediately?

Yes 67.2% 55.3% 57.1%
No 30.8% 43.2% 40.1%
Can"t say 2.1% 1.5% 2.7%
Source: CVoter

Do you want to change chief minister immediately?

Yes 65.6% 51.7% 55.4%
No 33.2% 47.2% 42.3%
Can;t say 1.2% 1.1% 2.2%
Source: CVoter

But blame is directed at the state leadership and not the central leadership, where Modi can still do little wrong.

Do you want to change Central government immediately?

Yes 41.9% 42.6% 37.8%
No 55.4% 55.1% 59.1%
Can't say 2.7% 2.4% 3%
Source: CVoter

Do you want to change prime minister immediately?

Yes 38.4% 40.9% 36.8%
No 59.2% 57.7% 61.2%
Can"t say 2.4% 1.4% 2%
Source: CVoter

“The majority of people in these states want a change of CM but not a change of PM,” Deshmukh said. The latest CVoter data still shows Modi polling twice as high as Gandhi. It is conceivable that the BJP will lose these states in 2018 but still win the bulk of their parliamentary seats in 2019.

There is a year between the May 2018 Lokniti survey numbers and the next election, and nearly everything will depend on the election campaign the two parties mount. Will Modi’s still strong personal popularity push the BJP to victory again? Possibly, but his ability to create another Modi wave, in the face of rising dissatisfaction with his party, looks unlikely.

Visualisations by Anand Katakam.

Read all stories in the How India Votes series here.