Voters who dislike both candidates – who have been dubbed double haters – were especially likely to defect from Trump. Overall, Trump lost more than one-fifth of the double haters who once backed him. That group of defectors was about evenly split between moving to Biden and saying they were now undecided.

Politically disengaged voters, an area of growth for Trump in recent polls, are especially likely to shift from Trump to Biden.

With five months to go until the election, there’s still plenty of time for Trump to regain his standing. The verdict is still fresh in the minds of voters, and shifts in public opinion following a major news event can prove fleeting. The study offers no reason to assume that Trump has lost these voters for good, and many still haven’t made up their minds about the verdict.

A plurality of those we called back for the study approved of the verdict, but a sizable share said they had not heard enough to say whether they approved or disapproved of the outcome. More than a quarter said they’d paid little or no attention to Trump’s legal battle.

The findings depict an unsettled electorate, one with many disengaged voters who might swing over the months ahead. Overall, 8 per cent of respondents offered a different response in the presidential race than they had when they were first interviewed no more than eight weeks ago – a tally far higher than many might imagine in today’s polarised country. Even Biden retained only 96 per cent of his former supporters, with 1.5 per cent of those former supporters saying they would now back Trump, despite the news of his conviction, and the rest moving to undecided.

Contacting previous respondents may be an excellent way to track how people’s views change over time, but it’s not necessarily the best way to represent the whole electorate. On the one hand, Biden’s supporters were slightly likelier to retake the survey than those who backed Trump, 37 per cent to 35 per cent. The voters we reached again were generally older, more educated, more highly engaged and more likely to be white than those who did not respond.

On the other hand, these demographic groups were also more likely to stick by Trump than the younger, less educated, less engaged and non-white voters who were less likely to retake the survey.

In fact, the voters we spoke to who continue to support Trump appear to be more enthusiastic than ever. Many of his previously disengaged supporters seemed newly energised by the verdict, with 18 per cent of his supporters who previously said they were unlikely to vote now “almost certain” to do so, compared with just 3 per cent of Biden’s supporters who moved into that category.

But the slight movement overall towards Biden is broadly in line with other recent surveys. In a Reuters/Ipsos survey taken immediately after the verdict, 10 per cent of Republicans said Trump’s conviction made them less likely to support him in the November.

Another recontacting study by Echelon Insights, a Republican firm, found Biden gaining 2 points compared with its previous survey.

In a Times/Siena poll of six battleground states conducted in November, about 7 per cent of Trump’s supporters said they would switch their support to Biden if Trump were to be convicted and sentenced to jail in an unspecified criminal trial. Other pre-verdict polls asking specifically about the Manhattan hush money trial found a similar share of Trump’s supporters nationally who said they intended to switch their support if there were a guilty verdict.

Here are the key things to know about how this study was conducted:

— On June 3-4, we reached 1897 registered voters who had taken a Times/Siena poll in the previous two months.

— The study was conducted by telephone using live interviewers. More than 93 per cent of those we spoke to were reached on a mobile phone.

— Respondents were called if they took part in the Times/Siena surveys of six presidential battleground states in May or of the nation in April.

— While recontacting studies can help answer important questions of whether individuals are changing their minds, this study is not necessarily representative of the entire electorate. It is not possible to calculate a conventional margin of sampling error. And while all surveys have sources of error beyond sampling, such as non-response bias, this study in particular may be more likely than the typical Times/Siena poll to over-represent the most politically engaged voters.

This article originally appeared in The New York Times.