Some people who voted for Donald Trump in 2024 now say they didn’t, a shift that may be causing polls to understate Democratic support, according to pollster G. Elliott Morris.
Morris says many pollsters adjust their samples so the portion reporting a 2024 Trump vote lines up with the real result. The trouble comes if disillusioned Trump voters stop admitting it. Then those who still say they backed him are more devoted than his actual coalition, and bumping them up to match his true share would exaggerate how many hard-core supporters there are.
In September, Morris used simulations to estimate this “recalled vote bias” was trimming Democrats’ lead on the generic ballot by roughly 0.8 points. His October Strength In Numbers/Verasight poll let him test it directly. Verasight keeps each panelist’s 2024 vote from when they joined, so he compared those answers to what 2,028 respondents said now.
Most answered the same way. But about 3 percent of people listed as Trump voters now say they backed Harris, which is over double the rate of Harris voters switching to Trump. Overall, 6 percent of Trump voters changed their story, 3.2 percent of them to Harris. Those switchers mostly disapprove of Trump and almost all plan to vote Democratic.
The groups are tiny, 22 people moving one way and 11 the other, so uncertainty is large. Morris says the pattern held across each cut he looked at, including weighted and unweighted results, registered voters, and demographic groups.
Ultimately, he argued, this could be biasing polls slightly against Democrats — and election results could end up coming in better for them than recent surveys are predicting.
This comes as polls already show danger numbers for Republicans, with voters backing Democrats on economic issues. It also comes amid a shocking development with one recent upstart pollster admitting to have been faking its numbers.