UK MPs Saw a Climate Backlash That Polling Showed Did Not Exist

For a policy to help homes become more energy efficient, MPs pictured a country with far more opposition than an Ipsos poll found. For a proposal to vary product prices according to environmental damage, their picture was darker still. The difference was not simply whether a policy seemed popular. It was whether it seemed dangerous to say aloud that it should happen.

In an online survey experiment, 100 sitting UK MPs were asked to estimate public support for climate policies, then to say how willing they would be to defend two of them in a heated television debate or promote them among parliamentary colleagues. Their estimates were compared with nationally representative polling of 4,201 adults conducted earlier that year. The study, led by Lisa-Maria Tanase at the University of Cambridge’s El-Erian Institute of Behavioural Economics and Policy, found a Parliament that, on average, saw less appetite for climate action and more political division than the polling did.

The Opposition MPs Expected

The first two policies were a scheme offering grants or loans for households with lower energy-efficiency ratings to improve their homes, and a measure to make product prices reflect their environmental damage. Before any intervention, MPs underestimated public support for the efficiency policy by 7.1 percentage points and for the pricing policy by 14.9 points. Across both, the average shortfall was 10.97 points. Seventy-eight of the 100 MPs underestimated support overall, while 61 underestimated it by more than 10 points.

The MPs didn’t merely mistake support for indifference. They also overestimated opposition. In the paper’s discussion, the authors argue that this distinction matters because a politician who sees voters as actively hostile has more reason to avoid a policy than one who sees them as undecided. The result echoes research across four countries in which elected representatives made large errors when estimating public preferences and were only modestly more accurate than citizens themselves. That comparative study examined 866 politicians and more than 10,000 estimates of policy support.

Cost did not remove the gap. MPs were told to estimate support after respondents in the public survey had been made aware of personal inconvenience or financial cost. Under those descriptions, MPs still underestimated support for energy efficiency by 15.41 points and for environmental pricing by 21.68 points. They did correctly rank the policies: the energy-efficiency subsidy was more popular than the tax-based proposals. But they expected public support to fall much farther when costs entered the picture than the polling indicated.

A Wider Divide Than the Poll Found

The perceived electorate was also more sharply divided by party than the measured one. MPs underestimated support among Conservative voters more than among Labour voters, Liberal Democrat voters or the public overall. In the press release’s policy-level breakdown, MPs put the Labour-Conservative gap at about 25 percentage points, while the Ipsos figures placed it at 12 points. They saw Conservative voters as 12.6 points less supportive than the public overall, rather than the 2.5-point difference in the polling.

Tanase calls the result a “funhouse mirror” effect, in which politicians encounter a distorted version of public views on climate policy. “Politicians, like all of us, are often exposed to information environments that distort perceptions of public opinion,” she said. “Social and traditional media can disproportionately reflect particularly vocal segments of society rather than the full distribution of opinion.”

This is not evidence that every disagreement over climate policy is imagined. Specific policies produce different coalitions, and political representatives can themselves be more divided than voters on some climate questions. A recent study using matched surveys of British MPs and voters found limited overall differences on broad climate attitudes, but found the public more supportive of some specific mitigation measures and less polarized over them. The distinction between broad goals and concrete policy choices is central to interpreting the new study.

Within the new sample, personal opinion mattered too. MPs who were less personally supportive of a policy tended to make larger underestimates of public support. Yet the relationship was not simple projection. Even MPs who supported the policies still underestimated backing on average. Nor did party affiliation significantly predict the average baseline error, although the authors caution that the small Conservative and Liberal Democrat groups limited the ability to detect party-level differences.

Silence Attached to the Tax

The study’s political mechanism appeared only for the more contentious policy. Among the 50 MPs assigned to the control group, lower perceived public support for environmentally informed product pricing predicted a lower stated willingness to speak in favor of it, even after the researchers accounted for the MP’s own opinion. The same was true for comfort promoting the policy to parliamentary colleagues when MPs believed fewer colleagues supported it.

The researchers call this pattern consistent with a spiral of silence: expecting little support makes someone less likely to voice a position, leaving that position less visible to others. But it was not detected for the home-energy-efficiency policy. That negative result is informative. Subsidies are generally expected to be less controversial than taxes, and the study suggests that perceived opposition may matter most when a policy already looks electorally exposed.

“MPs across parties who both support and oppose climate policies consistently overestimate public opposition,” Tanase said. “This is likely to limit the range of what they consider to be politically feasible options for taking action on climate change.”

The analyses measured intended advocacy in hypothetical scenarios, not parliamentary speeches, voting records or legislation. The authors therefore cannot show that misperception has already blocked a particular climate measure. Nor can a survey capture the full pressure of party discipline, constituency politics, media framing or reputational risk. Still, the association identifies a route by which a distorted impression of opinion could narrow the policies MPs consider safe to champion.

One Dose of Polling Was Not Enough

Half the MPs were then shown polling information for the first two policies, including overall support, opposition, party-voter estimates and the effect of making costs explicit. They subsequently estimated support for two different policies: a frequent-flyer levy and higher taxes on red meat and dairy. The design tested whether information could reduce a broader tendency to underestimate support rather than whether it could correct answers to the exact same question.

On the study’s preregistered frequentist test, the intervention group improved by 1.22 percentage points relative to the control group, a difference that was not statistically significant. A Bayesian analysis, informed by a prior pilot in members of the public, produced a larger estimated difference of 3.6 points and a 94.2% posterior probability that the effect was above zero. Its 95% credible interval, however, ran from -0.8 to 7.8 points. The data were compatible with a modest benefit, but also with no benefit.

That uncertainty resembles a wider problem for interventions aimed at elected officials. In two US field experiments, more than 2,300 legislators were invited to use dashboards of constituency opinion; few accessed them, and those who did were no more accurate afterward. Those experiments do not test climate policy or the same kind of correction, but they show why simple access to opinion data may not readily change elite beliefs.

The Cambridge study is limited by its size and timing. Its 100 MPs included 74 Labour MPs, 16 Conservatives, four Liberal Democrats and six MPs from other parties, an approximate rather than fully representative party distribution. Polling was collected in April 2024, while the MP survey ran from December 2024 to January 2025, although the authors cite tracker evidence suggesting broad net-zero attitudes were relatively stable across that period. The paper also relies on stated survey support, which may not predict behavior once a policy is implemented.

What the study puts in view is not a hidden consensus that can make every climate policy politically effortless. It is a more precise problem. MPs could tell that subsidies drew more support than taxes, yet still systematically underestimated the support that was there, especially when they imagined people paying a cost. If politicians act on those errors, the political debate can contract before voters have had the chance to reject, accept or reshape the policies being kept from it.

FAQ