Beyond the Ballot Box

Leveraging MRP models for real-world impact - a nuclear example.

My team and I have spent 2 years building and refining our political Multi-level Regression with Poststratification (MRP) model, but it’s not just because we love tracking elections. Elections are one of the few areas where our model projections can be quickly verified, making them ideal testing grounds. They allow us to refine the model’s ability to capture diverse demographic and geographic nuances. But they let us refine and adapt our approach in ways that go far beyond politics. We can transfer the learnings and models to new, equally complex topics tied to the same values and population patterns. And, not sure if I mentioned, ours is pretty great.

MRP’s unique advantage is in its ability to translate small, often biased samples into meaningful geographic insights. We are now able to break down reliably even to the ward level. There are roughly 8,000 wards in Great Britain, which means a standard nat-rep survey will get an average of a quarter of a person per ward – not quite enough to draw any meaningful conclusions without modelling. This precision helps us track shifting battlegrounds in real-time, both geographic and ideological. By capturing the essence of these shifts, MRP lets us see where issues resonate most strongly and what is influencing public opinion.

The Energy Transition: Nuclear MRP in Action

We have built over 20 GB models that track key issues for the electorate and our clients, from Labour unions to the ‘Nanny State’ to defence spending. A key megatrend our team is tracking and building models for is the energy transition.  One important part of this is nuclear power, and we often poll on public support here:

Support for nuclear power plants has gone from being a middling support ‘wedge issue’ - splitting the political ‘left’ and the ‘right’ - to having resounding net support unified across all the major political parties. Lower levels of support from Scottish National Party (SNP) voters indicates that there are some regional differences, and our MRP allows us to break this down further.

Average net support is high, but there are significant regional fluctuations. This is incredibly useful when designing and running national or regional campaigns, especially when broken down further to ward level.

Knowing that nuclear support is rising and where is useful, but understanding why is even more valuable. Even though the major parties are aligned in backing Nuclear, we have found that different societal subgroups back nuclear energy for different reasons: some are more focused on lowering energy costs, others on job creation, carbon/emissions reduction, or even just proving that the UK isn’t paralysed by planning restrictions. MRP lets us see where these motivations overlap, allowing projects to align naturally with public sentiment.

When campaigns align with genuine public concerns in this way, they build a foundation of trust that can strengthen their long-term impact. Data-backed messaging resonates because it speaks to what matters most to people, and MRP helps us identify those unique concerns. Clients can create campaigns that aren’t just persuasive but that foster trust and loyalty, reaching beyond immediate goals to create lasting public buy-in.

In addition, whether politicians and policymakers will take notice of increased nuclear support will be heavily dependent on politics and economics. In the few months since the election – we have seen a dramatic shift in voting intention, with Labour’s vote share not far off the Conservatives again.

There is a long time until the next election, but over the next few years, Labour will be looking for ways to bring back their “hero voters”. Linking together our MRPs allows us to see the relationship between Labour holding their seat and support for issues we care about, in this case, Nuclear.

Simply put, the more at-risk a Labour seat is, the stronger the support tends to be for nuclear energy—and the more sceptical people tend to be about the economic benefits of tackling climate change. So, if Labour wants to win back these constituencies, building nuclear is likely to be popular, and the best framing may not be about tackling climate change. Our latest MRP model lets us test different ways of framing these issues, helping us find the messaging that best aligns with each area’s priorities.

Based on our latest MRP update, we would classify about 157 Labour constituencies as “at-risk”. These “hero voter” constituencies are essential battlegrounds, and their opinions will be weighted highly over the next few years. They often differ from traditional party lines and are deeply linked to issues like energy and the economy.

Using MRP beyond elections gives us a robust tool for understanding societal currents. For instance, in the context of the energy transition, we don’t just measure public support; we offer a breakdown of what drives it. Whether a company is developing an energy project or a policy group is advocating for new initiatives, these insights let them tailor their approach to align with public sentiment on nuanced issues, ultimately making a stronger case for their efforts. With MRP, clients gain a fuller picture of what matters to people and how to frame complex issues in a way that resonates, helping them navigate and influence the shifting landscape of public priorities.