The 2026 Senedd Election Simulator: One Year to Go

This time next year, Wales holds its seventh devolved election – one that will be different from all previous contests. This election will be the first to use a new proportional electoral system (closed party list), the first to elect 96 Members rather than the current 60, and the first to use the new 16 multi-member constituencies. And that’s before we even start with modelling the general ups and downs for the political parties since the last set of elections in 2021.

The previous system, known as the ‘Additional Member System’, where voters cast one ballot for a constituency member and one ballot for a party standing on a regional list, has been scrapped. And although it was used for 25 years, it was perennially misunderstood by voters and politicians alike.

We now have a chance to reset the clock on public understanding of Welsh democracy. That’s why I’ve created a Senedd Election Simulator (https://jaclarner.github.io/senedd_etholiad_sim/)  that will show how votes might translate into seats in our new Party List system. Through it, I hope that citizens, journalists, students, and politicians alike can engage with our new electoral system right from the get-go.

What Does the Simulator Do?

The simulator allows you to adjust national vote shares for each political party and immediately see how these changes would affect seat distribution. It visualizes results using:

• A hemicycle showing seat allocations in the Siambr
• Comparative charts of vote share versus seat share
• A dynamic allocation of six seats in all 16 constituencies under the D’Hondt formula
• Analysis of possible government coalitions
• Statistical measures of how proportional a given result would be

When you enter new vote percentages, the simulator calculates results across all Welsh constituencies and shows which parties would gain or lose seats compared to the baseline. Users can also select from a number a ‘preset’ vote shares based on historic results and current polling.

Important Caveats and Limitations

As an educational tool rather than a prediction model, the simulator has several limitations you should keep in mind:

Different Swing Models Produce Different Results

The simulator presents TWO different calculation methods:

1. Uniform National Swing: ‘UNS’ applies the same percentage point change to all constituencies, uniformly for each political party. So, if Plaid Cymru were to see a five percentage point increase Wales-wide, their vote share would increase by five percentage points in all 16 constituencies. This is simple but can produce unrealistic results with high volatility – for example, it might show Labour dropping to 0% in some areas given current polling swings — that’s of course highly unlikely to happen in reality.
2. Proportional Swing: Unlike UNS, proportional swing varies predicted vote share proportionally and relative to a party’s starting position in each constituency. This can sometimes deal with the challenges mentioned above, but it tends to overestimate the likely vote share for new parties (such as Reform UK) and struggles with large swings.

Deliberate Model Transparency: I’ve intentionally chosen not to artificially “fix” the known issues with the swing models. While it would be possible to add arbitrary caps or floors to prevent unrealistic results (like Labour dropping to 0% in a constituency), such adjustments would either require complex modelling or introduce hidden “stops” based on subjective judgments. Who’s to say whether an artificial floor for Labour should be 0%, 2%, 5%, or 10% in a given constituency? Instead, I’ve kept the models mathematically pure (with one exception – see below), allowing users to see their genuine limitations.

Baseline Data Limitations: The starting vote shares are estimates based on the 2021 Senedd election adapted to new constituency boundaries. It’s important to understand that these are inherently uncertain – they are not definitive descriptions of how 2021 results would have looked under the new system. We can’t because we used a totally different electoral system in 2021! Even small changes to these baseline estimates can significantly affect projected seat outcomes, especially in closely contested constituencies. I plan to make further adjustments to these estimates and will make all code and methodologies publicly available on my GitHub repository for full transparency, and to invite feedback that might improve the baseline data.

Reform UK Baseline: Calculating a baseline for Reform UK is particularly tricky for obvious reasons. To get as close as possible, this model combines estimated 2021 Reform UK and Abolish the Welsh Assembly Party support, because Welsh Election Study data suggests an almost complete voter transfer between these two parties. This isn’t perfect, but it’s the best assumption we have for estimating the voter base that Reform UK are starting from at the 2026 election.

Other Simplifications:

• The models don’t account for tactical voting
• They assume consistent turnout across constituencies
• They don’t factor in local candidates or constituency-specific issues
• They don’t model demographic changes or differential turnout

Future Model Updates

I’m actively developing several enhancements to make the simulator more robust and insightful in the run-up to the 2026 elections.

Regional Swing Model Implementation: The first priority will be to add an regional variation option which will allow for different swings in North, Mid, South-West, and South-East Wales. This will better capture how parties perform differently across Wales’s diverse regions.

Refined Baselines: I’m working with Welsh Election Study colleagues to improve our baseline estimates using more sophisticated demographic modelling, which should provide more accurate starting points.

Enhanced Coalition Analysis: Include coalition formation models based on political science theories, including ideological compatibility measures and historical precedents in Welsh politics.

Uncertainty Visualization: Better ways to display uncertainty in results, including confidence intervals and sensitivity analysis that shows how small vote shifts could change outcomes.

Mobile UI Improvements: Making the simulator work seamlessly on smaller screens via redesigned interfaces for smartphones and tablets.

Educational Resources: I plan to add more explanatory content about Wales’s new electoral system for secondary school students.

As Wales approaches this historic electoral transition, the Senedd Election Simulator aims to democratise understanding of our new voting system. While no model can perfectly predict the complex interplay of political forces that will shape the 2026 election, this tool provides a valuable starting point to engage meaningfully with Wales’s evolving democratic processes. And importantly, I want to be transparent in showing both the possibilities and the limitations of elections modelling rather than hiding the forecasts behind a veneer of certainty.

Your feedback and engagement will be invaluable as we continue refining this simulator in the months leading up to this landmark election. Let’s make next May’s election the best yet in public understanding and engagement with Wales’ democracy.