Methodology

Our Objectives

  1. Get the Tories out!
  2. Help voters influence the next government
    Not just think tanks & press barons
  3. Support a diversity of voices in parliament
  4. Highlight that tactical voting wouldn't be needed if we had a fair voting system

Our Methodology

1 - We collate data from various sources to calculate our recommendations:

  • Parliamentary boundaries have changed since 2019, so we cannot use the actual 2019 constituency results. Instead we use the implied 2019 results for new boundaries based on Sky, BBC etc research.
  • We take a simple average of MRPs run in the last 6 months. You can see the exact polls we are currently using on our data page.
  • We look at any lists parties publish about the seats they are targetting (or seats they are not targetting). You can see the exact data sources we are using for this on our data page

2 - Using this data, we automatically judge each every seat:

  • Could the Tories realistically win this seat?

    If the Tory party was significantly behind the top opposition party in the 2019 implied results, and the Tory party are significantly behind the top opposition in the average of recent MRPs, and the Tory party has a low vote share in the average of recent MRPs, then we assume this seat cannot be realistically won by the Tory party.

    If any of the above are not true, we assume the Tory party could potentially win the seat.

  • Is there a clear lead opposition party in this seat?

    We look for the progressive party with the biggest implied vote share in 2019, the progressive party with the biggest vote share in the average of recent MRPs, and whether this is a target (or non-target) seat for any of the progressive parties.

    If all of the above agree, then there is a clear lead opposition party in the seat. If any of the above do not agree, then we err on the side of uncertainty.

3 - Having automatically judged every seat, we then publish advice using the following flow chart:

Flow chart showing the advice we will publish in different seats based on category of seat

4 - In seats where automatic advice cannot be given, or where we have specific local intelligence, we may manually override the advice:

  • We always make it clear on the website and in our published data when we have made a manual override.
  • We always cite the reasons for manually overriding advice.
  • When making a decision about whether to manually override, and the advice to give, we will look at various other data, including but not limited to:
    • Recent local election results in the parliamentary seat in question
    • Any recent parliamentary byelection in the seat in question
    • Historical voting patterns in the seat in question

Methodology FAQ

Westminster constituency boundaries were reviewed in 2023, and therefore many constituencies now have new geographic boundaries. Some constituencies have been abolished entirely and some new constituencies have been created.

The 2019 general election was fought using the old constituencies and boundaries, and these results cannot be easily mapped onto the new constituency boundaries. However, Sky & the BBC commissioned research which estimates the vote share each party would have achieved in 2019 using the new constituency boundaries. We use these implied 2019 results in our methodology (see above).

Byelections tend to have low turnout compared to general elections, and parties may be able to dedicate more resources to fighting a byelection campaign, because they&re only fighting byelections in a small number of constituencies at any one time.

Byelections will still also be using the old constituency boundaries, which means byelection results can be difficult to map directly onto the constituency boundaries which the 2024 election will be fought on.

In constituencies where we manually override our automated advice, we will consider any relevant byelection results.

Council elections tend to have low turnout compared to general elections, and the issues they are fought on can be significantly different from those which are the focus in general elections.

In constituencies where we manually override our automated advice, we will consider any relevant council election results.

We use the same methodology - described in detail above - on all seats in England, Scotland & Wales. In some cases, this methodology cannot give a tactical vote recommendation, because it's not clear which progressive party is best placed to stop the tories.

In these constituencies, we will manually make a recommendation. This may happen close to the election day, so that we can consider the most up-to-date information, and we can make the most informed & accurate recommendation possible. When making recommendations manually, we consider the data used by our automated methodology - implied 2019 results, recent MRPs, target seat lists - as well as other data such as westminster byelections, local council elections in the area, etc.

There may be other situations where we need to manually override our automated methodology, such as if the candidate for the recommended party stands aside.

When manually making a recommendation, we always make this clear on the website and in our open data, and we cite the data we have used to come to our recommendation.

For questions not related to our methodology, please see our main FAQ page.

@MVTFWD