How are the results calculated?
Why are the results different to another tactical voting site I saw?
Why should I trust your recommendations?
Are you trying to stop Brexit, or just an anti-Tory site?
But Labour supports Brexit!
What about the European elections?
What about polls?
What about historical general election results?
What about Leave/Remain votes in the EU referendum?
What about the surge of local support for such-and-such candidate?
What about a formula with all of the above?
I don't want to vote for the party you've recommended
The Tories don't stand in my constituency
I think you've got it wrong!
Who runs this site? Who funds it?
Can I help?
The basic method is as follows:
We may make specific, manual adjustments for other unusual seats, such as seats where there are independent candidates or electoral pacts. We have also switched to 'Not Sure' in some cases where other tactical voting sites were clashing with our advice, to try to increase the overall level of consensus. These reasons will always be stated clearly on each page.
Have a look through the constituency A-Z to see the results in different kinds of constituencies.
There are many different ways that you could potentially work out the 'best' tactical vote in each constituency, and many different factors to take into account. Each list and site circulating uses a slightly different method, but we believe ours meets these key criteria:
What's important here is that we didn't decide these results seat by seat – the same method is applied across the board. We apply any changes to the method to all constituencies, not one rule for some and a different rule for others.
We believe it is irresponsible that some sites are recommending supposedly 'tactical votes' for candidates who were in a distant third or even fourth place in 2017.
People have lots of questions and comments about our method, but it is the same method we used in 2017, when we were the most accurate tactical voting site.
This question is based on the idea that these are different aims that would lead to different recommendations, but really they are one and the same.
The most important thing in this election is to keep the Tories out in your constituency. That means voting for whichever non-Tory party has the best chance of winning. Depriving the Tories of a majority in this way is the most likely path to stopping Brexit (via a second referendum). Voting for a 'more Remain' candidate who cannot win the seat will do nothing at all to help stop Brexit, and so is not a tactical vote.
While Labour is not a 'full Remain' party, it is offering a second referendum on Brexit. Its manifesto says: "Labour will give the people the final say on Brexit", and one of the options on the ballot paper would be "the choice of remaining in the EU". This second referendum is now, in reality, the most likely route to stopping Brexit.
If you refuse to vote Labour in seats where they are the only party that can keep out Boris Johnson's hard-Brexit Tories, that's your decision – but it is precisely the opposite of tactical voting.
The European elections are a completely different voting system, proportional representation. This is the case even if you narrow down results to one general election constituency – people were voting in a proportional system and voted for parties that would not be able to win in a first-past-the-post general election seat.
If we took the European elections as a guide, then we would be expecting the Brexit Party (who won the European elections with 30% of the vote) to win the general election. Hopefully you'll agree that is a very unlikely outcome.
We do not base our recommendations on polling. We understand why people send us polls, especially particularly striking ones – but a striking poll is even more likely to be an outlier. Polls fluctuate all the time and were a notoriously unreliable guide in 2017, which was considered a 'shock result' mostly because the majority of polls called it wrongly.
The group Best for Britain is running a tactical voting site based on polling using the MRP method, which it says is highly reliable. But this claim is based on a few cherry-picked examples of success (eg. particular seats that the method predicted in 2017, without stating which ones it failed to predict).
For example, Anthony Masters of the Royal Statistical Society writes: "A curious case of selective memory has surrounded the use of MRP. In the 2017 General Election, people seem to recall YouGov’s accurate central estimation of a hung parliament. Using the same technique, the Lord Ashcroft model (which estimated a Conservative majority over 60) is sometimes forgotten. Just as surveys can have errors, models can too. Large sample sizes and MRP are not guarantors of reliability."
Some people argue that whatever the methodology, at least the data is more recent, but MRP is extremely sensitive to the assumptions and demographic factors build into its model – and Best for Britain are refusing to publish any of that information: their recommendations come from a 'black box' model that you're being asked to take on trust. They are making many claims for their accuracy and track record without acknowledging the error and uncertainty that are inherent in polling.
We do not claim to be experts on polls – another reason for us not to use them for recommendations! – but we believe that any claims by anyone that they have a super-accurate polling method should be treated with a healthy skepticism.
The common theme in messages of this sort is that they want us to look at 2010 and make a Lib Dem recommendation – but 2010 was a high water mark for the Lib Dems that they have not come close to since. The Lib Dems picked up 23% of the vote in 2010 – the biggest vote the party had gained in any general election since its formation (as the successor to the SDP-Liberal Alliance) in 1988. They then lost much of this vote by forming the 2010 Tory-Lib Dem coalition.
There is no evidence that the 2010 vote represents a historic 'base' that will return to the Lib Dems. We were sent the very same arguments at length in 2017, but the Lib Dems' 2017 result (7.4%) was nowhere near their 2010 vote, and in fact slightly fell compared to 2015 (7.9%).
The referendum was held before the 2017 general election, so the Remain-Leave affiliations are already well represented in that data.
We're not going to change recommendations based on anecdotes from campaigners about how well a particular candidate is doing or not doing – we believe our recommendations need to be based on hard data about actual votes, not speculation. While we understand that people want to offer their local experience, accepting it would open us up to lobbying from local party campaigners as well as arbitrary decision-making.
It's tempting to imagine that some kind of 'super-formula' with as many data points as possible would give more accurate recommendations – but there is no evidence for this. Including – for example – the previous five general elections would most likely make your results less accurate, not more accurate. Worse, the weighting and adjustment factors in such a model would make it difficult to replicate and therefore easy to manipulate to give a particular set of results, especially in any attempts to incorporate 'local factors' (ie. anecdotal evidence). Our simple method, in contrast, can be reproduced by anyone who wants to double check it.
The 2017 general election result – held along the same constituency boundaries only two years ago with largely the same electorates – is by far the best base for comparison. We made the same argument in the 2017 general election (in relation to 2015 results) and it turned out that our recommendations were the most accurate. Some other sites used methods were more complicated and so appeared more 'sophisticated', but ended up skewing their results and introducing errors.
The point of this website isn't to list who we'd like to win in each constituency, but who we believe is most likely to defeat the Tories – the best tactical vote. It will be no use to anyone if we advise tactical votes for candidates who are a long way behind on the basis of wishful thinking from supporters of one party or another – we have to stick to what the data tells us. Of course, if it needs to be said, your vote is still entirely your decision.
We're only aware of this being the case in Northern Ireland seats. In that case, we have considered which parties would and would not support a Tory government in a confidence vote. This boils down to tactical voting against the DUP, which propped up the Tory government through a confidence-and-supply arrangement.
Let us know on [email protected] if you see a result that you think is wrong. If you make an argument that's covered in the FAQ above then we won't change the result, but in the past people have spotted errors caused either by software bugs or recent changes in the situation in the seat, so do tell us if something doesn't look right.
This is a project from @votetools, a collective of coders and volunteers formed during the 2017 general election. You can see our list of spokespeople here. We are funded by donations to our crowdfunder.
If you can afford to donate even a small amount it would be very helpful. Donations will be spent on improving the site and promoting tactical voting as widely as possible.
You could also volunteer your time. Whether your skills are in coding/tech, design/illustration or media/social media, we'd love to hear from you - email [email protected]