Under our first past the post electoral system, many voters will be looking to cast their vote tactically in order to have the most impact.
We want to work out the extent to which voters will vote tactically and the parties to whom they will switch allegiance. And whether that changes the outcome in their seat.
In our recent MRP poll of over 10,000 respondents, we asked people four questions about tactical voting. We asked how likely they were to vote tactically if their first-choice party was unlikely to be amongst the stronger two parties in their seat. We also asked them who is their second-choice party to work out how supporters of each party would vote tactically.
Are Labour voters likely to vote Liberal Democrat in seats where Labour are perceived as weak? Are Reform voters willing to lend support to the Conservatives in seats where they are facing a strong challenge?
The results were stark and varied nationally.
In England, we discovered that Reform voters were the least likely to vote tactically with only 34% saying they would vote for another party. This bodes poorly for Rishi Sunak, especially given recent policy announcements that are designed to cajole Reform voters into lending him their support. Fewer than a quarter of Reform supporters would consider voting Conservative tactically.
On the other hand, we found that many Liberal Democrat, Labour and Green voters were minded to vote tactically with 45%-50% of them saying they would. This tactical voting is almost entirely anti-Conservative.
In Scotland, the group most likely to behave tactically were Labour and Liberal Democrat voters who appeared to be willing to lend support to one another in a likely bid to boost unionist candidates. On the pro-independence side, SNP and Green voters were quite inclined to switch between each other's parties.
People in Wales are less inclined to vote tactically, although the sample size is fairly small.
Another key finding is that voters are confused about which party to vote tactically for. This is partly due to the new constituency boundaries, and partly due to the big changes in public opinion since 2019.
Overall anti-Conservative tactical voting could cost the Conservatives around six more seats. But voter confusion saves the Conservatives in nine more seats.
In Scotland, pro-Independence tactical voting could help the SNP is a handful of seats.The Liberal Democrats could gain 10-20 extra seats through anti-Conservative tactical voting.
In summary, the results show that the impact of tactical voting on the upcoming general election takes the Conservatives' electoral fortunes from very bad to slightly worse. With that said, where the Conservatives may benefit is from the ambiguity of voters as to who is likely to be the main challenger to the Conservatives, neutralising some of the impact tactical voting may have.
In addition to voting intention, we asked respondents four additional questions about tactical voting.
Q3. Suppose all the of following political parties were equally likely to win your seat at a general election. Which party would you most like to win? Q4. Other than the party you just picked, which party would be your second choice? Q5. Suppose your first-choice party is unlikely to win the seat, but your second-choice party could. How likely are you to switch your vote to your second-choice party? Q6. Which pair of parties do you think are likely to come in first and second place (in either order) in your seat at the next general election. The purpose of these questions is as follows:Our algorithm for analysing these questions goes like this.
1. For each party, look at the universe of people who match all these criteria:
This defines the universe of supporters of a particular party who live in seats which that party is unlikely to win, and have not already decided to vote tactically.
2. Of this universe of people, we look at the subsets, for each alternative party to switch to, of the people who:
3. We calculate the size of the group defined in stage 2 divided by the size of the universe in stage 1. This gives the fraction of people who support the first party but would be likely to switch to that alternative party. The results of this analysis are tactical transition matrices – one each for England, Wales and Scotland – which are given in the Appendix below.
We can also use the answers to work out if voters have a clear idea of which parties will do well in their seat. The results show a degree of confusion. Partly this is because of two extraneous factors:
When voters are asked to think about which two parties will do well in their seat, they may be thinking of the old boundary seat or the new boundary seat, and they may be thinking of the 2019 result, or a predicted result. They may also have an unrealistic view anyway.
There are also some anti-Conservative tactical voting sites, who have their own advice. This is sometimes in favour of the largest anti-Conservative party in the implied 2019 new-boundary results.
Slightly less than half of people (48%) correctly identified the two likely best-placed parties in their own seat. There was a very slight trend in favour of using the implied 2019 new-boundary results as a determinant, but the predicted new-boundary results were about as good a match.
To explore the difference, we will run both scenarios to see what difference it makes.
We can apply these tactical propensities to the predicted seat vote shares, to see what impact tactical could have on actual seats.
We have a separate forecast of seats win based on three different scenarios:
| Party | No TV | TV 2019 | TV 2024 |
|---|---|---|---|
| CON | 72 | 66 | 57 |
| LAB | 493 | 476 | 495 |
| LIB | 39 | 59 | 51 |
| Reform | 0 | 0 | 0 |
| Green | 2 | 2 | 2 |
| SNP | 22 | 26 | 23 |
| Plaid | 4 | 3 | 4 |
Table 1: Predicted seats won under various tactical scenarios
Under the "TV 2019" scenario, the Conservatives lose six more seats, and the Liberal Democrats gain twenty more seats, slightly reducing Labour's total. In Scotland the SNP gain 4 seats from pro-Independence tactical voting.
Under the "TV 2024" scenario, the Conservatives lose fifteen seats compared with the baseline, but the Liberal Democrats only gain twelve more. Labour gains two seats.
In terms of maximising anti-Conservative tactical voting, the focus on the implied 2019 general election is sub-optimal. It achieves less than half the effect of optimal tactical voting.
Of course, many voters may already have decided to vote tactically anyway. We estimate about 16% of voters have already decided to vote tactically, based on the difference of responses between the VI question and Q3 above. In particular, many Lib Dem supporters (30%) and Green supporters (49%) have already decided to vote tactically, mostly for Labour.
Further tactical voting may only have a limited effect. The main limits to tactical voting are:
The following are what we call Tactical Transition Matrices. Each row indicates what natural supporters of one party would likely do if that party was not strong in a seat. The diagonal entry indicates the fraction of those voters who would not vote tactically, and the other elements indicate the second preferences of those who would vote tactically.
The transition matrix for England is given below.
| Party | CON | LAB | LIB | Reform | Green |
|---|---|---|---|---|---|
| CON | 64% | 3% | 9% | 19% | 5% |
| LAB | 0% | 54% | 24% | 0% | 22% |
| LIB | 8% | 14% | 50% | 5% | 24% |
| Reform | 22% | 6% | 2% | 66% | 4% |
| Green | 2% | 27% | 14% | 1% | 55% |
Table A1: Tactical Transition Matrix for England
For example, the Labour row says that 54% of Labour supporters would not vote tactically even if Labour was unlikely to win the seat. But 24% would switch to the Lib Dems, and 22% would switch to the Greens if those parties were competitive. No Labour supporter would switch to the Conservatives or Reform.
We see that most (64%) of Conservative voters would not vote tactically, and most (66%) of Reform voters would not either. Liberal Democrat and Green voters are more prepared to vote tactically and would mostly share their votes between those two parties and Labour.
There is a separate matrix for Wales:
| Party | CON | LAB | LIB | Reform | Green | Plaid |
|---|---|---|---|---|---|---|
| CON | 69% | 3% | 19% | 8% | 0% | 7% |
| LAB | 0% | 71% | 29% | 0% | 0% | 0% |
| LIB | 0% | 0% | 78% | 0% | 6% | 16% |
| Reform | 21% | 3% | 0% | 76% | 0% | 0% |
| Green | 0% | 6% | 2% | 2% | 89% | 3% |
| Plaid | 0% | 8% | 15% | 0% | 10% | 67% |
Table A2: Tactical Transition Matrix for Wales
In Wales, there is less inclination to vote tactically, with high values on the diagonal.
In Scotland, the transition matrix is different.
| Party | CON | LAB | LIB | Reform | Green | SNP |
|---|---|---|---|---|---|---|
| CON | 59% | 0% | 20% | 21% | 0% | 0% |
| LAB | 0% | 31% | 66% | 0% | 3% | 0% |
| LIB | 0% | 53% | 18% | 0% | 15% | 13% |
| Reform | 26% | 11% | 0% | 58% | 0% | 5% |
| Green | 0% | 2% | 0% | 2% | 56% | 40% |
| SNP | 0% | 0% | 0% | 0% | 0% | 100% |
Table A3: Tactical Transition Matrix for Scotland
There is clear pro-Union tactical voting between labour and the Liberal Democrats, which is even stronger than in England. There is also strong pro-Independence tactical voting by Greens to the SNP. SNP voters have no opportunity to vote tactically because the SNP is competitive in all Scottish seats, and the SNP row has (unused) dummy data.