The user-defined prediction also gives a (simple) model of tactical voting, which lets you control the size and direction of any effect.
The idea of the model is that some voters whose first-choice party is likely to come in third place would rather vote for their second-choice party which is more likely to win.
See the Tactical Voting Model for more details about the model itself and the historical experience.
We call this a 25% "tactical fraction" of Lib Dem supporters.
In the user-defined prediction entry page, there is an area for entering the tactical swing levels:
In this example, the user specifies a zero tactical fraction of Conservative voters, a 25% tactical fraction of Labour voters and a 35% tactical fraction for Lib Dem supporters.
At the 2019 general election, there were significant amounts of both pro- and anti- Conservative tactical voting. In some seats, tactical voting was as high as 50pc, but in other seats it could be around 20pc.
Voters who vote tactically do not all pick the same destination for their tactical vote. It depends on which party they basically support, and also on the characteristics of the seat. For example, non-Green voters are unlikely to vote Green tactically unless the Green party is competitive in that particular seat.
The general trends of tactical preferences at the 2019 election were estimated as follows:
Additionally in Scotland, there was some pro-Union tactical voting against the SNP, but this is not captured by this model.