The fight is on for the Gorton and Denton byelection on February 26. It is a three-way contest between Labour, Reform and the Greens. According to Electoral Calculus, a site which runs regular MRP (multi-level regression post stratification) surveys, Reform will win with 32% of the vote, the Greens will come second on 23.3% and Labour third with 22.6%.

There is however a problem with MRP polls. They produce results that are far more variable than can be accounted for by standard sources of errors in surveys. For example, a different MRP conducted by Electoral Calculus in December last year gave Reform 335 seats, the Greens 52 and Labour 41 in a hypothetical general election.

In contrast, an MRP conducted by More in Common at about the same time gave Reform 381 seats, the Greens nine and Labour 85. Both results are therefore highly questionable.

An alternative approach to forecasting involves conducting small sample surveys in constituencies holding byelections. The trouble is that these are not very accurate – as the recent Caerphilly byelection for the Welsh Senedd proved. Based on polling, many believed Reform was a contender to win but it ended up some way behind Plaid Cymru.

The relationship between the Labour vote in general elections and subsequent byelections, 2010 to 2026:

A third approach is to use historical data on the relationship between byelections and........

© The Conversation