Turnout will be pivotal, and recent history bodes ill for Labour on that score
In general, by-elections are the hardest political markets to predict. Opinion polls are even less reliable than usual, asking a mid-term question that only a small minority of engaged voters have spent more than a few minutes considering.
Recent UK by-elections have seen some highly unpredictable results, particularly in supposedly safe Labour seats.
In 2012, George Galloway’s Respect pulled off an enormous upset in Labour-held Bradford West on a whopping 53% swing. In 2008, the SNP took Glasgow East from Labour on a 22% swing. Labour were rated in excess of 90% likely to win on the Betfair market on both occasions.
In 2006, the Lib Dems won Dunfermline and West Fife, after Labour had been matched at the minimum odds of 1-100. In all three cases, the seats returned to their natural owners at the next General Election.
Galloway aside, the big by-election story of the last parliament was the rise of UKIP. Nigel Farage’s insurgents took two Tory seats by wide margins, and finished second in a spate of contests in safe Labour seats. In Eastleigh, they came within a whisker of taking this high-profile Lib/Con marginal.
Thursday’s contest in Oldham West & Royton would, in normal times, be very low-key. Labour would expect to hold it, with UKIP almost certainly finishing a respectable, if distant, second. But these are not normal times.