Wednesday 30 April 2008
Turnout is always the crucial factor in London Mayoral elections. We know from previous election data that people are persistently poor at predicting whether or not they themselves will vote on election day! Turnout in 2004 was only 36% but in most recent polls a much larger proportion have told researchers they expect to vote this time.
Therefore in order to understand how different turnouts will affect the election outcome, we have had a more in-depth look at our final poll before the election, carried out for the trade union UNISON from 23-24 April. From this data, we can model the balance of support between the candidates at differing levels of turnout, and therefore make an estimate about how the turnout could ultimately affect the results.
The charts below illustrate a step-by-step description of how we have made our calculations. In sum, a turnout of 73% or higher (which is extremely unlikely to happen), would give Ken Livingstone a statistically significant lead. As turnout falls, Livingstone's lead falls and Johnson's share increases. A turnout of just over 40% is literally a 'dead heat' statistically. Below this level of turnout, the scales tip towards Boris Johnson — although his lead is not statistically significant until turnout falls as low as 10%-15%, also a very unlikely outcome.