United Kingdom Election - five days to go

The United Kingdom election will be held in five days on Thursday. The current UK Polling Report (UKPR) poll average has the Tories on 34%, Labour on 33%, the Liberal Democrats (Lib Dems) on 9%, the UK Independence Party (UKIP) on 14% and the Greens on 5%; this represents a 1% gain for the Tories since last week.

Based on uniform swings in England and Wales, and a separate uniform swing for Scotland, UKPR has Labour winning 290 of 650 seats, to 263 for the Tories, 19 for the Lib Dems, 60 for the Scottish National Party (SNP) and a few Others, and 18 for Northern Ireland parties. My adjustment to this is to take 20 seats off Labour and give them to the Tories owing to sophomore surges; this method gives the Tories 283 seats to Labour's 270. If this is the result, the Tories would be the largest party by 13 seats, but Labour would form some sort of government with SNP support.

There are some signs that Labour has regained ground in the last few days. Four of the last five polls have given Labour a 1-2 point lead, and the fifth was tied. As a result, the May 2015 poll average now has a dead heat between the major parties, after showing a slender Tory lead. A caveat is that the most recent polls are Internet based panel polls, and phone polls have been better for the Tories.

The last three Scottish polls all give the SNP over 50% of the Scotland vote, with Labour in the low to mid 20's. As a result, there has been some speculation that the SNP could win all 59 of Scotland's seats. The latest forecasts have the SNP winning between 49 and 56 seats

I have some concerns about the proportion of undecided voters in the UK polls. This chart shows the percentage of men and women in the various polls who say they are undecided, and it implies an undecided rate between 9 and 20% depending on the pollster. By contrast, the latest Federal Newspoll had a total undecided/refused rate of only 6%.

I think the UK pollsters have many more undecided voters than Australian pollsters because they do not follow up their first voting intentions question with a question that pushes weakly committed voters into saying which way they are LEANING.

Many UK pollsters assign some of the undecided voters to the party they recalled voting for at the last election. I think this is flawed because many of the undecided voters will not accurately remember who they voted for five years ago! There is in fact a tendency for people to say they voted for the winning party, even if they did not. As a result, more of the current undecided voters probably remember voting for the Tories than actually did.

This problem could explain why phone polls have been more Tory-friendly than Internet based polls. Internet polls use a panel, and when someone new joins the panel, I think they would be asked then how they voted in 2010. Phone polls need to rely completely on recall of an event now five years old.

How votes are counted

Polls will close in the UK at 10pm Thursday night local time (7am Friday Australian Eastern). An exit poll will be released as soon as the polls close that will project the number of seats for the Tories, Labour, the Lib Dems, the SNP and UKIP. The exit poll only samples key seats, and does not provide information about the national vote share of the various parties.

Unlike Australia, votes are not counted in individual polling stations. Instead, all polling place votes are transported to a constituency's counting centre, and it is only after all polling place votes have arrived that official vote counting begins. Once all votes have been counted, barring a recount request, a winner is declared.

This method of vote counting means that there will be few actual election returns until at least two hours after the polls close. Most seats will be declared in the small hours of Friday morning UK time, though a few seats will not be declared until well into Friday. By about 4am UK time (1pm Aus), we should have a good idea of the result.

Adrian Beaumont, PhD Student, Department of Mathematics and Statistics, University of Melbourne

This article is republished from The Conversation under a Creative Commons license. Read the original article.