Monday, June 05, 2017

Why Polls May Overstate The Likelihood Of A Conservative Government


In the 2015 election the opinion polls understated the Conservative vote. It was felt by many that this was due to under representation and also to a failure to adequately adjust for turnout. This time the opinion poll companies are adjusting the raw voting intention data from their samples and coming up with a headline number by adjusting for factors such as turnout.

As we can see above these adjustments tend to significantly increase the Conservative vote. Taking the average of the headline figures you get a Tory lead of around 7%. This many seem like a significant lead but at the last election the Tories received a 6.5% higher vote share than Labour which only translated into a very slim majority with 331 seats.

The latest YouGov current election forecast has a Tory lead of only 4% which means they would fall short of a majority by 21 seats. See chart below.


Given we haven't had any major boundary changes the Tories may need more than a 6.5% lead to increase their current majority. The average headline lead from all the polls is just 7%. If the poll adjustments are even slightly out we could see a hung parliament, as anything less than a 6% Tory lead means they may fail to get a majority.

Nate Silver, the US election analyst has recently reviewed the new polling adjustments and says "Given the poor historical accuracy of U.K. polls the true margin of error on the Labour-Conservative margin is plus or minus 10 points."

This is a large margin which means within the margin of error we could have a Tory lead of 17% or even a labour lead of 3%. So a Conservative win is still the most likely outcome but a hung parliament also appears to be a real possibility from the polling data.