Politics | June 12, 2017 9:23 am

Why Polls Fail to Accurately Predict Voting in UK Election

After flawed polls in U.S. and U.K. elections, methods are re-examined.

Britain's Prime Minister and leader of the Conservative Party Theresa May, accompanied by her husband Philip, delivers a statement outside 10 Downing Street in central London on June 9, 2017 as results from a snap general election show the Conservatives have lost their majority. British Prime Minister Theresa May said Friday she planned to stick to the timetable for starting Brexit negotiations in 10 days, with a new government that would lead Britain out of the EU. (Justin Tallis/AFP/Getty Images)
Britain's Prime Minister and leader of the Conservative Party Theresa May, accompanied by her husband Philip, delivers a statement outside 10 Downing Street in central London on June 9, 2017 as results from a snap general election show the Conservatives have lost their majority. British Prime Minister Theresa May said Friday she planned to stick to the timetable for starting Brexit negotiations in 10 days, with a new government that would lead Britain out of the EU. (Justin Tallis/AFP/Getty Images)

Polls just aren’t what they used to be.

Statisticians are scratching their heads to find ways to conduct public opinions polls that more accurately the results of an election. The hand-wringing comes after pollsters whiffed yet again when their pre-election surveys erroneously projected the results of Thursday’s British election.

Prior to voting day, polls put Prime Minister Theresa May’s Conservative Party one to 12 points ahead of the rival Labour Party, the Wall Street Journal reports. Instead, the Conservative Party ended up losing seats and fell short of securing the majority in Parliament.

Polls can easily be skewed if small sample groups that don’t reflect the common sentiment was relied upon, but the problem is finding a group that’s diverse enough to be indicative of public opinion—and willing to be surveyed.

Surveying public opinions by cold-calling random households has historically been a reliable way to take the pulse of a nation. Now that only half of American homes have a landline, companies are relying on the Internet polls to supplement since they’re cheaper. But, they’re also less accurate.

Pollsters disagree on the best way to correct his, but there’s a consensus on one point: Something needs to change.