This morning Interel hosted one its regular breakfast briefing events. We were joined by the respected pollster and joint founder of YouGov, Joe Twyman.

It has been a difficult couple of years for pollsters. The industry as a whole has come under fire for failing to accurately predict the outcome of the 2015 General Election or the EU referendum last year, leading to many questioning their methodology and leaving the public wondering what purpose election polling really serves if it is so inaccurate.

Polling throughout the election campaign has been broadly consistent; no poll so far has doubted that the Conservatives will be the most popular party. The polls have however narrowed over the past six weeks, and this week YouGov put its head above the parapet, publishing a forecast showing that a hung Parliament is a distinct possibility, with another poll putting the Conservatives only 3% ahead of Labour.

Twyman spoke about the methodology behind YouGov’s model, which was extrapolated from 70,000 surveys and based on known demographic and attitudinal data to give the first constituency-by-constituency estimate of the result. He was keen to point out the wide margin of error, however, with the potential result ranging from the loss of 56 seats at worst, to winning 15 seats at best.

Why, Twyman was asked, does YouGov’s forecast point to such a different result compared to that predicted by other polling companies. The key difference, he explained, comes down to turnout. Whilst it is usually taken as read that young people are the least likely to turnout to vote, YouGov’s data predicts a much higher turnout amongst 18-24 year olds than would usually be expected, two-thirds of whom favour the Labour Party. Conservative support is therefore being overstated in other polls.

This highlights one of the key caveats with election polling in general. Whilst polling data does give an indication of the trends in public opinion, unless voters turn up to cast their votes on the day, the data can very quickly become redundant. It is ultimately turnout, rather than voting intention, that is the core variable in election polling.

Twyman’s assessment of the General Election campaign also gave much food for thought, with his description of the campaign as having moved through comedy, farce and then to tragedy with the events in Manchester last week, providing an apt characterisation of the twists and turns we have seen so far.

He offered caution about any Conservative attempt to run an aggressive attack campaign against Corbyn in the final week, highlighting the need for a positive message and pointing to the failures of the Remain campaign and the Clinton campaign using this same tactic. His scepticism about the impact of tactical voting may however provide a glimmer of reassurance for the Conservatives.

What this election has shown so far is that politics is impossible to predict. Whilst Twyman conceded that the underlying data still favours the Conservatives, even in spite of yesterday’s poll, big divisions still remain between demographic groups and uncertainty over the expected turnout makes the result even more difficult to anticipate.

We all know a week is a long time in politics. Anything could happen between now and next Thursday, but the Conservatives in particular will be hoping they don’t put a foot wrong in that time.