My opinion on opinion polls

Now I don’t claim to be Nate Silver with his predictive/analytical abilities but I have made reasonable living from working with data. So I feel duty bound to give my views on the election and what’s the point of the polls.

How useful are these polls, they are simply measuring the share of the vote. They claim to be weighted to represent the British public. Each poll surveys approximately 1100 respondents (some as low as 500), this equates to less than 2 people per constituency. With each Seat being contested by 5-7 parties how can the media and commentators possibly base their analysis on this basis. The poll of polls has become a popular method for ironing out any discrepancies between the polls but again this is simply just measuring the most popular candidate within the sample at that time.

There has been a huge shift in the voting patterns of the British public from a position where the two main parties accounted for 95% of the vote in 1951 to 65% of the vote in 2010 and probably a smaller share on Thursday. In a simple two horse race share of the vote polling is more effective as you can be default assume a loss for one is a gain for another.

However predictive polls continue to be the main fodder for media outlets and commentators to base their analysis of who will form the next government come May the 8^th.

The election will be decided quite simply by the marginal seats and in effect we should concentrate on polling within these seats to determine the outcome of who may or may not gain/lose here.

I would advise anyone to watch Nate Silver and Richard Bacon breaking the law in the Airstream caravan, still available on iPlayer to understand what might happen on Thursday. You don’t have to be a data geek to enjoy it.

Paul Kemp