Thursday, 17 October 2013

Adroit-e Polling: The Weighting Game

Adroit-e Polling: The Weighting Game
It can be very frustrating.
Why do most National Polls more or less agree on the state of the parties at any given time, but then one Poll produces very different Results? The answer often lies with the interpretation of that data or its analysis, via a weighting system
The Adroit-e constituency Polling system provides the scores to you as found – then you can ask us to weight them in a particular way – or not:
Let’s suppose that the question asked is ‘if the election were tomorrow ,who would you vote for?’ The answer (simplified) might be Labour, Tory, Lib Dem, Undecided. On looking at the Undecided Vote the Polling company discovers that 60% of those undecided, voted Tory last time.  ‘Aha!’ they say having an ‘Aha!’ moment, they may say they are undecided but we know if they voted Tory lasttime they are still more likely to vote Tory next time.
So we’ll achieve greater accuracy by ’weighting’ their answer: so in the analysis it’s not shown as a 1 but us a .8 or some such device – This weighting produces more accuracy, allegedly. But as there is no agreed weighting standards, different companies weigh them differently.  And hence, probably, the differences between Voting companies – The Adroit-e view – Weighting claims to bring greater accuracy but actually it merely adds a further layer of subjectivity.
So with  Adroit-e Polling we devise a questionnaire with you, take it out on to the streets to have up close and personal conversations with your voters, provide you with a rapid topline report, then produce an extensively analysed full report 

There’s a difference between accurate and very accurate polling – meaning possibly the difference in getting first past the Post – or not

Jonathan Brill