Former Conservative Party election advisor Michael Moszynski is claiming that YouGov's recent polls are 'irresponsible', following its forecast of a hung parliament in the UK's forthcoming general election.
In an article in online news publication The Drum, Moszynski (pictured), who is the CEO of global ad agency LONDON, called out YouGov's forecast, which is based on 'Mulitilevel Regression and Post-stratification' analysis. This method takes what one group of potential voters say and applies it to others, and projects the outcome in individual constituencies based on a sample size of just 75 per seat, which Moszynski says is 'not statistically robust'. YouGov applied this technique to the EU Referendum and US election, but failed to predict either result. For the current campaign, the YouGov model puts the Tories on course to win 310 seats, down from the 330 they held when the election was called.
In his article, Moszynski states that there would be 'more success tossing a coin to guess the results of those elections than relying on YouGov's predictions', adding that as well as 'creating confusion', 'getting it wrong so consistently is damaging the economy'. Using his firm's own Polling Calibration technique, Moszynski is predicting a Conservative majority of 103-108 seats.
In an interview on Sky News, Joe Twyman, YouGov's Head of Political and Social Research (EMEA), rejected the idea that his firm's modelling technique is anything other than a scientific approach, adding that given the fluidity of public opinion, results may very well be different on polling day. He refuted the suggestion that YouGov always gets its predictions wrong, and said he was confident that the firm's model is accurate.
A link to the article and interview can be found at goo.gl/Lmz7Wt .
All articles 2006-23 written and edited by Mel Crowther and/or Nick Thomas, 2024- by Nick Thomas, unless otherwise stated.
Register (free) for Daily Research News
REGISTER FOR NEWS EMAILS
To receive (free) news headlines by email, please register online