Owen Smith has attempted to discredit Jeremy Corbyn’s claim that Labour was ahead in the polls until the attempts to oust him began – by using aggregated data from a voluntary project, which two years behind schedule.
In a debate with Jeremy Corbyn, Owen Smith quoted data from Britain Elects, which on their own website admit they are not as up-to-date as planned. It says on their website:
Our site, two years behind schedule, is currently under construction, but progress is being made! Slowly. It will hopefully be live within the next few months, although please don’t put that in your diary. If you wish to get in touch in the meantime, be it data, media or somesuch requests you can contact us on Twitter or throw us on email on firstname.lastname@example.org. We’ll aim to get back to you as soon as possible, but as we’re currently an entirely voluntary service, expect a delay by up to a few days.
Edward Parker, a supporter of Owen Smith, attempted to justify Owen Smith’s choice of data. “[I]t’s the average poll of polls, put together by Britain Elects,” he said. “[A]lso we haven’t been ahead in an opinion poll since April.”
Jonathan Bishop, a Fellow of the Royal Statistical Society, says the method used by Britain Elects is suspect. “In simplest terms, a thousand flies does not make dog crap good,” he said. “Crocels Research predicted the Welsh Assembly results accurately by using a linear regression of the past performance of the parties when they contested elections against each other.
“Aggregating different polls does nothing to help improve the statistical significance of the data, so if this is the only way Owen Smith can attempt to discredit Jeremy Corbyn then we should be asking whether his £200bn New Deal figures add up also.
“At Crocels Base we have been looking into what Owen Smith has said in the past and what he is saying now, and whilst he has a good record of announcing big budget projects, he has never been in a position to actual implement them.“