After #Election2015: How can we save Big Data? The Skype window crackled into life, beaming a little piece of Manhattan live into my Shoreditch loft workspace. The ceiling, I could see, was decorated with empty vodka bottles and men’s underwear. A woodcutter’s axe, a signed souvenir from Marcus Mumford, formed the centrepiece, buried ironically into an upside down 6-core …
I saw that programme and it looked like a nice jolly for Nate Silver, paid out of our licence fee, that didn't really mean very much. He chucked in a few caveats regarding the info available here compared with the data he uses in the US, but he was definitely out of his comfort zone.
However I shall certainly be looking at fivethirtyeight.com as the contest hots up over there.
Labour's internal surveying based on interactions showed it pretty early on (different from a methodology perspective in that we made sure we called the named person we needed, even if that took 10 attempts, other pollsters go with who is in). But even the forecasts we produced were still at around 300-305 for the Tories and 260-265 for Labour... closer than any of the other polling (and we got the single digit for Lib Dems correct for the past 12 months, how no one else got that one is beyond me). The issue was that it was a multiple swing dynamic to take account of. Looking at the Labour data, just post the Nichola "judas kiss" on Ed in the 2nd debate you can see the only direct swing of the campaign, UKIP to Tory of 6pts. All others were bouncing around like mad. Lib Dem to both Labour and Tory minus the UKIP from both, plus some anti-SNP voters, minus an army of SNP punishers for the big-3 (formerly). It was a dogs dinner and then people were uncertain too (other than in Scotland) so a significant number voted on how they felt in the booth, not shy, just meh.
Labour didn't actually loose that much from the English "Northern heartlands"... about 4-6% and mostly to UKIP, partly for some of the reasons as in the article. But they did pick up a lot of drifters from Lib Dem voting middle classes (about 8-10%, more than half of Lib Dem lost votes) and they had very little swing from Tory in London (4%) and to Tory in southern England (3%). Biggest impact issue was the SNP "alliance" that took about 30% of undecided voters to a firm position (mostly against Labour). But still around 40% of undecided voters remained undecided in the day or two before... that was much higher than the typical 10-20%. Labourvoter turnout was also down too (estimated at about 2-3%, ironically enough to have swung the election substantially), suggesting that some of those supporting Labour just never got motivated enough.
Anonymous, just in case someone dislikes sharing the "insight", not that it's particularly insightful.
Downvoted for missing the obvious point that if only our new Chief Data Officer had been appointed earlier the pollsters could have been scrummed into shape.
... Nate was hatched from an egg at the Sea Life Centre in Weymouth, England, then moved to a tank at one of the chain's centres at Oberhausen in Germany. His name derives from the title of a poem by the German children's writer Boy Lornsen: Der Tintenfisch Nate Silver.
According to Sea Life's entertainment director, Daniel Fey, Nate demonstrated intelligence early in life: "There was something about the way he looked at our visitors when they came close to the tank. It was so unusual, so we tried to find out what his special talents were."
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
