What’s that margin of error again?
September 27, 2011
Poor Ed Miliband. Not only does the live TV feed break down just minutes into his speech to the Labour Party conference in Liverpool, but a poll in the Independent reports that “the Tories enjoy a lead for the first time since October last year”.
The Conservatives are on an impressive 37% while Labour languishes behind with a paltry 36%. So that’s that, then!
Except that it isn’t.
Our old friend, “margin of error”, raises its weary head yet again to object that no poll is entirely accurate: so what’s the margin of error on this one? The Independent doesn’t tell us, although there is a link to the Comres website, who conducted the poll. That doesn’t state the margin, either.
However, there is a link to handy “Margin of error calculator” on the Comres site. So, taking the UK adult population to be 51m, and with a sample size of 1,000, that gives a margin of error of 3.1% (actually, the population is the adult population who intend to vote – let’s be generous and assume that’s 25m. It makes no difference to the margin).
So we end up with a margin of error of around 3%. Hence before we can conclude the Tories have clear water between them and Labour, their lead would have to be at least 6%, not the 1% found in the survey.
Perhaps it’s not so bad for Ed, after all – now who unplugged that cable …?
(PS: Wikipedia gives the number of people who voted in the 2010 election as 27,148,510, which sounds about right – doesn’t alter the 3.1% margin of error, though)