I read this by Anthony Wells:
As you’ll probably know, the Wythenshawe and Sale East by-election was also last night, and was a comfortable Labour hold. This means today will be full of people saying what it *means* and trying to draw some wider conclusions based upon it. I’ll only repeat my normal warning about not reading too much into by-elections. They are extremely unusual beasts – an election in just one single seat that won’t be representative of the whole country, intensely fought but often with low turnout, and where who wins does not make any difference to who the government is the next day. Essentially, if a by-election performs in line with the national polls it doesn’t tell us anything we didn’t already know, if it performs in some way different to the polls it’s probably because of the unusual circumstances implicit in a by-election.
Please, I beg anyone, read that last sentence again. With care.
Because I think it *means* the polls are more reliable than votes-in-the-box. And that is a very, very dangerous assumption.
On the whole, an opinion poll is a sample — typically a bit over a thousand bods, chosen by gender, age, and past experience, to represent a wider population. So those 1,000+ are to be treated as an “accurate” representation (+/— the margin of error, which is a convenient cop-out) of the 42 million electors in Britain.
Wow! Suddenly I’m with the opening Prologue of Henry V:
Piece out our imperfections with your thoughts;
Into a thousand parts divide one man,
And make imaginary puissance…
Except, by the statistical legerdemain of the pollster, one man comes to be about 42,000. Which is an imaginary puissance beyond even Will Shagsper at his prime.
Where Anthony Wells has a point
He is correct to say that any by-election is an unfair “sample” of the whole of this nation.
Or, were we to do so, we should only concern ourselves with those bellwether constituencies which blow with the prevailing political wind: Dartford, Basildon, Luton, Reading — all conveniently an hour or so on the train from central London, and so of metropolitan importance.
Where Anthony Wells goes too far
Of course Wythenshawe and Sale East matters. Even if only 24,ooo voters turned out on a foul night in mid-February.
Mr Wells should look carefully at what happened there yesterday, compare it with “Lord” Ashcroft’s poll, review the electoral history of this constituency over the last four General Elections.
He would see that:
- The Labour vote (and conventionally, working-class Labour voters are more of a problem to turn out than others, especially in a rock-safe constituency) was right in the middle of predictability;
- The Tory vote plummeted. There has been a solid 25% Tory vote here since 1997. Last night, four-in-ten Tory voters didn’t bother — or went elsewhere.
- When four out of every five “natural” LibDems absent themselves, something is badly, sadly amiss.
- And suddenly the UKIP “surge”, as the sump for hopeless-but-motivated dissident, has some meaning.
Where Anthony Wells is OTT
He claims that polling is more “representative” than elections.
That his statistical tweekery is *means* something, is of more reality (his own implication!) than real people going out on a miserable day to make real votes at a real polling station, to be counted by real polling clerks.
Nonsense. Rubbish. Crap.
Such arrogance exceeds the mark.
After hubris comes nemesis.