What a wonderful word psephology is! It’s defined as “the study of elections and voting, and their statistical analysis in the prediction of results.”
Thanks to The Guardian’s Newsblog (the comments are hilarious – as a result, I suspect, of Operation Clark County) I today discovered MysteryPollster, a wonderful website put together by Democratic pollster Mark Blumenthal. In a completely non-partisan way, he helps decipher what the polls mean, and the factors behind them that mean you can’t take everything they say at face value.
He explains, in some detail, how margins of error work (it’s not as simple as you think), and raises some interesting questions. It seems that two things could really lead us to be misled by the polls:
1) Are pollsters making the right judgements about who is likely to vote. There are some staggering figures: of 203m voting age people at the last election, only 130m were registered to vote (64%), and 105m actually voted (52%). That seems like a small number to me. However in the UK, the turnout at the last election was 59%, although that’s calculated as votes cast as a percentage of registered voters – 44m or so. The difference is that most people are on the electoral roll in the UK. This isn’t the case in the States. What this could mean is that if all the initiatives from people ranged from REM and the Dixie Chicks, to Michael Moore, across to P Diddy, have some kind of real effect, then we could see many more new voters than in recent times. If you know you can make the difference, you have an added incentive to vote.
2) Are the Republicans seeing a larger part of its support coming from areas where the votes are already safe. So if Texas is definitely going to go Republican, it makes little difference if everyone in the State votes Republican – it’s the battleground states that matter.