Monday, 4 January 2010

Political Polling and Political Prediction

Political polling was one of those areas of political study that failed to command my undergraduate attention.  It is a kellnerish area of study - ie it requires high levels of data handling skills and  niggly attention to detail, remote from political theory and, even more distressingly, immediate political events; its  objectivity was always being called into question which removed  its interest as a particularly valid external indicator of the effects of political choice and policies on voters; being such a thing would have made it fascinating, but as merely another political propaganda means of affecting voters' decisions, it wasn't worth the effort of mastering its arcane difficulties and subtleties.

So which is it: a boys' toy that keeps them entertained fiddling with  statistics and modes of comparison, and sample reliability and conformity that could be substituted with a well-informed guess; or a powerful predictor of voter behaviour that interacts with that behaviour.  That the answer might now be the latter (after technique refinements and the instigation of a professional rules-giving body)  is underpinned by the publication of political polls being prohibited, just before elections take place, in many democratic countries.

Political polls are only interesting when they are technically neutral, and most interesting when the questions asked are emotionally intelligent.   Emotionally intelligent questions are intrinsically part of truth-finding.  A true question requires a question that is capable of being answered; and a true answer requires maximum co-operation from the answerer.  To ask: "how will you vote at the next election?" is a question  incapable of  evincing a 'true' answer.  And the answer: "Labour", is not a 'true'  answer.

What is needed is a profile of a Labour voter, which is then matched or not onto  respondents by a series of much narrower and less loaded - culturally, historically, psychologically  etc., - questions; questions that are tailored carefully to the circumstances of the respondents.   Framing truth-seeking emotionally intelligent questions and evaluating the emotional intelligence of the answers is not yet highly developed in political polling.  A lot of what looks like neutral data-gathering and sophisticated adjustment and interpretation is still close to any informed, intuitive grab on what is being picked up from many and varied sources.

I think. But then I should have been paying more attention when I had the time and opportunity.   


Odin's Raven said...

Banning opinion polls before an election makes sense to those who want to manipulate the results of the election. There is no longer a comparison with a reasonable expectation of the outcome and 'media scripts' like 'too close to call' or 'hung parliament' become more easily accepted. Our current rulers have specialised in media manipulation, so it has obvious attractions for them. Welcome to post-democracy.

Caronte said...

The publication of poll results, unavoidably, paves the way to strategic voting: people do not express their preferences in absolute, but their preferences conditionally on to the voting intentions revealed by the polls.

Faced with the prospect of Labour likely defeat, for instance, I may respond by mobilising my vote and that of my friends and relatives in support, or - more likely - I might give up in despair. Either way, the result is corrupted by games unleashed by the pollsters.