Wednesday, 10 August 2011

Biased Polls

I have some limited experience of polling, and one of the most important features of any honest poll designed to seriously test opinion and obtain useful information is that it must never guide the respondent to a specific answer. The wording, categories, descriptions etc must always be kept neutral so the respondent gives a true answer without prompt, nudge or emotional knee-jerk reaction.

Of course this is what the MSM don’t do, they will blatantly write a scare story, usually couched in emotive terms, about some issue, then follow with a poll. For example the crime rate and someone being mugged, then asking if people want more policemen on the beat. Inevitably they get the desired answer because the poll was heavily loaded from the start to obtain that answer.  Any other aspects of the question are not mentioned in the article.

The same tactic is used by almost every lobby group and pressure group and of course politicians. Find one specific example of a practice that people won’t like, publicise that aspect, then ask a leading question about the activity. Anti-smoking, anti-alcohol, anti-salt and suchlike movements always work to this pattern.  Use something simplistic and irritating to people, load the language of the question with carefully biased wording. “Do you know how much scarce NHS resource is wasted on alcohol related injury? (note the words “scarce and  “wasted”, which are emotionally loaded). Then follow up with a poll whether the NHS should charge drunks YES/NO. You’ll probably get a “yes”.  I’m ignoring the deliberate lies and misleading statistical distortions often used. All it needs is some carefully loaded emotive reporting and a leading question.

There is a wonderful section in “Yes Minister” to illustrate the technique of using loaded questions to achieve an outcome but I can only find this version online, via someone else who notes distorted polls.

But there are other far more subtle ways that distort polls, Things that a true seeker after opinion works hard to avoid, and that is biasing results by removing an element of choice, or by having confusing choices or ill thought out questions.  Asking for a YES/NO answer when most people have a nuanced attitude is a simplistic example. Often the problem is swept under the carpet by having an ‘other’ option, but even then many people will try to fit themselves into the tick box options.

This leads me to the Total Politic poll of blogs and bloggers, which prompted my train of thought.  There is a free choice of blog entry vote but then you are asked to categorise them. Look at the categories.

Right wing        Left Wing         Libertarian        Non-aligned          Labour
Conservative    Lib Dem           Green                Group blog       Scottish
Welsh               Northern Irish   Councillor         MP                   Media

Are we meant to categorise as we interpret them or as the blogger describes them? No idea!

But more importantly notice how the choices control the responses. A blog may be libertarian but not authoritarian. It cannot be localised, there is no local category although local politics may be just as political. The blog can be Welsh, Scottish or from NI but can’t be recorded as English, are we not allowed to describe blogs that concentrate on English affairs?  Read any good English based UKIP, anarchist (or even marxist) blogs anyone? 

These restrictions suggest to me that Total Politics, while not defining the winning choice in any way, have a very preconceived set of ideas as to the appropriate contents and background of a political blog.

No comments:

Post a Comment