Used properly, opinion polls can be an important way of measuring what voters think about particular issues, parties and candidates. Newspapers and broadcasters often commission their own polls to give them information about voters' intentions. An opinion poll can also be a way of finding out what voters think about a particular issue - or what they think the important issues are. These might not be the same as the issues that seem important to politicians. Arguably, opinion polls help to enhance democratic choice. For example, in a first past the post system they may assist those who wish to vote tactically to ensure that a certain candidate fails.
The danger with opinion polls is that they are subject to manipulation at many levels: choosing the questions, choosing the sample, choosing the time to ask the questions and so on. This is quite apart from the normal margin of error in any opinion poll - which in a closely fought election may be greater than the margin that separates the parties. A well-conducted poll is usually remarkably accurate. It is reckoned that a sample of 1,000 can accurately reflect the views of more than 200 million US adults to within a few percentage points. But their limitations must be clearly understood too. That is why media reporting of opinion poll findings demands the highest professional standards. It is also why, arguably, reporting of opinion polls should be subject to regulation by the electoral supervisory body to make sure that the media are not communicating deliberate falsehoods.
In the United Kingdom, internal guidelines remind British Broadcasting Corporation staff to downplay the significance of opinion poll results and to emphasize that they represent only a snapshot of opinion at a particular moment. (See 'mey24'.) Accuracy is clearly improved if media report the result of all opinion polls, especially if they then average the results, hence minimizing the impact of 'rogue' polls, which can occur even with the most professional of polling techniques.
Professional coverage of opinion polls means asking a number of key questions about how the survey was conducted - and telling the audience the answers:
- Who conducted the poll? Are they reputable and independent?
- How many people were interviewed?
- How were they chosen?
- Are the published results based upon the answers of all those interviewed?
- When was the poll conducted?
- What is the sampling error?
- What questions were asked - and how were they worded? In what order were they asked?
- How do the results of this poll compare with other findings?
See Questions to Ask About Opinion Polling and 'mey29'.