|
Media FAQs on Opinion PollingApril 2005 Q1 What is an opinion poll? A1 An opinion poll is a scientific survey designed to measure the views of a specific group at a point in time – for example, in the case of a general election, to measure the views of the country’s electorate at the time the interviewing was done. A2 Yes, there are many differences, although an opinion poll and a conventional market research survey use the same tools and have the same limitations. Whilst both these activities are conducted as a means to measure an audience’s opinions or feelings, the subjects on which they are questioned and the methods deployed can vary considerably, in turn affecting the reliability of results (as with all market research studies). Crucially, opinion polls set out to provide a snapshot of public opinion at one point in time – the point at which the fieldwork was carried out – and unlike some market research studies do not use this information to try to predict the future. Public opinion and voting intentions tend to fluctuate rapidly in the run-up to a general election, especially in marginal constituencies and over highly controversial issues. Hence two opinion polls conducted on consecutive days can have marked differences in outcome. By contrast, market research surveys have much broader uses, and are undertaken to research people’s past, present and future behaviour as much as their opinions – whether it be charting evolving shopping expenditure or looking at favourite grocery brands. Market research is conducted as a means to gauge people’s longer-term attitudes and opinions, usually over less volatile topics than opinion polls. Today, market research is conducted through a variety of mechanisms – from face-to-face and telephone one-to-ones to focus groups – and is generally carried out over a longer period of time than opinion polls, which are revisited daily in the run up to an election. The MRS Code of Conduct governs acceptable interview practise for market research and opinion polling, and all MRS members must abide by the Code. A3 Reputable political polling is conducted by several major polling organisations who use a number of different techniques for sampling and fieldwork. The three main techniques for conducting polls are face-to-face interviewing in respondents’ houses, telephone interviewing and internet-based polling, the latter of which is becoming more widely used. Face-to-face polling is conducted within a number of ‘primary sampling points’ (usually parliamentary constituencies). These are often selected proportional to the number of electors, and should reflect a cross section of social class and political preferences.
A4 There are two methods for ensuring a poll chooses a sample that is truly representative. ‘Random’ sampling involves a polling company using a list of randomly-generated telephone numbers, email addresses or names and home addresses (if they are being drawn from an electoral register or similar list). These people are then contacted and requested to partake in the survey. Theoretically this selection process gives every elector the same chance of being selected for interview. ‘Quota’ sampling involves setting quotas – for example, of age and gender – and seeking out those people in each location who, together, match those characteristics. Quota polls are frequently used by internet-based pollsters who use quota samples to select representative samples from a database of people who have already provided information about themselves. All polling samples seek to be representative of the total electorate of Great Britain by carefully selecting samples from a variety of social groups. The accuracy of a survey depends upon the absolute sample size rather than the proportion it represents of the population it is drawn from. This means that a survey of 1,000 people for the whole country is as reliable as a survey of 1,000 people for a single constituency. Conventionally, Northern Ireland is omitted from British electoral polling as it has a different party system.
A5 Yes there is. Mathematical and statistical theories prove that, in a random poll of 1,000, 19 times out of 20 a poll will be accurate to within 3 per cent. But this does leave the one in 20 chance of the result being outside this margin of error – hence opinion polling is accepted as an inexact science, and it is sensible to weigh up the results of several different polls when calculating the most likely outcome.
A6 No. Polls may not be perfect, but they are proven to be the best available way of measuring what the public thinks. They should be viewed as a means to provide a good guide to the general opinion of the populace; however, by the nature of polling mechanics and sample sizes, the margin for error does mean that in the case of a closely-fought general election, the final outcome may differ from the pollsters’ predicted results. A7 If respondents did often lie, opinion polls would never get their predictions correct, and the industry – which is steadily growing as a key source of information on public opinion – would not have become the valuable tool it is today. Furthermore, opinion polling conducted by MRS members is done in accordance with the MRS Code of Conduct, meaning that results are collected on an unattributable basis so there should be no reason for respondents to lie.
A8 Actually, for the vast majority of elections, most polls have reported correctly to well within the accepted margin for error. It is important to remember that opinion polls set out to provide a snapshot of public opinion at a point in time – the point at which the fieldwork was carried out – and are not necessarily and automatically a predictor of future attitudes and actions. In the general course of events, public opinion and voting intentions are unlikely to be very volatile over a short period. However, they are likely to be more so in the run-up to a general election, especially in marginal constituencies and in situations where tactical voting may be important. When planning and commenting on the results of a poll it is crucial that everyone involved should have this potential volatility clearly in mind. In situations where parties are very close to each other in voting support, it is sensible not to regard any single survey as a totally reliable indicator of opinion, but instead to look at the consistency of the range of polls being conducted at the same time, using different methodology. To ensure results are as accurate as possible, pollsters must also bear in mind expected voter turnout for each political party and ensure their opinion polling results recognise this.
A9 There are several reasons why results amongst pollsters may differ – quite separate from the general issues relating to sampling error. Polls may have been conducted at a different time, even if they are published at the same time; in a run up to a general election people’s views fluctuate considerably as they hear different campaign manifestos, so a single day can make all the difference. Questions may be worded differently between pollsters, which in turn can affect the responses. In the same way, the ordering of questions can affect an outcome – respondents tend to ‘warm up’ to the interviewer as the survey progresses, so they may answer sensitive questions more honestly later in the interview. The method by which the poll is undertaken can also affect responses. The increasing popularity of internet polling has come about largely because some people may be happier to give their true opinions and feelings to a machine than to a person – and are choosing to respond to the survey rather than being caught ‘on the hop’. So two reputable companies asking the same person the same question via telephone and email may receive two different answers.
A10 Today’s opinion polls are highly scientific and make use of proven techniques to eliminate bias. Whilst well conducted random and quota samples should present a broad cross section of society, there are many reasons why they may contain slightly too many of one social group. To manage this, polling companies will always ask respondents for detail about themselves as well as their opinions in the initial survey (for example, their type of employment) so that this information can be compared with the census statistics being used. The raw numbers from the poll are then adjusted to match the profile of the population being sampled. Opinion polls can also be reweighted to correct for bias. If, for example, a poll finds that it has 100 responses from one demographic group and it should have 110 to match other demographic groups in the study, each response can be ‘weighted’ so that they count for 1.1 people. However, excessive reweighting is avoided as this could otherwise impact on the accuracy of the results.
A11 It is possible – but they would soon get caught out, and would be subject to severe disciplinary action. Professional interviewers are trained to exclude bias from the questions they ask, and indeed from every stage of the research, including sample selection. They also frequently have their techniques assessed by a supervisor. The MRS Code of Conduct governs acceptable interview practice and all MRS members must abide by the Code. A researcher’s failure to abide by it may result in disciplinary action against them.
A12 An ‘exit’ poll records how votes have been cast, it does not anticipate how people will vote, as other types of opinion polls do. For an ‘exit’ poll a sample of voters is interviewed as they leave carefully selected polling stations during the day of an election. Usually, interviewers select the respondents by deciding to interview one in a pre-selected number of voters (for example, one in ten). To ensure the results are representative of the whole population, they are reweighted (mathematically adjusted) once the survey is complete. The record of the exit polls has been remarkably accurate in the last two elections.
A13 The findings from opinion polls are most commonly published in newspapers and are broadcast on television and radio shows, and on the internet. The media is often selective with the results they choose to publish, tending to use one chosen pollster for all their opinion figures, and it is impossible for a researcher to avoid editorial judgement. However, the full text of the questions asked by a pollster and methodological details must be available on request to any serious enquirer. Since the formation of the British Polling Council, and coinciding with the 2005 General Election, all pollsters have also agreed to make detailed data from their polls available on their websites or on request. As with all professional market research studies, the sample size, interviewing method and fieldwork dates of the poll must be published with the findings to enable the audience to gain a fair understanding of the findings.
A14 It has been widely demonstrated that the direct effects of opinion polls (such as changing people’s voting pattern or encouraging them to abstain from voting) are limited. By their very nature, opinion polls are designed to reflect the overall views of the population, not to alter them. However, some people may consider the findings of surveys in order to determine how to vote tactically.
A15 This is a hard argument to sustain given the high number and wide diversity of polls now conducted and published. While it is possible for clients to commission polls with a specific aim in mind, they are unable to specify the results. Moreover, with the popularity of opinion polling, it is unusual for any apparently controversial findings not to be cross-checked by other interested groups. Agenda setting is carried out by the media, not by pollsters. Pollsters are only able to set agendas when the media publish them.
A16 Opinion polls have a central role to play within the media, during the run-up to an election. They are often commissioned by newspapers who need statistics to back up their own reporting of events, and to support their political and editorial agenda. They also provide the public with the opportunity to express their views on the competing political parties and air their opinions on the issues that are really important to them before they have to cast a vote. In this sense opinion polls are central to modern election campaigns: they provide a context within which the battle is fought. Opinion polls have also become part of the armoury of political parties as they seek to gain momentum and drum up support before an election. They guide the parties as to how they are viewed by the public and the areas in which they are excelling or need to improve.
|
|
What's New - Membership - Company Partner Service - Members' Area - Code/Guidelines - Qualifications - Training - Awards © Copyright 2010 MRS - Privacy Statement - Terms and Conditions - Legal Information | ||
