Audience Dialogue

Issue-based polls for broadcasters

Broadcasters often make "vox pop" interviews with members of the public, to discover public opinion on specific issues. This page shows how to go beyond vox pop, by carrying out small surveys that can provide useful information to public decision-makers.

To carry out a successful survey, you need three things: a representative sample, a clear set of questions, and a method of analysis. Let's cover each of these in turn.

1. A representative sample

It's important for a survey to represent all the types of people who might be affect by an issue. A simple way to do this is to use quotas, and select your sample from a wide variety of sources. Don't rely only on street interviews, or only on telephone calls to people at home. The more methods you use, the harder it will be to challenge the results.

About the minimum sample size for a survey to have credible results is 100 people. If you use quotas, you could contact 25 by phone, and 75 in person: 25 at home, 25 at work, 25 in public places. As a further safeguard, you could try to make sure that close to half of those surveyed in each place are men and half are women. For many issues, it's not necessary to survey children, so people aged under 15 can be excluded. In developing countries, about half the people over 15 are aged 15 to 34, and the other half are aged 35 and above. The sample should reflect this.

If the questionnaire has about 10 questions, it will take 5 to 10 minutes per interview. Because not everybody agrees to be interviewed, and because the interviewer has to move around, about a third of the time is not spent doing interviews. That means you can do about 4 interviews per hour (if they average 10 minutes each) and about 8 (at 5 minutes each). Usually, two interviewers can do 100 short interviews in one day.

See this page on our quota surveys in Indonesia. The method worked very well, and could easily be adapted for use in other countries.

2. A questionnaire

To begin with, describe the proposal or issue, in a a few sentences - perhaps 20 to 30 words. If the description is too short, some people won't fully understand what you are asking about. If it's too long, some people will get bored, and not pay full attention. 10 words is usually not enough, and 50 is usually too much. When respondents understand the proposal, you can ask questions like this...

1. "Have you already heard about this proposal?"
[] Yes
[] No
[] Not sure

2. "What's your opinion: are you in favour of it, against it, or not sure?"
[] In favour
[] Opposed
[] Not sure

3. "How strongly do you feel about that?"
[] Very strongly
[] Not so strongly
[] Not strongly at all
[] Not sure

4. "If you agree or disagree, why is that?"
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5. "What do you see as the advantages of the proposal?"
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6. "What do you see as the disadvantages of the proposal?"
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

7a. "What types of people do you think would benefit if this went ahead?"
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7b. "How would they benefit?"
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

8a. "What types of people do you think would be worse off if this went ahead?"
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8b. "How would they be disadvantaged?"
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

9. "Do you think it will go ahead anyway?"
[] Yes
[] No
[] Not sure

10. "Is there anything else you'd like to say about this proposal?"
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

11. "Which age group are you in...
[] ...under 25, or
[] 25 to 44, or
[] 45 or over?"

12. Record respondent's sex
[] Male
[] Female

The best questionnaires mix multiple-response questions and open-ended questions. The counted answers from the multiple-response questions indicate the extent of support or opposition to the proposal, while the verbatim answers to the open-ended questions help you understand why people feel as they do - and give some details that can be reported on air.

Verbatim answers are often similar to each other, and can easily be grouped - effectively transforming an open-ended question into a multiple-response question.

Demographic questions

The questions on age group and sex are known as demographic. The answers are not of great interest in themselves (50% of people were male and 50% female? Big deal!) but useful for checking the accuracy of the sample, and also for seeing whether different types of people have different opinions on the central issue.

You don't need many demographic questions, but choose some that are relevant to the issue you're studying. If it's related to a particular location, you could ask what area respondents live in. If the issue is work-related, you could ask respondents their occupations. Other demographic questions can cover education level, family status, religion, etc.

Keep the questionnaire short. It's tempting to ask more questions, but long questionnaires mean that the survey takes longer to do, costs more, and takes longer to produce results. A good principle is to limit the questionnaire to one page: this makes printing and analysis much easier.

When the questionnaire is finished, and has been tried out on a few people, you could print 100 copies. In fact, some copies will be wasted: a survey fo 100 people usually needs about 120 questionnaires.

3. A method of analysis

With only 100 questionnaires, it's not hard to count the results by hand - specially if the questionnaire takes only one page. It's also possible to enter the answers on a computer. If you have access to a spreadsheet program such as Excel, you can use that. The time taken up by typing in the answers is more than saved in the tabulation.

Relevant pages on this website

by Dennis List - updated 22 July 2003