Audience Dialogue

Know Your Audience: chapter 9
Visitor surveys

The term "visitor survey" - though not something you'll find in other books on research methods - is a useful way of referring to a survey of people who attend a specific event. Sometimes these are called "event surveys."

There are two main types of event:

Because of the different ways in which the audience come and go to these two types of event, two different survey approaches are needed.

1. Surveying performance events

The main problem with surveying a performance is to distribute questionnaires quickly enough, because most of the audience arrive and leave in a very short time. So everything must be ready beforehand:

Try to estimate the sample size you need (100 may be enough), allow for non-response (assume 50%, in the absence of other data), then calculate how many questionnaires must be printed and handed out. There may not be enough time to give questionnaires to all people entering or leaving the event. If tickets have been sold in advance, you can often get a fairly accurate estimate of the numbers attending the event. For example, if an audience of about 1,000 is expected, and 200 questionnaires have been printed, try to give every 5th person a questionnaire.

The goal is to hand out as many questionnaires as possible. If some of your staff run out of questionnaires, while others have plenty left, you’ve wasted some opportunities. When questionnaires run out, this is often in the last few minutes before the show starts, so there’s very little time to find out who has extra questionnaires, and to get these to somebody who has run out.

Prepare for this by having a staff member who will hold about 20% of the questionnaires back till the last 10-15 minutes, then rush around the other staff, distributing questionnaires as needed.

People who fill in the questionnaire while they are still at the venue will vastly outnumber those who take it home and post it in later. It seems that a lot of people fully intend to send in the questionnaire later, but never get around to it.

Therefore, find some large postboxes, each about a metre high, with a questionnaire-sized slot in the top. For example, we have used cardboard cartons that filing cabinets were delivered in. Paint the boxes bright colours, so that they stand out in a crowd of people. Put one box at each side of each exit from the venue, so that everybody who has filled in a questionnaire will pass a box on the way out, and be able to drop their questionnaire into it. Put signs above each postbox, too, so that everybody can see where to put their questionnaires, even if the box is surrounded by people.

Using these methods, we have achieved response rates around 65% to 70%, in a wide variety of events. There seem to be four key factors in getting a good response rate:

1. Questionnaires must be personally handed out. In the few seconds available, the interviewer should try to convey to each respondent that their opinion is very important.

2. Try to get people to fill in questionnaires while they are still at the event - but make it easy for them to return the questionnaires later. A business reply envelope, printed with your address) works well. The questionnaire should also state a cut-off date, about a week ahead. If the cut-off is too soon, some people won’t bother returning the questionnaire, but if it’s too far in the future, they’ll forget. One week is about right, I’ve found - but questionnaires will still arrive for another week after that.

3. Offer a prize for some respondents: because they are all at a particular venue, tickets to another show at the venue are often a good choice. The prize needn’t be large to be effective, but you should make it clear that everybody who returns a completed questionnaire (by the cut-off date) is equally eligible. If people think that giving certain answers will give them a better chance of winning the prize, this will distort the results.

4. The questionnaire must be short, interesting, and easy to complete. Follow the principles laid out in the chapter on mail survey questionnaires. Always offer people the chance to give their opinion of the event they have come to. Even if that’s not your main interest in the survey, it will be the main interest for most of the audience.

A good format for an event questionnaire is to print it on a piece of A4 paper, folded down the middle to make a 4-page A5-size booklet. Inside the fold you can put a business reply envelope, so that it won’t easily be lost.

Though this is a small questionnaire, people don’t usually have a large flat surface available at most arts events. A small questionnaire is easier to write on legibly. Nor have people come to the event to fill in a long, complex questionnaire. On a 4-page A5 questionnaire, you can fit about 12 main questions - that’s plenty.

Don’t make the print or the answer boxes too small. At many theatres, the lights are dim, and older people find small print difficult to read. About 12 points (approximately this size) is the practical minimum.

2. Continuing events

This type of event (such as an art exhibition) is easier to survey than a performance event. There is no sudden rush of patrons in and out, and this means that the people who hand out questionnaires have to stay around for much longer.

Therefore, at continuing events, it is often more efficient to do personal interviews. These are generally easier for respondents, and you get better-completed questionnaires. However, this requires interviewer training, as described in the Interviewing chapter of this book. Another possibility is to interview some audience members, and also hand out questionnaires to people who are willing to fill in their own.

Interviewers may not need to be present for the whole time the event is open. If so, the interviewing times should be carefully selected so that no type of visitor is excluded from the survey. In an exhibition continuing for several weeks, interviews should be approximately evenly spread between the beginning, middle, and end period - and also distributed roughly in proportion to attendance between normal business hours and weekend and evening times. The goal is that anybody attending the exhibition should have an equal chance of being interviewed.

Continuing events, with many visitors and many entry points

If people must buy a ticket to attend the event, there will be a fixed number of entry points, and interviewers can be stationed just inside each entry point.

But if entry is free, the area may not be enclosed at all, and it will be much more difficult to keep track of the people entering. In this case, self-completion questionnaires are not a good idea. The response rate will be low, specially if the exit points are very wide, or there are too many to put postboxes at them all. In this situation, you will usually obtain a more accurate sample by having the interviewers walk around the whole area of the event, interviewing at as many different places as possible.

Some events - such as farm shows - often have a lot of exhibition stalls. Sometimes there are as many people working on the stalls as there are visitors. You need to decide on your sampled population: will it be all people who are present, or all visitors? If you include people working on stalls, it’s usually easier to interview them by adapting the principles of door-to-door interviewing (as explained in the chapter on interviewing), pretending that each stall is a house. The visitors can be interviewed separately. The opinions of the two sets of people are often quite different, and need to be considered separately. Therefore each questionnaire should show whether the interview was with a visitor or a stall-worker; this can be done by using two different coloured questionnaires.

At this type of event - as with any survey done in a public place - it’s very difficult to get a representative sample. You can’t even do an accurate quota sample, because the population of people visiting the show may not be representative of the whole population - so you can’t assume that (say) 50% of interviews should be with men and 50% with women.

Interviewers in these situations tend to approach people who are alone, but not those who are talking to friends. If the only people interviewed are those attending alone, this is not a good sample: usually, people who go to an event by themselves are much more interested in it than those who go with others. So interviewing too many people who are alone may produce too favourable a result.

A good solution is to do the research in two stages. In the first stage, nobody is interviewed, but the interviewers record whatever information they can about each person who passes them. For example, they should count 100 people: how many are men, and how many are women? Then count another 100, and divide them into say three broad age groups: perhaps under 20, 20 to 50, and over 50. With this 100, count the number in each age group. Then count a third hundred: how many are alone, and how many are with others?

If the area is a large one, counts should be made by several interviewers, in different parts of the area. Now the interviewers should meet, and their counts can be added together. You will now have an estimate of the kinds of people attending the show, and enough information to work out the quotas. If 50% are men, and 40% are aged 20 to 50, and 30% are alone, then 50% x 40% x 30% (or 6%) of the interviews should be with men aged 20 to 50 who are alone. Though multiplying the three percentages together is not as accurate as counting the three variables separately for each person, there’s often no time to do that.

With two sexes, three age groups, and two categories of company (alone, or with others), that’s 12 categories of person. Each interviewer can then be given some quotas, based on the numbers already counted, and sent off to do that number of interviews with each type of person.

Continuing events with few visitors, indoors

For some types of event, the above approach isn’t appropriate: for example, an art exhibition, in a single room, with only a few visitors at a time. If the interview can be overheard by the other visitors, the respondent won’t be at ease.

With a small number of visitors, the interviewer can count them and ask (say) every third one to fill in a questionnaire: either as they arrive, or as they leave - it depends on the purpose of the survey.

The goal is to have the respondent fill in the questionnaire on the spot, but without being stood over by the interviewer. In a situation like this, it’s best to have a space available where respondents can fill in the questionnaire - perhaps a counter they can stand at. You can use a large postbox (as described above), in which completed questionnaires can be dropped.

With this type of setting, there’s no need to offer any prize, but the interviewer by his or her manner should make it clear that the respondent’s opinion will be very much appreciated. As with the performance event surveys, a self-completion questionnaire for a continuing event survey should be short and simple - but if there’s a place where respondents can fill in the questionnaire, it needn’t be as small as A5 size. A4 is easier for coding and data entry, and gives more space for comments.

If the questionnaire looks too long, or asks difficult questions, many people won’t bother to complete it. I suggest a maximum of one A3 piece of paper, folded to form a 4-page A4 booklet, with generous margins, and reasonably large and well-spaced type. The average respondent should take no more than about 5 minutes to fill it in: that’s a maximum of about 15 questions. For most purposes, that’s plenty.

If most people at a venue have an interest in the organization sponsoring the survey, a good incentive to increase response rates is a promise to notify people of the survey results. It’s very expensive to notify all respondents individually, by mail, but other alternatives exist: