Avoid Ambiguous Questions: Ambiguous questions mean different things to different people. Naturally, comparable replies cannot be received from respondents who take a question to mean different things. The question how often do you serve soup at home? Obtained answers that showed soup was served less frequently than was believed to be the case. Further study showed that too many respondents served meant a special occasion such as when entertaining. Soup may have been eaten when the family was alone, but it was not served. A better wording turned out to be; how often do you use soup at home?
Questions that use such terms as usually, normally, frequently and regularly are ambiguous: What brand of cigarettes do you smoke regularly? What does regularly mean – always almost always, more than any other, or what? Some respondents will interpret it one way and some another. When the researcher wants to find out typical behavior, it is probably best to inquire about a specific time, such as the last time. What brand was the last cigarette you smoked? The last brand smoked by some respondents will be different from the one they usually smoke, but if the sample is adequate this will average out.
Avoid Leading Questions: In a study to evaluate the service of automobile insurance companies, a series of question on claim service was preceded by a statement that began as follows: It has been alleged that some low rate companies are much tougher in adjusting claims than standard rate companies, and that you are more likely to have to go to court to collect the sum due you. It is obvious that this statement would influence the answers to the questions on claim service that followed. Most marketing research studies do not have the obvious bias of the above, but it is easy to lead respondents toward one answer unless care is taken not to do so.
In one experiment half of the respondents were asked: How long was the movie? and the other half was asked: How short was the movie? The average response to the first was 130 minutes while to the second wording it was 100 minutes. A better statement of the question would be: For how many minutes did the movie run?
Do you have a Frigidaire refrigerator? This question will result in more reports of Frigidaire refrigerators than will a question: What brand of refrigerator do you have? When respondent get an idea that a survey is being made for a particular company or product, they have a tendency to respond favorably toward the sponsor. In a mail survey among FM radio owners in the area covered by a small FM radio station, respondents were asked to list the four FM stations to which they listed most. The sponsoring station came out on top by far. A study of the individual questionnaires, however, showed that many of the respondents had associated the survey with that station, apparently through the address to which the questionnaire was mailed. Many of the returned questionnaires had additional comments such as ‘We think you’re wonderful and we like your station but not quite as well as X’.
Do not ask questions in a way that will involve Generalization: Questions should always be stated in specific terms. If generalization is desired the researcher should make them from the specific data obtained. Consider the following question: How many machine tool sales representatives have called on you in the last year? The only way the respondents could answer this question would be to estimate about how many sales representatives came in during a typical week or possibly a month and then to multiple.