There is no interviewer in mail surveys to explain the purpose and to induce cooperation, to ask the questions, to record the answers and in general to cope with any problems. This is the main difference between mail and the other two methods. It puts a great deal of importance on the construction of the questionnaire ad any transmittal letter that may accompany it.
Mail surveys are particularly versatile in reaching all types of people, in all geographic areas, in rural or urban areas at the same cost. On the other hand, while mail may make it easy to reach a particular group, and only that group, this is true only where a mailing list is available. Members of specific organizations or all licensed automobile owners are examples where such lists are available. The purchasers of a particular product can be reached by attaching questionnaire to the product package, and subscribers to particular magazine or newspapers can be reached by including questionnaires in the publications. Mail cannot be used to conduct an unstructured study where an interviewer is relied on to improvise the questioning as each interview progresses.
Information Obtained: Many commercial research organizations operate on the rule that questionnaires longer than six pages should not be used to collect data by mail. This seems to fit with common sense, but a number of experiments suggest that longer questionnaires can be used effectively by mail. Questionnaires as long as 32 pages or containing 1,100 questions have been used successfully. The manner in which the proposition is presented and the appearance of the first page may have more to do with completing the questionnaire than length.
From a specific respondent, a mail questionnaire probably produces as good or better information than a personal or telephone interview where an interviewer can bias the results The Bureau of the Census has shifted from the use of personal interviews to mail surveys because it believes they provide more reliable answers supplied directly by respondents instead of through more or less inhibiting intermediary, the enumerator.
Mail questionnaires have been shown to be generally superior to either telephone or personal interviews in collecting data on topics that might be embarrassing. Data from one of the best studies of this problem are shown below. Even though the questions asked appear to be only, mildly embarrassing at most, the differences in answers obtained by mail and by personal interview are large.
Responses to the Same Questions by Mail and Personal Interview
Subject Personal Interview Mail
Have used hair rinse 37% 51%
Have used eye shadow 46 59
Have purchased margarine 75 82
Have borrowed money at regular bank 17 42
Have borrowed money at credit union 16 22
Have borrowed money at small loan company 11 13
Sample size 200 100
Sequence bias is a threat in mail surveys – respondents can change their answers after seeing later questions. This bias may not be as great as it is usually thought to be. In an experiment to determine the extent of sequence bias, a questionnaire was loaded with references to a particular brand of gasoline before a question as to brand of gasoline used. Only a 6 percent increase in the proportion reporting use of the specified brand occurred with the mail questionnaire.
In answering open ended questions that is, questions that respondents must answer in their own words – respondents tend to be more brief and more general in mail surveys than in personal or telephone interviews. Complex questions with rating scales or other procedures that can be confusing tend to draw more ‘no answers’ in mail surveys.