Collection and Evaluation of Interview Data

    Administering
interview schedules and questionnaires requires different skills and involves
different considerations. In this section, we examine issues in the
administration of structured instruments, and ways of handling difficulties.

Collecting
Interview Data

    The
quality of interview data depends heavily on interviewer proficiency.
Interviewers for large survey organizations receive extensive general training
in addition to specific training for individual studies. Although we cannot in
this introductory book cover all the principles of good interviewing, we can
identify some major issues. 

    A primary task of interviewers is to put
respondents at ease so that they will feel comfortable in expressing opinions
honestly. Respondents’ personal reactions to interviewers can affect their
willingness to participate. Interviewers, therefore, should always be punctual
(if an appointment has been made), courteous, and friendly. Interviewers should
strive to appear unbiased and to create a permissive atmosphere that encourages
candor. 

    All opinions of respondents should be accepted as natural; Interviewers
should not express surprise, disapproval, or even approval. When a structured
interview schedule is being used, interviewers should follow question wording
precisely. Similarly, interviewers should not offer spontaneous explanations of
what questions mean. 

    Repetitions of the questions are usually adequate to
dispel misunderstandings, particularly if the instrument has been properly pretested.
Interviewers should not read questions mechanically. 

    A natural, conversational
tone is essential in building rapport with respondents, and this tone is
impossible to achieve if interviewers are not thoroughly familiar with the
questions. When closed-ended questions have lengthy or complicated response
alternatives, or when a series of questions has the same response alternatives,
interviewers should hand subjects a show card that lists response options. 

    People cannot be expected to remember detailed unfamiliar material and may
choose the last alternative if they cannot recall earlier ones. Closed-ended
items are recorded by checking or circling the appropriate alternative, but
responses to open-ended questions must be recorded in full. Interviewers should
not paraphrase or summarize respondents’ replies. 

    Obtaining complete, relevant
responses to open-ended questions is not always an easy matter. Respondents may
reply to seemingly straightforward questions with irrelevant remarks or partial
answers. Some may say, “I don’t know” to avoid giving their opinions on
sensitive topics, or to stall while they think about the question. In such
cases, the interviewers’ job is to probe. 

    The purpose of a probe is to elicit
more useful information than respondents volunteered during their initial
reply. A probe can take many forms: Sometimes it involves a repetition of the
original question, and sometimes it is a long pause intended to communicate to
respondents that they should continue. 

    Frequently, it is necessary to encourage
a more complete response by a nondirective supplementary question, such as,
“How is that?” Interviewers must be careful to use only neutral probes that do
not influence the content of a response.  The ability to probe well is perhaps the greatest test
of an interviewer’s skill. 

    To know when to probe and how to select the best
probes, interviewers must fully comprehend the purpose of each question and the
type of information being sought. Guidelines for telephone interviews are
essentially the same as those for face-to-face interviews, but additional
effort is usually required to build rapport over the telephone. 

    In both cases,
interviewers should strive to make the interview a pleasant and satisfying
experience in which respondents are made to understand that the information
they are providing is important. Collecting Questionnaire Data Questionnaires
can be distributed in various ways, including personal distribution, through
the mail, and over the Internet. 

    The most convenient procedure is to distribute
questionnaires to a group of people who complete the instrument together at the
same time. This approach has the obvious advantages of maximizing the number of
completed questionnaires and allowing researchers to clarify any possible
misunderstandings. 

    Group administrations are often possible in educational
settings and may be feasible in some clinical situations. Personal presentation
of questionnaires to individual respondents is another alternative. Personal
contact with respondents has a positive effect on response rates for SAQs. 

    Furthermore, researchers can help explain or clarify particular items or the
study purpose. Personal involvement may be relatively time-consuming and
expensive if questionnaires have to be delivered and picked up at respondents’
homes. The distribution of questionnaires in clinical settings, on the other
hand, is often inexpensive and efficient and likely to yield a high rate of
completed questionnaires. 

    Questionnaires are often mailed to respondents, but
this approach tends to yield low response rates. When only a subsample of
respondents return their questionnaires, it may be unreasonable to assume that
those who responded were typical of the overall sample. That is, researchers
are faced with the possibility that people who did not complete a questionnaire
would have answered questions differently from those who did return it. 

    If the
response rate is high, the risk of nonresponse bias may be negligible. A
response rate greater than 65% is probably sufficient for most purposes, but
lower response rates are common. Researchers should attempt to discover how
representative respondents are, relative to the selected sample, in terms of
basic demographic characteristics, such as age, gender, and marital status. 

    This comparison may lead researchers to conclude that respondents and
non respondents are sufficiently similar. When demographic differences are
found, investigators can make inferences about the direction of the biases. 

    Response rates can be affected by the manner in which the questionnaires are
designed and mailed. The physical appearance of the questionnaire can influence
its appeal, so some thought should be given to the layout, quality and color of
paper, method of reproduction, and typographic quality of the instrument. The
standard procedure for distributing mailed questionnaires is to include a
stamped, addressed return envelope. 

    Failure to enclose a return envelope can
have a serious effect on response rates. The use of follow-up reminders is
effective in achieving higher response rates for mailed (and Internet)
questionnaires. This procedure involves additional mailings urging
non respondents to complete and return their forms. Follow-up reminders are
typically sent about 10 to 14 days after the initial mailing. Sometimes
reminders simply involve a letter of encouragement to non respondents. 

    It is
preferable, however, to enclose a second copy of the questionnaire with the
reminder letter because many non respondents will have misplaced the original or
thrown it away. Telephone follow-ups can be even more successful, but are
costly and time-consuming. With anonymous questionnaires, researchers may be
unable to distinguish between respondents and non respondents for the purpose of
sending follow up letters. 

    In such a situation, the simplest procedure is to
send out a follow-up letter to the entire sample, thanking those who have already
answered and asking others to cooperate.