Beruflich Dokumente
Kultur Dokumente
Introduction
Many authors have drawn attention to the increased use of qualitative techniques by commercial marketing
researchers over the last decade (Greenberg et al., 1977; Cooper and Branthwaite, 1977; Goodyear, 1978;
de Groot, 1986). These methods have been recommended for use in both consumer (Goodyear and Vineall,
1976) and more recently in industrial settings (Welch, 1985). Despite their importance, qualitative research
methods are frequently misunderstood and even ignored by students of marketing. Fahad (1986) stressed
this omission and advocated the use of one kind of qualitative research, namely, the focus group interview,
at least by the postgraduate researcher. This, however, seems over-cautious. The benefits obtained from
qualitative research methods render them worthy of consideration by all students engaged in research projects.
These techniques can be used to conduct a preliminary exploration of an under-researched area, to sort and
screen ideas as the project progresses, to explore complex behaviour or to experience a consumer's eye view
of the world. The accepted advantages of speed, flexibility and economy, together with the rich data generated,
make qualitative methods eminently suitable for student research at both undergraduate and postgraduate level.
Of all the techniques available, the focus group is probably the most familiar, having provoked a wealth of
comment from both commercial and academic researchers; as such, it offers a realistic starting-point for the
study of qualitative research.
Focus group interviews are a qualitative method in which a small sample of respondents discuss elected topics
as a group for approximately one to t w o hours. A moderator focuses the discussion onto relevant subjects
in a non-directive manner. The method is a development of the group discussion. This article presents a practical
guide to the procedures for conducting focus groups. A full discussion of the theory, role, uses, advantages
and disadvantages of qualitative research in general, and focus groups in particular, can be found in Drayton
and Tynan (1986).
There is a dearth of useful, practical information on how to run group discussions. The Market Research Society
Research and Development Sub-committee on Qualitative Research (1979, p. 116) reported that "there are
no widely agreed rules on how discussions should be conducted". Axelrod (1975a, p. 52) concurs: "There
are no prescribed guidelines for qualitative research, no books of rules, no formulas, and no strategems". A n
added complication is that the process of running a group discussion seems simple to the onlooker, ". . . to
the uninitiated, the amateur, or the outsider, group interviewing appears to be a simple casual process. . ."
(Payne, 1976, p. 69), yet to organise a group discussion yielding useful and usable data requires a thorough
understanding of the process and very careful planning.
Preparation point that the moderator must consider all the avenues
The first step in conducting qualitative research is to the group might discuss on a given subject, and how
discover "what information is being sought, and how he/she might present them in such a way that will not
insights derived will be used" (Axelrod, 1975b, p. 55). influence the results (Axelrod, 1975b). Hannah (1978,
The researcher, or moderator, as they are usually p. 77) suggests that the moderator and the client, i.e.
known, should then prepare a discussion or interview the student and supervisor, should review and discuss
guide which is essentially an outline of the key issues, the proposed guide thoroughly to ensure that they are
areas and questions to be covered (Payne, 1976). both "comfortable with the outline" before proceeding.
Merton and Kendall (1946, p. 541) see the interview
guide as "setting forth the major areas of inquiry and
the hypotheses which locate the pertinence of data to G r o u p Size MIP 6,1
be obtained in the interview". It is important that the The most widely recommended size for a group 1988
questions are open and non-committal, leaving the discussion is between eight and 12 respondents
respondents to discuss the issue without being biased (MacFarlane Smith, 1972; Bellenger et al., 1976; Cox 5
by the wording or presentation of the topic. It is at this et al., 1976; Qualitative Research, 1977; Prince, 1978;
Fern, 1982). For more difficult projects, or more Recruitment
sensitive subject matter, using fewer respondents is The recruitment of respondents is an issue which is
advised (Falconer, 1976; Mendes de Almeida, 1980). currently the subject of much debate between
However, it should be noted that for normal topics, too marketing research practitioners (Market Research
small a group may result in the loss of useful data. An Society Qualitative Research Study Group, 1985; Firth
experiment by Fern (1982, p. 12) proved "focus groups and Goffey, 1986; Willis, 1986). Many of the instances
of eight members generated significantly more ideas of bad practice quoted, stem from the use of
than focus groups of four members". At the other inadequately supervised recruitment agencies, a
extreme, 12 respondents are viewed as "a maximum situation which is unlikely to be paralleled in an
for decent moderation" (Mendes de Almeida, 1980, p. educational setting.
119). With more than 12 members, the group becomes
difficult to manage and may disintegrate into two or However, the recruitment of group members can be
even three small groups, each having their own difficult and expensive. To locate respondents who meet
the sample criteria quickly, efficiently and accurately,
independent discussion.
the Market Research Society R&D Sub-committee
(1979) and, more recently, Welch (1985) advise the use
of a recruitment questionnaire, whereas Wells (1974)
Sample Selection suggests recruiting by telephone or door to door, but
The relatively small number of respondents used in a only as a last (expensive) resort. All recruiting methods
focus group study makes it important that the sample should mask the precise subject matter of the
is properly selected (Market Research Society R&D Sub discussion, and provide demographic and "usership"
committee, 1979), but it should not be assumed that data which can be used in interpreting the research
the limited numbers in any one group preclude the use findings. Researchers should take steps to "minimise
of scientifically drawn samples. Normal practice is to exclusion of busy, impatient . . . hard to get people"
use brand or product usage characteristics and (Falconer, 1976, p. 21), and be aware of the "volunteer
demographic controls as a basis for selecting bias" indicated by Belson's research (1960, 1963).
respondents (Prince, 1978). One issue of the debate which "has not been
The issue of inadequate sample size is one of the adequately resolved", according to Fern (1982, p. 12),
criticisms most frequently levelled at focus group is the effect of acquaintanceship on the results of group
research. The most convincing case to refute this discussions. Mendes de Almeida (1980) indicated that
group discussions are vulnerable to any "group
criticism comes from Lipstein (1975, p.34). He used
cultures" that existed prior to the discussion because
data from the US Census of 1970 to prove that
they may enable members to bring prevalidated ideas
"increasing sample size is not a guarantee of increasing
to the debate. On this basis, he excludes the possibilities
accuracy. On the contrary, the sheer size makes many
of recruiting from churches, clubs or associations, but
kinds of errors highly likely". Errors that increase with
is not unduly worried about the possibility that a couple
sample size are called non-sampling errors. They
of the participants of a group may be acquainted. Kaden
include errors due to factors such as interviewer bias, (1977) is not perturbed about acquaintanceship among
mistakes in coding, respondent and interviewer fatigue. his group members, as he advises the use and re-use
Sampling theory states that sampling error decreases of the same respondents, claiming they become more
as a function of 1/n. Therefore, increasing the size of useful as they learn their art. This use of trained
a sample decreases the sampling error; but it also respondents has, more recently, been advocated in a
increases the non-sampling error. So, for any given British context by Schlackman (1984). The widely held
sample size, there is a trade-off between sampling and middle view is expressed by Wells (1974, p. 137). He
non-sampling errors. reports that "the assets and liabilities of allowing friends
to participate almost balance out".