6 Approaches to Interviewing
It has already been shown that, despite their common features, survey and
qualitative interviewing serve different purposes and have different character-
istics. This chapter develops those differences as more detailed consideration is
given to the different approaches to interviewing. The principal argument is that
there are no hard-and-fast rules about which approach to use, nor is any approach
a prescription for practice: the choice of approach and the way any approach is
worked out in a particular research setting is very much a matter of designingthat
which is fit for the research purpose. That said, ideas about survey interviewing
tend to be more fixed and more prescriptive than with qualitative interviewing,
where practices are much more flexible.
In this chapter,q e explore some of the ways in which you canvary the standard
approach of having one interviewer in a private setting with one informant. It is
possible to have more than one interviewer and to interview more than oneperson
at a t i m ~ , ~ p Q ointerview face-to-face, or over the phone. We conclude this
chapterwith a fairly detailed description of one interview-based study. This draws
together many of the themes that have run through these first six chapters.
One-to-one interviews and variations on the theme
More than one interviewer
It can be very useful to have two researchers present during an interview. For
example, when interviewing groups of North American students about whether
they would choose to do postgraduate study in the UK, Peter found it helpful to
have a colleague in the room. Their roles were explained to the interviewees. At
any one time, one would chair the discussion while the other would keep notes,
referring to the counter on one of the tape recorders that was running, but saying
nothing.Advantages included:
The silent researcher could notice things of interest that were missed by the
chairing researcher, whose attention was often held by the management of
group dynamics. At the end of the interview each interviewer had a chance to
raise points that had struck him while observing.
The researchers had a good account of the interviews and a guide to where the
most interesting quotations were long before the tapes could be transcribed.
Having two views of what had happened helped to clarify key themes and
areas for enquiry and analysis.
Approaches to interviewing 75
In this case the researchers were 'on the road', going fromcity to city,hotel to
hotel, doing four sets of interviews in four days, and had to give aprelinzinary
report to sponsors on the fifth. It was only possible to work to this schedule
because two interviewers were used.
More than one intewiewee
Joint interviews Joint interviewing involves one researcher speaking with two
people simultaneously to gain both perspectives on the same phenomenon. It is
an approach that tends to be used in work primarily of a qualitativenature rather
than in structured surveys. Family life is one area of enquiry where joint inter-
views can be used to advantage (Allan, 1980), suggesting that participants
will generally be people in marital relationships or living as couples. Quite often,
joint interviews are used as part of a triangulated study (see Chapter 2), involv-
ing separate conversations with the individuals concerned either before or after
the joint interview, and with interviews repeated over a long period of time
(Radley, 1988; Backett, 1990).
In Box 6.1, we detail some of the potential advantages and disadvantages of
conducting planned joint interviews, which are quite different from unplanned
joint interviews (see Chapter 5).
There are various practical difficulties with a joint interview research design.
First, this sort of technique is associated with a low response rate. In two studies
involving family finance-admittedly a sensitiveissue-men were less likelythan
women to be willing to participate (Pahl, 1989; Jordan et al., 1992). Secondly,
organizingjoint interviews can be a difficult and drawn-out process, particularly
if both partners work. Accommodating two people's schedules may mean that
joint interviews are more likely to be arranged for evenings or weekends.Finally,
there may be financial implications. The areas where increased expenditure may
be expected include the training and time commitments of the interviewer(s),
fieldworkactivities, transcription and coding, and data analysisand interpretation.
Unfortunately, in small-scale studies with limited time and resources available,
implementing a joint interview approach may not be appropriate, even if it does
appear to be particularly suited to the research questions.
lntewiewing an intact social group There can be interviews with groups as well as
with individuals. Group interviews can show something of the dynamicsof social
relationships amongst group members - for example, who gets to speak and
who does not, what forms of speech characterize different members in the group
setting, and whose ideas are listened to. When the group is a naturally occurring
social group, then things can be discovered by talking with all members together
that could not be gleaned through one-to-one interviews. The researcher
may observe the processes of consensus formation and of the rules by which
disagreements are played out.
Bruner (1990) describes a method in which family members were individually
interviewed about their lives. As is often, but not invariably the case, the
informants 'later remarked spontaneously that they had enjoyed the interview

76 Interviewing for sociaI scientists
,' Box 6.1 Advantages and disadvantages o fjoint interviews
.
(Developed from Arksey, 1996)
Advslntagq
May establish rapport and an atmosphere of confidence more easily.
Can obtain two versions of events rather than one, which may, or may
not, produce a coherentjoint account. The distinct forms of information
andknowledge are likely to corroborate and supplement each other,but
at the same time may contain points of divergence. Inconsistencies
betweenperspectives are llkely to be missed if one partner is left out of
the study, and his or her views are inferred from the other's data.
The story that emerges may be more complete as interviewees fill in
each other's gaps and memory lapses.
The information obtained may be more trustworthy as bias in one
account may counterbalancethat in the other.
Researchers may gain insights into the interactions and nature of
(power) relationships between couples through observation of verbal
and non-verbal modes of communication; for example, it may be
possible to witness how couples support, negotiate and influence each
other, as well as manage disagreements and areas of tension.
Disadvantages
One informant may dominate,to the extent of silencing the partner; the
literature (McKee and O'Brien, 1983;Jordan et al., 1992)suggests that
men are likely to be the more vocal and overbearing.
The risk of stirring up antagonisms and conflicts of interest.
If the research topic is especially sensitive, or there is any likelihoodof
provoking friction, individuals may not be willing to disclose detailed,
honest information in front of their partner and instead provide a more
acceptable, 'public' response (Cornwell, 1984).
Partners may collude to withhold information from the interviewer.
Interviewees may not concentrate as well when two people are present.
andlor that they had found it personally informative. Several said that they had
been quite surprised by what came out' (p. 125). These interviews were then
followed by a meeting of the whole family where they heard the researchers'
account of what it was like to grow up in that family. The subsequent discussion
was still going strong after three hours.
This use of a group interview had considerable advantages in terms of the
research interest, which had much to do with the interplay of the individual and
Approaches to interviewing 77
the family unit. The small size of the family group (six members) and their known
and distinctive voices allowed transcripts to be made that identifiedwho had said
what. It also helped to validate the individual interviews, as well as highlighting
problem areas in them. It is regrettably uncommon for researchers to build on
theoretical insights into the interplay between individuals and social units by
using both individual and group interviews.
Focus group interviews A focus group is a selection of people who are invited to
respond to researchers' questions, findings from earlier studies, policy docu-
ments, hypotheses, concerns and the like. They may comprise people who are a
cross-section of the population, or they may be homogeneous, comprising, say,
retired women, 16-18-year-olds, or clients of the probation service.
Focus groups originated in market research. Suppose that a firm wished to
advertise a new product: which features should be stressed, how might it be
packaged and what price might be charged? Some purchase can be had on these
problems by assembling small groups of consumers and asking them, for
example, what they looked for in buying a new car. In this case, it is probable that
they would be shown mock-ups of different cars, descriptionsor drawings,and be
asked to say which they preferred and why.
The results of these market research focus groups would not prove anything,
especially if participants were paid for their views. The number involved would
be small and the generalizability would be quite low. There would also be
problems because of group dynamics, where dominant individualsmight obliter-
ate alternative points of view. Again, perceived status differences may lead some
group members to dissemble and conceal views. An example of this from Canada
is a focus group that was discussing road safety, specificallyrelated to motorists'
behaviour in areas patrolled by pedestrian crossing attendants (who were being
injured at an unacceptable rate). As focus group members introducedthemselves,
it emerged that one man was an officer in the Royal Canadian Mounted Police
('Mountie'). At first, people said that they conformed to speed limits and road
signs. Only when the RCMP man said that he regularly drove his private car 10
kph faster than the speed limit, even though he knew he should not, did others
begin to talk more openly.
Despite the drawbacks, focus group interviews are very useful for some
purposes. For example, researchers need to be sure that their instruments do
explore the way people feel, think, and say they act, and not simply test out the
researchers' view of how these things ought to be. Focus groups are a low cost
way of getting the range of the informants' perspectives and of getting some,
tentative purchase on who holds them. As long as the results of these groups are
treated as material for thought, reflection and hrther investigation, they can be
very useful and relatively inexpensive. They complement but do not remove the
need to pilot and refine research instruments.
One of us has used focus groups to understand the way students choose jobs
andpostgraduate courses (Hesketh and Knight, 1997).At first sight,this approach
is riddled with flaws. Groups can be harder to manage than individuals, so the
discussion can loop around on itself and some issues can get quite brief attention.

-.. .- . \
78 Interviewing for social scientists
Groups can be large (unexpectedly, one of ours had 25 people, was not easy to
manage, and produced a low quality tape recording). It would be more usual to
think of focus groups of about seven people, and even then it is important to have
two, good-quality tape recorders with separate stereo microphones carefully
placed to ensure usable recordings will be made. Even with good acoustics, good
equipment and manageable groups, it can still be impossible to ascribe views to
individuals, and some people simply say little. There is no scope to ask for a show
of hands to see how many people agree with any particular point, since that would
disrupt the flow of conversation and put the researcher in a far more directive role
than is consistent with hearing as many ideas as possible. Despite such problems,
the Hesketh and Knight study heard a consistent and very important message that
people applying for postgraduate courses were not getting the sort of information
from higher education institutions that they wanted. That was very helpful to
information providers who were able to think better about providing material to
potential postgraduate students.
Focus groups are also used to validate research reports. In Chapter 11 we
describe some of the problems of analysing interview data. It will be seen that this
is not an objective, judgement-free process, which means that the interpretations
that come out of the analysis might make sense to the researcher but not to the
research subjects. One way of reducing this risk is to validate the findings by
giving participants a summary of them. Where this can be followed by focus
groups composed of participants, the researchers have a better chance to hear how
well the interpretation fits their understandings. It also provides a chance for the
researchers to seek explanations of unexpected findings and to clarify details.
Fuller discussion of issues in focus group research is provided by Morgan
(1988), Krueger (1994), Greenbaum (1998) and Wilkinson (1998). The set of six
books in the Focus Group Kit (edited by Krueger and Morgan, 1998)provides a
comprehensive treatment in a practical and accessible way.
One-to-one interviews
Most interviews are conducted one-to-one. To illustrate the range of approaches
to one-to-one interviewing, we review here telephone interviewing, as an example
of survey interviewing, and oral history interviews to give a sense of less-focused
alternatives.
Telephone interviewing Surveysproduce data that are far more closely related to the
quality of the questions that are asked than is the case with exploratory interviews,
where rich data are often defined as those coming from people sharing easily what
is foremost in their thinking. There, the exact wording of the questions is less
important, given the great flexibility of qualitative approaches.
At first sight, self-administered questionnaires should be the preferred survey
method, and it is hard to see why anyone would go to the expense of paying
people to read out scripts that could be sent out as questionnaires. Yet, survey
interviewihg is often the method of choice in market research. We examine
telephone interviewing to illustrate the case for survey interviewing.
C...,'
Approaches to interviewing 79
Once, telephone interviews were treated with distrust, rather as internet-based
research is now (Selwyn and Robson, 1998). Those with a phone were better off
than others, so telephone surveys provided information about the attitudes and
beliefs of more affluent people only. Now, telephone ownership is widespread,
although the very poorest are still excluded, more so in the UK than in the USA,
and telephone interviewing is now a major method in survey research in market-
ing, although it is still, in the opinion of Frey (1989), not sufficientlywell known
as a social science research technique.
In many ways, telephone interviewing does not feel like interviewing. The
respondent cannot be seen and the visual cues that are so important in establishing
an interviewing relationship are lost. The rapport depends on what is said and on
the voice manner of the interviewer. For this reason, careful attention has to be
paid to the introductory patter and to the interviewing schedule, and interviewers
need to be chosen for the quality of their phone manner. Despite these
impediments, Lavrakas (1987) hopes to gain a 90 per cent response rate from
those of the target group who can be contacted.
Telephone interviewing has several advantages:
It is well-suited to random and structured sampling, far more so than
clipboarding (which is also known as 'intercept interviewing': where people
are stopped in malls, going to football games, leaving churches, and so on. The
major problem is that the people at those places are only representative of
people who go to such places at such times).
Telephone interviewing is ahead of its main rival, the questionnaire, because it
is quicker.
- Telephone interviewing usually has higher response rates than do
questionnaires, especially where people have had a letter saying that they will
be called and outlining the purpose of the coming call. (It also means that
people can have to hand any files they might need when they are telephoned.)
The interviewer can help respondents who have difficultieswith any question,
which is not true of questionnaires.
Literacy, which is necessary for questionnaire response, is not a limiting factor
in telephone interviews.
* The conventions of phone use work for the interviewer, since people feel a
pressure to answer the phone (but not to respond to an intercept interviewer), it
is customary for the initiator to terminate the call, and there is an expectation
that the person answering the phone will then participate actively in the
conversation (Frey, 1989).
* It has the advantage over questionnaires that the interviewer can encourage
reluctant phone subscribers to participate. Ways of doing that include having
a good 'patter', by stressing how helpful it would be to have their opinions,
by saying that the interviewer will call back at a more convenient time,
by saying that the interview will be brief, by pointing out the value of having
more information on an important issue, or by saying that cooperation helps
the interviewer to earn a living (Lavrakas, 1987). Box 6.2 illustrates some of
the things that the interviewer might say in these circumstances.

80 Intewiewing for socia/ scientists
Box 6.2 Examples of explanationsthat phone interviewers give
(This material relates to the issue of informed consent, covered in
Chapter 9)
The purpose of the survey
This is a short survey b v e a length of time here].The questions are about
your attitude towards [add name of topic].We want to know how you feel
about this because it will help us to understand what people think. That will
help us to get a better idea of [whatto do/how to inzprove things/what the
firnz, government etc. could concentrate on in thefuture]. I'm doing this as
a part of my work at [name of institution] and, as well as being useful to
[namethe users of the$ndings] it is a part of my coursework. Whatever you
tell me will be confidential. Your cooperation is voluntary but it would be
very helpful if you would answer these questions for me.
How will the survey be used?
I shall write up a report for the course tutor. I will not use anyone's names
in the report, or say anything that could identify anyone. A summary of the
findings will also go to [addname of user].The idea is that these findings
should help us to understand [insertname of the topic]better so that better
decisions can he madebetter action can be taken. Your cooperation is
voluntary but it would be very helpful if you would answer these questions
for me.
How did you get my number? I'm not in the book
Your number was chosen by a computer which randomly generates a list of
all the numbers that might.bein use in this area. I then dial the numbers that
the computer comes up with until I find one that is in use, like yours. This
method is used because it is the only way we can be sure that I get to talk to
a fair sample of people in [addname of the area].If I do not talk to a fair
cross-section, then my results will be misleading.
WFy do you want to talk to [someone of the opposite sex, sonleone
younger or older] and not me?
I need to makegure that I get to talk to a good mix of men and women, older
and younger people. The computer helps me to do that by telling me who
to ask for each time I make a call.
".. .,' +,
Hmm. I'm still not sure.
If you want to check that what I'm saying is accurate, why not call [rzaine
and number of academic supervisor] and I'll call you back afterwards.
Approaches to lntewlewing 81
* Interviewer reliability should be high, since a supervisor can monitor calls and
spot cases where the interviewer diverges from the script.
The researcher can quickly see how the work is progressing by reviewing the
completed response forms as they are passed on by the interviewers.
The main disadvantages are more or less those common to survey methods:
Respondents will not be prepared to spend a long time answering questions.
There is a consensus (Frey, 1989) that the questions need to he fixed-response
ones, since open-ended questions are harder to manage over the phone than
face-to-face and answers tend to be less complex and shorter.
The interviewer has little guarantee that the respondent's mind is really on the
questions and not distracted by TV, children, pets, or the dinner that is burning.
Phone surveys take more interviewer time than do questionnaires, involve
phoning outside of normal work hours, and require repeated attempts to
contact the right person at some numbers.
They demand a lot of concentration and energy to keep to the script and to
sound bright.
Interviewer training is necessary and interviewer supervision is common.
Each call is more expensive than the cost of sending out each questionnaire.
(However, the cost per response of the two methods may favour telephone
interviews.)
And, of course, the data produced will be of the quality that comes from all
survey methods.
In this balance of opportunities and problems, there is one important aspect
of telephone interviewing that commands serious attention, and that is the con-
struction of the sample. As with mailed questionnaires, careful construction of the
sample is very important where the intention is to generalize with statistical
confidence from the sample to the population. Using telephone interviewing, it is
broadly possible to target specific groups of respondents, such as middle-income
or low-income groups, by matching residential information (perhaps taken from
census summaries) to phone codes, although this is complicated where the codes
do not align with the residential districts to be surveyed. Computer-assisted
telephone interviewing (CATI) is a powerful way of generating random samples.
The area codes that are to be surveyed are entered and a computer program
randomly generates telephone numbers, within the range of numbers used by the
telephone company. This random sampling will select numbers that are unlisted -
a substantial proportion of numbers in some areas, especially in the inner city
- but telephone subscribers can become quite irritated at being 'cold called' in
this way and decline to participate in the survey. There are also manual ways of
generating a sample of telephone numbers, although some of the numbers,
whether generated manually or by computer, will not be in use, so the size of the
sample will always be smaller than that pool of numbers.
If interviewers only speak to the adult who first answers the call, the sample
will be systematically biased, since it has been found that olderpeople and women

82 Interviewing for socid scientists
are more likely to be on hand. So, sometimes interviewers are told to ask to speak
to the eldest adult male, or to the youngest adult female in the household. How-
ever, a requestto speakto the youngestmale, or to a female wage-earner (strategies
designed to get a balanced sample of respondents) will often be met with the
response that he or she is not available, so a further call will be needed.
Although the telephone interviewing method has many attractions, it is
nevertheless little used in social science research. One reason is because social
science researchers are often interested in people who have a characteristic that is
not related to their phone number or address. For example, the phone numbers of
single parents, widows, recent immigrants and highly educated people cannot
be generated by computer or picked out from the phone book. Secondly, where a
researcher is interested in people who share an occupation, such as nurses, shop
workers, or laboratory technicians, it is not satisfactory to call them at work.
Teachers, for example, usually share a staffroom phone, are seldom close to it,
and have neither the time nor privacy to deal with a phone interview.
Oral history interviews There is no reason why an oral history interview cannot
take the form of a survey, although it is quite rare, since the historian is usually
preoccupied by understanding the details and meanings of the particular rather
than by trying to generalize (which can be a pretty risky business in history, given
the effects of forgetfulness and decay, of time colouring the past, and of death
carrying off witnesses). Unstructured interviews are avoided since they tend
to produce a mass of incompatible data, which can be analysed, but which can
- leave the researcher wondering whether other informants would have endorsed or
rejected points that some had made but which they themselves did not spon-
taneously volunteer. It is more usual for oral history to use semi-structured
interviews that allow informants to depict the past in their own words, following
their own sense of what was important. The researcher, guided by the literature,
documents and other interviews, will have a loose agenda of questions to ask and
themes to explore, but the answers will be open-ended, and the interview will not
be dominated by the researcher in the same way as is the case with surveys. Yet,
oral historians are often anxious to get a picture that looks as though it might
credibly k$&?nt the experience of people other than those who were inter-
viewed. FO; that reason, they may take considerable care to try to get a sample
of informants that mirrors the structure of the past in some way, for example, in
terms of social class, occupations, geographical area or gender. So, their work
may combine careful but non-random sampling with a semi-structured format.
It might be expected that informants' memories of things that happened up to
half a century ago might be particularly unreliable (and there are questions about
the reliability of our memories of yesterday, let alone of 50 years ago). This has
been investigated and the opinion of two authorities (Lummis, 1987; Thompson,
1988)is that informants can speak in detail about things that were salient to them,
such as critical incidents and processes they lived day in and day out. In fact, their
memory of distant times can, in some respects, be better than their memory
of more recent times. However, nothing should be taken on trust, so as much
as possible should be checked against other sources. This is a bread-and-butter
Approaches to interviewing 83
matter for historians, who were triangulating data long before social scientistshad
invented a geometrical term for the Renaissance practice of critical document
study. Interestingly, oral historians have concluded that the errors they detect can,
in themselves be revealing, showing a lot about myths and explanatory frame-
works. Arguably, historians have an advantage here over social scientists. The
social scientist investigates the contemporary world where informants may feel a
need to put up a front. When the subject is something that happened half a century
ago, that need is likely to be less strong.
As with social science researchers, oral historians tend to pay a lot of attention
to building rapport in interviews so that they get the best quality data, although
Lummis reports that he has 'had good open interviews with people with whom I
felt little rapport and had no personal liking' (1987: 68). (We develop this theme
of rapport in interviews in Chapter 7.) Since they will often interview the same
informant repeatedly, there is a fair chance of confidence developing, so that
the informant edges from 'public talk' to 'private talk'. A common aim in these
interviews is to get as full a description of an event, milieu or time as possible,
which means that the interviewer needs to be skilled at listening and probing. The
list of probes and prompts is as long as the researcher's imagination. The impor-
tant thing is to probe for clarification and detail witlzout turning the interview into
an interrogation and ceasing to listen. Box 6.3 shows some commonly used
probes, which might be used by social science researchers as well as by oral
historians.
Oral historians, like social science researchers, have considerable problems
with data management and data analysis (see Chapter 11).However, the idea that
there can be different readings of an archive is again a well-established one in
history. The discipline handles this by requiring researchers to substantiate their
claims with explicit, footnoted reference to an archive that is publicly available
and submits interpretations to the test of the extent to which they are plausible to
others in the disciplinary community. Again, there are parallels here with some
social science approaches to establishing the credibility of research findings.
Lastly, the results of oral history work may most frequently appear in the form
of a conventional academic report, but they also appear as broadcasts and some-
times lead to films, novels or other creative representations of a discovered past.
Such practices are becoming more common in some social science departments.
Designing an interview study: an example
These first six chapters have concentrated on the design of interview research.
The following account of a large-scale study in which Peter was involved in
1994-6 draws together a number of themes that have been developed in these
chapters. As you read it, you might consider whether the study could have been
better designed to address issues identified in the literature; whether a different
interviewing approach would have been more productive; whether the research
team should have taken seriously a more positivist approach to reliability and
validity; and whether they could have done anything to allow them to make