I

I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I

CHILDREN & SOCIETY VOLUME 10 (1996) pp. 145-154

Researching Children:
Methods and Ethics
Ann Mahon and
Caroline Glendinning
National Primary Care
Research and
Development Centre
University of Manchester
Karen Qarke
Department of Social
Policy, University of
Manchester
Gary Craig

School of Sodal Policy,

University of
Humberside, Hull

The appropriateness and desirability of researching children have

been issues of some debate. Children may be perceived as noncompetent or vulnerable, and proxies have been used as children's
representatives. Increasingly researchers are speaking to children
directly. Why is this so and what are the methodological and ethical
implimtions of researching children's views? In this paper the
authors draw on their own experiences of researching children in the
fields of child carers and the impact of the Child Support Act 1991.
A number of social, political and legal trends are identified which
form a background to the growing interest in children as potential
and actual participants in the research process. The theoretical,
methodological, ethical and practical issues involved are then
identified and described, using examples from two separate studies
conducted by the authors.

his paper reflects on the experiences of two research

studies involving semistructured interviews with
children. The first study focused on the experiences
of children and young people caring for a family member and
evaluated the impact of three projects specifically designed to
support young care givers and their families. Face-to-face,
semistructured interviews were conducted with children and
the person who they were caring for, in 25 families. The
yoimg carers' ages ranged from nine years to mid-twenties,
although issues discussed here generally relate to yoimger
carers in this Scimple. Wherever possible, interviews with
children and parents were conducted on separate occasions.
The study was commissioned by the former Mersey Regional
Health Authority, which had funded the young carers'
support projects (Mahon and Higgins, 1995).

Address for correspondence:

Professor Gary Craig, School of
Policy Studies, University of
Humberside, Inglemire Avenue,
Hull HU6 7LU

CCC 0951-0605/96/020145-10
1996 by John WUey & Sons, Ltd.

The second study focused on the impact of the Child Support

Act 1991 on lone mothers and their children receiving certain
means-tested benefits. Semistructured interviews were carried
out with 12 children aged between 10 and 17, whose mothers
had already been interviewed earlier in the study. The
interviews explored the children's perceptions of family
relationships, their views on parental obligations and their

146 Ann Mahon and others

knowledge of financial arrangements associated with the Child Support Act. Again,
every attempt was made to interview the children separately from other family members.
This was the last stage of a part-longitudinal study (Clarke, Craig and Glendinning, 1993,
1994,1996) and was funded by the Joseph Rowntree Foundation.
In designing and undertaking these studies we had to address a number of difficult
ethical and methodological questions about involving children directly in research in this
way. The first was about the competence of children to provide answers to questions of
both fact and opinion: would they know, for example, about the payment of maintenance
and other sources of household income; would they have views on the respective rights
and obligations of parents and children in separated families? Could equally valid or
more accurate responses not be obtained from the adults involved? A second issue was
the particular vulnerability of children and their need for extra protection from potential
exploitation by researchers. Should not children be shielded from intrusive and
potentially distressing questioning, with proxiesparents, teachers, sodal workers
representing the views of children instead?
In addressing these issues, we came to the view that in many research contexts it is
neither theoretically nor methodologically appropriate to rely on proxies to represent the
views and experiences of children. On the contrary, children's views can and ought to be
taken seriously. However, accessing and understanding these views raise a number of
important issues for researchers across a wide range of disciplines.

Social^ political and legal pressures

The trends outlined here, whilst not new, together create a set of imperatives for
researching children's experiences within a wide range of theoretical and applied social
research.

Changing family patterns

Fundamental changes in the structure and stability of families in recent decades have
major implications for how we approach family research. Even so, there is considerable
discrepancy between official statistics on family status and the actual experiences of
women, men and children. For example, whilst numbers of both lone and single
(never married) parents appear to have increased rapidly, these mask growing trends to
cohabitation prior to or instead of marriage. Evidence firom birth registrations suggests
that increasing numbers of children are bom into stable cohabiting relationships (Kieman
and Estaugh, 1993). There is no official documentation of the breakdown of such
cohabitationsan increasingly common cause of lone parenthood (Bradshaw and Millar
1991; Ford and others, 1994)nor of the entry of women and children into new cohabiting
relationships.
Official figures on remarriage are equally unhelpful, as they do not include growing
numbers of children who live in unmarried step-parent families. Nor do they reveal ways
in which adults and children may experience a number of different family forms during
their lives. For example, a fifth of the lone mothers in the Child Support Act study had
had children with more than one partner: some lone mothers had been married, divorced
and subsequently cohabited with another partner with whom they had had further
children; others had been married (and divorced) more than once; yet others had had

Researching Children: Methods and Ethics 147

children in a succession of short-term, non-cohabiting relationships (Clarke, Glendinning
and Craig, 1994).

It is therefore not appropriate to rely solely on information about family structure and
status to tell us how children experience their families and, through them, their wider
social worlds. Assumptions about the impact of, say, marital breakdown and entry into a
step-familyeven about the effects of such basic characteristics as age and birth order
may be increasingly unreliable. For example, one lone mother in the Child Support study
had had her first son while married to a violent partner. After this marriage broke down,
she had three more children with another man who treated the older boy exactly the same
as his own children; when interviewed, the boy confirmed that he regarded his mother's
subsequent partner as more of a 'father' to him than his biological father. Even children
who share the same biological parents may have very different relationships with each
parent; such differences in affection and allegiance become increasingly salient in shaping
the children's experiences if the parental relationship is in difficulty. We therefore cannot
even assume similarities in the experiences of children who are currently members of the
same family. The diversity and fluidity of family forms as children move through
childhood thus impel us increasingly to seek their ovm accounts and experiences, rather
than assuming that we can automatically know them from demographic and structural
information we have about their families.
Political trends
The 1980s saw a political emphasis on 'consumerism' within public sector services,
reflecting the introduction of a market ethos as a means of promoting user responsiveness.
This has added to the pressures for the 'consumers' of public servicesincluding,
potentially, young 'consumers'to be involved directly in the evaluation of service
initiatives. Thus in the study of young care givers, the Regional Health Authority which
commissioned the research assumed that the evaluation of the young carers' projects
would of course involve those service 'users' whom the projects had been established to
support, rather than simply relying on reports from parents or social services staff.
Concern to involve young people directly in the research also reflected the child-centred
approach adopted by the projects that were the subject of the evaluation study (Mahon
and Higgins, 1995). This, in tiun reflects, inter alia, the impact of legal trends outlined later.
Legal trends
The rights of children and their status within the family have been addressed by several
major recent pieces of legislationthe Children Act 1989, the Child Support Act 1991 and
the Criminal Justice Act 1991. These Acts originated from very different policy concerns
and have resulted in some marked inconsistencies in the current legal status of children
among the separate social security, criminal justice and child welfare systems.
Nevertheless, the flurry of legislative activity is itself a reflection of the rapid sodal and
political changes referred to earlier.
Of most significance here, the Children Act represents the culmination of a shift from
children being the passive objects of parental rights to becoming legal subjects in their
own right. In any decision under the Act, the welfare of the child must be the paramount
consideration and his or her views and wishes must be ascertained. Although the latter
are to be 'considered in the light of his (sic) age and understanding' (S. 6.1), this nevertheless represents an important formalisation of children's rights to be heard within the
legal system.

148 Ann Mahon and others

Theoretical problems
'Normal' childhood and the construction of social 'problems'
Academic acceptance of diversity in family form is widespread. However, the funding of
applied research is still too often focused on situations and behaviours that are assumed,
implicitly or explicitly, to be deviant and problematic. For example, studies of the effects
of separation and divorce have tended to focus on the negative outcomes for children,
compared with that in children in intact families; only recently has research begun to
challenge this implicit acceptance of negative outcomes (Burghes, 1994) or to focus on the
factors that might help children adapt and cope (Cockett and Tripp, 1994). Yet in view of
the diversity of family forms and relationships described earlier, the prefiguration of some
types of family experience as 'normal' and some as 'deviant' seems increasingly
inappropriate.
Similarly, much research on child carers has assumed that informal caregiving as a whole
is necessarily problematic and may have pathological consequences for the people
involved. This model has itself been vigorously criticised by disabled people, on the
grounds that it denies their autonomy and agency by casting them as wholly passive and
dependent (Keith, 1992; Morris, 1991). It is equally debatable whether this is an
appropriate model to use when investigating the kinds of support that younger people
may give to adult family members with physical or mental disabilities. Quite simply, we
know very little about the range of responsibilities that children feel and carry towards
their various family members, nor how these vary with age, gender and ethnic group. In
the study of child carers a wide range of caring scenarios were encountered, including
variations in the illness or disabilities experienced by those being cared for, the t)rpes of
caring tasks and responsibilities being undertaken by the young people, and their
attitudes towards the tasks and consequences of caring. Indeed, one of the difficulties in
interpreting these sorts of findings is that of distinguishing those tasks considered
'normal' from those that are out of the ordinary, both within the particular families being
studied and in the wider social context.

Validity
A second issue which emerged in our studies concerns the concept of validity. As
indicated earlier, the increasing diversity of feimily situations means that we cannot
assume that adult 'proxies' such as parents or teachers will be able to give valid accounts
of children's own social worlds. In the Child Support Act study, lone mothers were often
aware that their child had very different experiences of and feelings towards an 'absent'
parent from those of their own. Moreover, these mothers knew that children would
sometimes conceal such feelings so as not to hurt them, even to the extent of hiding
apparently straightforward 'facts' such as recent contact with their father. Valid accounts
of children's attitudes and experiences could therefore only be obtained by engaging
directly with the children and treating them as independent actors who were engaged in
negotiating a complex set of relationships and loyalties.
Similar questions of validity arise in relation to the use of siblings as 'proxies' for each
other. Again, the Child Support Act study revealed that siblings could have very different
relationships with each of their parents. Consequently, the attitudes of children within
the same family about parental obligations and the payment of maintenance could be as
polarised as those of some lone mothers and their former partners. The study of child
carers also found that children within the same family assumed different caring

mm

I
'9
A
^
_

Researching Children: Methods and Ethics 149

I
I
I
I

responsibilities, held different views about their roles and experiences as carers and used
the support offered to them through the projects in different ways.

Methodological issues
A key issue in researching children is the extent to which the researcher is seen as some
kind of expertwith the risk that 'public' accounts will be given that are thought to be
acceptable to the 'expert' point of view of the researcher (Comwell, 1984). This distinction
between public and private accounts may have particular relevance for children and for
the relationship between researcher and child. Younger children may be more suggestible,
or more inclined to respond to an interviewer with answers that they think are expected
and wanted than are adults; but the problem is essentially the same as that faced in relation
to adults. The options available in response to the problem are rather different however.
With adults, the interviewer can be matched to the person being interviewed on a range of
social characteristics such as age, gender or ethnic origin, in order to create a relationship
in which the person being interviewed feels rapport and identification with the
interviewer. However, this is not always possible with (younger) children because of
one of the most salient social characteristics in the situationage (and, deriving from this,
in most cultures, authority). Increasingly, however, children are being involved in research
projects, not just as subjects of research, but as researchers in their own right (Alderson,
1995). The appropriateness of involving children as researchers is not always apparent and
should be judged on the basis of the research topic, the research methods, and the degree
of skill and responsibility (and, hence, age or maturity) required of the researcher.
In the research conducted with the children of lone mothers, the two interviewers
employed both had extensive experience of working with and talking to children in social
work and legal context. We have no way of knowing how the identity of the interviewers
that children constructed affected the kinds of issues that they were wiUing to talk about
or the ways in which they expressed their views. However, it seems very likely that the
nature of the conversations which took place were affected by the interviewers' identity
in the child's mind. The 'problem' of adult authority in relation to children may be more
acute when the child and the researcher are together on a one-to-one basis. The adoption
of more varied and imaginative research methods may make it possible to overcome
these problems to some extent; for example, the use of group interviews and focus
groups, interactive research methods such as video and drawings, and the involvement
of young people themselves as researchers.
When interviewing children it is important that children understand the questions asked
of them and that researchers understand what children are saying; that the children are
understood as well as (literally) understanding. So far, we have referred to 'children' as if
they are a homogeneous group. Clearly, the problems that arise in relation to validity and
other aspects of research methodology are very different for children of different
chronological and developmental ages. The research with child carers, for example,
involved children across a wide range of ages from as young as nine years. Less
'successful' interviews, in terms of the depth of response and rapport established, were
conducted with those in younger age groups, and with boys. In retrospect, alternative
and more imaginative methods, perhaps including the use of a male researcher, might
have yielded richer and more valid data. For the older age group, face-to-face interviews
generally seemed an appropriate method. Obtaining iriformation from children in
younger age groups may therefore require an approach different to that from older age

150 Ann Mahon and others

groups and this involves different skills and different research methods. Recent research
exploring children's and young people's perceptions of cancer, for example, devised a
'write and draw' technique for the collection of data from children aged nine to ten years
(Oakley and others, 1995).

Ethical issues
In some of the discussions with colleagues and advisers around our respective studies, we
encountered anxieties that the children we planned to interview would not fully
appreciate what was being asked of them; or, as discussed earlier, that they would
construct the researcher as an authority figure to whom they had to acquiesce, even
though it might not have been in their own best interests to do so. In other words, it was
argued, children somehow needed extra 'protection' from the invasive questioning of
researchers, who otherwise risked exploiting or abusing the children's lack of understanding and/or lack of power. Thus, in the Child Support Act study, we were warned
about the risk of interviewing children about relationships with their (separated) parents
and their views on parents' continuing financial obligations, on the grounds that this
would prove too upsetting for the children.
However, the researchers' view was that children are, in a variety of different ways,
capable of deciding whether or not they wished to be interviewed. The children were
given fuU information beforehand about the scope of the proposed interview and had a
number of opportunities to decline. Children were approached directly for their consent,
with parents first being asked only for permission to make contact with the child. Parents
were given vratten information about the aims and scope of the interviews with the
children. A separate approach was then made to the child at which his or her consent was
sought and each child was asked again at the start of the interview whether he or she was
willing to take part. At the end of the interview, in acknowledgement of the time that they
had given, a small payment was made to the chUd, although care was taken not to
mention this payment in negotiating the child's consent. Children who did not want to be
interviewed either refused outright or did not keep appointments (here the interviewer
had to exerdse a considerable amount of discretion in deciding whether to call back and
try and fix another interview).
In the study of child carers it was decided at the outset that the children involved shotdd
be the main focus of the study rather than adults; indeed this was assumed by the
commissioners of the research and reflected the child-centred focus adopted by the
projects. In this study parents and children were contacted by letter at the same time and
both were asked for their consent to approach the child to request consent to be
interviewed. As the research also involved interviews v^dth parents or another family
member, children were asked for their consent to approach their parent/relative for
interview. All of those contacted were made aware of the right to refuse and it was made
dear by the researcher that participation was not something which was expected. Careful
reading of body language revealed some conflict at times with children's verbal consent
and further probing led to a refusal to be interviewed in at least one case in this study. As
experienced in the Child Support Act study, 'failure' to keep appointments may be an
indirect way of refusing to participate.
Perhaps the greatest anxieties or concerns that emerged from our respective studies were
those relating to confidentiality and disclosure, both in relation to parents and in relation

I
|
_
I

I
^

H
W

"

^
^
M
I
m
M

_
I
"
^
|
M

Researching Children: Methods and Ethics 151

I
I

to child protection authorities if a child was considered to be at risk. There appears to be

an emerging consensus amongst researchers that complete confidentiality can never be
guaranteed to child research subjects. The National Children's Bureau, for example, takes
the view that confidentiality can never be guaranteed to children because the researcher
has a duty to pass on information to the appropriate professionals. Our own studies
reflected two slightly different ways of putting this into practice. In the study of child
carers it was decided provisionally to guarantee total confidentiality to all the children
and young people interviewed. It was recognised that issues relating to child protecfion
might emerge during the course of the interview and a pragmatic decision was taken to
renegofiate guarantees of confidentiality, should the responses given by the child begin to
raise concern. In the Child Support Act study, children were guaranteed confidenfiality
but with the proviso that, if it emerged during the interview that someone in the family
was at risk, the interviewer might have to tell a third party. The children were, however,
told that nothing would be done without consulting or informing them.
A third ethical issue concerned the potendal effects of parfidpating in the research. In the
Child Support Act study, there was considerable concern that being involved in the
research would result in distress to the children and create ongoing problems for their
mothers, which the researchers would have no power to assist with and might not even
know about. We tried to meet this concern by giving careful thought to the process of
interview closure. The interviewers therefore checked with the child how he or she felt at
the end of the interview and whether he or she felt the need for any further help or
support. None of the children said that they did; had they done so, the interviewer would
have referred them to an appropriate local service rather than attempting to respond
directly to the child's needs him or herself. Letters sent later to parents and children
thanking them for their participafion reiterated the interviewer's willingness to identify
an appropriate local source of support if it was requested.
In the study of child carers, concern about possible distress resulting from the direct
questioning of aspects of children's lives that they perhaps had not previously discussed
with others was considered to be less problematic, since all of the children and young
people had varying degrees of contact with workers attached to the projects. Consequently, many of the young people had discussed family illness and their roles as
carers on previous occasions and an accessible source of support was already available to
them.
One argument that has been counter-posed to this concern about possible distress is that
it is beneficial for the child to have an opportunity to talk to someone outside the family
about issues which are important, if distressing, to him or her. However, it seemed to us
extremely important not to allow such arguments on their own to justify the involvement
of children (or adults) in research. It is crudal to mark the boundary between research
and therapy very clearly. Any therapeutic effects of partidpating in research carmot be
used as a justificafion for conducting the research. This boundary between research and
therapy may be less easy to maintain where qualitative research involves a longer term
relafionship with a child than that created in the course of a single interview. Children
may of course ask for help or advice from the researcher. Whilst it would seem
appropriate for researchers to try to meet these expressed needs by referring the child to a
local advice, counselling or support service, it is perhaps important that the researcher
does not attempt to take on such a role him or herself. This is no different from many
qualitative research scenarios where researchers come across situafions (for example,
failure to take up benefit entitlements) which may require expert advice from others.

152 Ann Mahon and others

Finally, it is important to bear in mind that children, more than adults, may have
expectations that issues and problems which they identify in the course of the interview
might somehow be put right as a result of taking part in the research. It is clearly very
important when explaining the purpose of the research to spell out its limitafions and the
restricted role of the researcher in relation to the child's day-to-day situation.

Practical issues
Many of these methodological issues have important practical implications for research
with children. For example, the age and development of children should be considered as
part of the selecfion criteria for a study population because of the implications that these
have for the choice of research methods and the skills and understanding required of both
researcher and child. Having dedded on the study population, the researcher must be
able to gain access to the children. Again the process of gaining access will vary according
to the characteristics of the children being studied. For example, child carers are often
referred to as a 'hidden' or 'secret army' and studies attempting to gain access to child
carers have often experienced difficulfies. In the study of child carers children were
recruited via supporting projects; other studies of children have recruited them through
schools (Alderson, 1995; Oakley and others, 1995).
The process of gaining informed consent and maintaining motivafion are of particular
importance when researching children and young people. In the study of child carers it
was made dear on the first approach and at the beginning of the interview that the child
or young person was not under any obligafion to parfidpate. This was important because
the young people may have regarded the interviewer as an authority figure or
succumbed to parental pressure to partidpate in the research. Requesting consent from
children should, whenever possible, be done privately, away from other family members.
In terms of mofivafion, the choice of research method may have different implications for
the length of time that children with different interests, personalities, gender and so on
can concentrate and maintain an interest in the interview. In the study of child carers, the
age of the child was directly related to the length of the interview, which in turn was
related to the depth of response. Declining mofivafion during an interview could also be
understood as an indirect way of withholding or withdrawing consent to parfidpate, if a
young person found it difficult to do so directly.
Finally there are pracfical considerafions relating to privacy, location and the need to
conduct interviews away from parents or other family members. This is important, but
not always feasible. Where should interviews with children be carried out? At home, at
school, in a burger bar? It is frequently desirable to interview people in their own homes
and this often provides important observafional data for the researcher. In the child
carers study, several interviews were conducted in burger bars or other 'neutral' settings,
as duldren did not want to be interviewed at home. On other occasions interviews were
conducted under very difficult circumstances with constant interrupfions and background noise. The locafion and environment in which interviews are conducted has
implicafions not just for privacy and the quality of data collected but also for how the
data are recorded. In both the child carers study and the Child Support Act study,
interviews were taped and it was anfidpated that there might have been some problems
in doing this. In practice no problems were encountered in gaining consent to taperecord. Giving children control over the tape recorderby giving it to them to hold and
to turn on and off as they wishedwas found to aid not orJy the willingness to be taped
but also rapport between researcher and respondent.

I
9
am
I

I
^

'
M

I
w
M
||
^
M
"
B
M

^
M

Researching Children: Methods and Ethics 153

Conclusions
We have started from the viewpoint that, for important theoretical, social and legal
reasons, children's perspectives on the issues and experiences which affect them can and
should be investigated. It is no longer appropriate (if indeed it ever was) to rely on the
accounts of others or to try and make deductions from knowledge about children's
location within a given family structure. At this stage, we would like to draw two main
conclusions, although our work in this research area is continuing.
First, we do not claim that any of the problems we have encountered and the solutions we
have described in this paper are particularly original to sodal research. The ways in
which we have been forced to think these issues through has brought home to us a
number of practical and methodological issues that are important and relevant to
research in other contexts as well. In other words, the experience has had an additional
heuristic value as well, in helping us to think our way more dearly through some of the
issues involved in a wider variety of research contexts.
Secondly, it has made us aware of the importance of drawing on and building on the
experiences of others. We were lucky in that we had available to us work carried out, for
example, by the National Children's Bureau and the major national children's welfare
organisations, which have already developed guidelines on involving children in
research. We were able to recruit sldlled and sensitive interviewers. We were also able to
talk to people with experience of carrying out research on very difficult subjects,
induding the immediate aftermath of marital breakdown. We were also aware that there
is a wealth of experience of researching children to draw on in other disdplines, such as
developmental psychology. These contacts should help us to develop good practice in
what is likely to become a growing research area. They may also prove valuable allies
against those who maintain that children's voices cannot or should not be heard.

References
Alderson, P (1995) Listening to Children: Children Ethics and Social Research. Bamardos
Bradshaw, J and Millar, J (1991) Lone Parent Families in the UK. DSS Research Report 6. HMSO
Burghes, L (1994) Lone Parenthood and Family Disruption. The Outcomes for Children. Family Policy
Studies Centre
Qarke, K, Craig, G and Glendinning, C (1993) Children Come First? Bamados
Clarke, K, Craig, G and Glendinning, C (1996) Small Change: Turn Years of the Child Support Act.
Family Policy Studies Centre
Clarke, K, Glendinning, C and Craig, G (1994) Losing Support: Children and the Child Support Act.
Children's Society
Cockett, M and Tripp, J (1994) The Exeter Family Study. University of Exeter Press
Comwell, J (1984) Hard Earned Lives: Accounts of Health and Illness from East London. Tavistock
Publications
Ford, R and others (1994) Lone Parenthood in 1989,1991 and 1993: Three Cross-section Survey Report.
Policy Studies Institute
Keith, L (1992) 'Who cares wins? Women, caring emd disability'. Disability, Handicap and Society, 7(2),
167-76
Kieman, K and Estaugh, V (1993) Cohabitation. Family Policy Studies Centre
Mahon, A and Higgins, J (1995) 'A life of our own . . . ' Young Carers: An Evaluation of Three RHA
Funded Projects in Merseyside. Health Services Management Unit, University of Manchester
Morris, J (1991) Pride against Prejudice: Transforming attitudes to Disability. Women's Press
Oakley, A and others (1995) 'Health and cemcer prevention: knowledge and beliefs of children and
young people', British Medical Journal, 310,1029-33

154 Ann Mahon and others

Contributors' details
Ann Mahon is a research associate at the National Primary Care Research and Development Centre,
University of Manchester.
Caroline Glendinning is a seriior research fellow at the National Primary Care Research and
Development Centre, University of Manchester.

I
tm

Karen Clarke works in the Department of Social Policy, University of Manchester.

Gary Craig is Professor at the School of Social Policy, University of Humberside, Hull.
Professor Gary Craig, Caroline Glendinning, and Karen Qarke collaborated in the study of the
Child Support Act, 1993-96, and Ann Mahon conducted the study of young carers, 1994-95.