CHOOSING

RESEARCH
METHODS
DATA COLLECTION
FOR DEVELOPMENT WORKERS

Development Guidelines No. 7

Brian Pratt and Peter Loizos

First published in 1992; reprinted in 1995 and 2003
Oxfam UK and Ireland 1992
ISBN 0 85598177 6
A catalogue record for this publication is available from the British Library.
All rights reserved. Reproduction, copy, transmission, or translation of any part of this
publication may be made only under thefollowingconditions:
with the prior written permission of the publisher; or
with a licence from the Copyright Licensing Agency Ltd., 90 Tottenham Court Road,
London Wl P 9HE, UK, or from another national licensing agency; or
for quotation in a review of the work; or
under the terms set out below.
This publication is copyright, but may be reproduced by any method without fee for
teaching purposes, but not for resale. Formal permission is requiredforall such uses,
but normally will be granted immediately. For copying in any other circumstances,
or for re-use in other publications, or for translation or adaptation, prior written permission
must be obtained from the publisher, and a fee may be payable.
Available from:
Bournemouth English Book Centre, PO Box 1496, Parkstone, Dorset, BH12 3YD, UK
tel: +44 (0)1202 712933; fax: +44 (0)1202 712930; email: [email protected]
USA: Stylus Publishing LLC, PO Box 605, Herndon, VA 20172-0605, USA
tel: +1 (0)7036611581; fax: +1 (0)703661 1547; email: [email protected]
For details of local agents and representatives in other countries, consult our website:
www.oxfam.org.uk/publications
or contact Oxfam Publishing, 274 Banbury Road, Oxford OX2 7DZ, UK
tel: +44 (0)1865 311 311; fax: +44 (0)1865 312 600; email: [email protected]
Our website contains a fully searchable database of all our titles, and facilitiesforsecure
on-line ordering.
Published by Oxfam GB, 274 Banbury Road, Oxford OX2 7DZ, UK.
Printed by Information Press, Eynsham.
Oxfam GB is a registered charity, no. 202 918, and is a member of Oxfam International.

This book converted to digital file in 2010

CONTENTS
Foreword v
Introduction 1
1 Starting research: some basic issues 5
Identifying a research need 5
The level of participation 8
Expectations aroused by research 11
Constraints on conducting research 13
Government attitudes 13
Conflicts and emergencies 14
Cultural factors: gender relations and power structures 17
Resources 19
2 Strategic issues in planning sound research 21
The unit of analysis 21
Questions of time 26
The research timetable 28
Continuous research or a snapshot? 30
Objectivity, subjectivity and the control of bias 32
Representative results 34
Some pitfalls to be avoided 36
Choosing the researchers 39
Local participation and experts 41
Training the field team 43
Styles of research 45
3 The range of research methods 47
Making the most of existing resources 47
Asking questions and interviewing 52
Group interviews 55
Oral testimonies, oral histories and life-histories 55
 Surveys 58
Sampling 60
The pros and cons of a survey 62
Participant observation 63
Rapid Rural Appraisal 66
The variety of RRA techniques 69
Collating and presenting RRA findings 71
Limitations of RRA 71
An example of a RRA 72
Participatory Rural Appraisal 74
4 General issues affecting most research methods 78
The use of appropriate indicators 78
Validation 79
After research: some crucial questions 81
Bibliography 85
Appendix 1 Checklist of questions 87
Appendix 2 Questionnaire design 89
Appendix 3 Some high-tech, high-cost research methods 92
Appendix 4 Logical Framework Analysis 95
Appendix 5 Women and evaluation 97
Appendix 6 Action Research Programme: the methods
and approach 100
Index 115
FOREWORD
This book started life in a 'brainstorming' workshop in which the
following people took part: Dr Jo Boyden, Roy Cole, Donnacadh
Hurley, Dr Peter Loizos, Dr David Marsden, Anne Muir, Dr Brian
Pratt, Dr Allan Stanton, Roy Trivedy, and Dr Tina Wallace. The
ideas of the group were then written up by Brian Pratt and Peter
Loizos.
The book is intended as an introduction to research for small-
scale development programmes. It will not make you instantly
proficient in any of the methods described, but it should help
you to decide what sort of research you might do, and what the
advantages and disadvantages are of particular methods. So it
should guide you to a choice of methods, and prepare you to go
deeper into the methods of your choosing.
If you have taken a course in Social Research Methods, this
book will be probably be too introductory for you. It is
designed to complement other volumes in the Development
Guidelines series. Once many of the questions posed here have
been answered, the reader may well decide that (for example) a
full-scale survey is required. Development Guidelines 6, Social
Survey Methods: A fieldguide for development workers, by Paul
Nichols, gives a full explanation of methodology for carrying out
surveys with limited resources. Number five in the series is
Evaluating Social Development Projects, edited by David
Marsden and Peter Oakley, and this, too, may prove helpful. We
have, for the most part, assumed for the purposes of this book
that research is a part of the project cycle; however, we have not
specifically considered the different elements of that cycle and
their needs.
Throughout, we have tried to concentrate on a small numberof
well-tried methods, rather than provide a catalogue including all
minor ones. For example, we have not discussed the 'logical
framework' approach in any depth, because it is a way of organ-
ising the data collection rather than a separate method of collect-
ing it. (Appendix 4 outlines the principles of logical framework
analysis.)
Inevitably, one method tends to overlap with another, so some
degree of repetition has been unavoidable. We have also been
concerned to present the issues in an accessible way rather than
to pursue them in all their technical complexity. There are refer-
ences to other useful books for the reader who wishes to get
deeper into particular methods or issues.
The authors would like to express their thanks to all the partic-
ipants in the original group workshop. The book would not
have been possible without the experience and ideas they con-
tributed. Peter Loizos would like to thank Paddy Barwise, Martin
Bulmer, Robert Dodd and John Harriss for their most helpful
comments.

VI
INTRODUCTION

This guide is about research which is action-linked, that is, tied

to some form of practical intervention. By 'research' we mean no
more than the systematic collection, analysis and dissemination
of information. Research might be carried out in response to a
request by a local group to help them meet some problem, or to
help a funding agency determine its priorities or, later on, to
assist the planning and operation of a specific project. The
research may be tied to feasibility studies for a proposed pro-
gramme, monitoring and evaluation of an ongoing programme
or alternatively to the final evaluation of a programme which has
ended.
One argument of this book is that research should be integrat-
ed into the programme cycle and not be set apart as an activity
carried out by people unfamiliar with the programme. A second
contention is that there is no perfect research model which can
be used for all situations: to be effective, research must be tai-
lored to fit the specific problem. A third assumption is that gen-
der-related questions should be raised for most if not all
research projects, at all stages of research initial formulation,
research design, data collection, and analysis. And a final
assumption is that responsible and sensitive research is always
conducted consultatively, together with the people affected,
rather than inflicted upon them from above, or outside.
The word 'research' should not conjure up some very difficult
and specialised activity which can only be done by people with
higher degrees. We all do research with a small V constantly,
since we collect information in our daily lives and use it to guide
our decisions. Sometimes, however, we need to consult people
with special skills to help us in this process, such as engineers,
surveyors, accountants, health specialists of many kinds, just as

1
nobody trying to offer an especially delicious meal would think
twice about reaching for a cookery book. But the more we learn
for ourselves the skills for which we sometimes turn to experts,
the better placed we are to pursue our goals and to evaluate the
contributions of specialists. Research with a capital 'R' is a sys-
tematic approach to the gathering of information to solve specif-
ic problems no more, but no less. It is the systematic char-
acter, the desire to eliminate errors, which marks out serious
research from normal problem solving.
Attitudes to development are undergoing continuous change,
and in the last 20 years or so there has been a major change in
thinking on several related issues: first, to what extent is a devel-
opment programme one which has developed in close response
to the expressed needs of a local group, and how far should it
remain the brain-child of planners and policy makers at several
removes from the group in question? Put in another way, this
asks whether experts are serving the local people or are prison-
ers of their own interests and disciplines. Thus, 'expert-driven'
research has been called into question, and the idea is gaining
ground that the best development research both begins and
ends with the expressed needs of local groups who will be most
crucially affected by it. Experts should be on tap, but not on top.
The second issue involves the political values which underpin
an agency's involvement with a local group. How fully is the
agency consultative, democratic, and sensitive to local percep-
tions and wishes? 'Top-down' development programmes are
often accompanied by research which is insensitive towards the
very people it is supposedly meant to assist. It is therefore
increasingly the case that research is carried out in closer collab-
oration with the people affected, using their knowledge, skills
and, where this is their wish their energies; and at the very least
in a way which gives them as strong a voice as they think they
want about what research is planned, why it is planned, and
how it might proceed.
The third issue concerns the nature of expertise itself. In the
period before the first atomic bomb was dropped upon
Hiroshima, there was a general optimism about science in all the
industrial countries and among the leaders of independence
movements. Science and social science were then perceived as
expert systems which would solve most human problems within
a foreseeable future, and people who had not received any kind
of scientific training tended to have their knowledge downgrad-
ed. Since 1945 this optimism has been challenged by a far more
balanced view of scientific 'progress', and a gradual reduction in
what might be called the misplaced over-valuing of experts. It is
now increasingly realised that in many matters, from local medi-
cines and local knowledge of animals, plants, crops, and the
environment, to local scepticism about the market, money, and
profit-taking, the understandings of small-scale pre-literate soci-
eties, are often as pertinent as the most up-to-date scientific
thinking, and often a good deal more environmentally sensitive.
In the past, scientists and policy makers have backed many
ideas which have later been shown to be dangerously wrong,
and we are in the process of trying to find a new balance in our
appraisal of the very nature of knowledge about the world, a
balance which will no longer overvalue formal, official science
and undervalue indigenous knowledge.
Since neither experts nor locals are always right, or. always
wrong, the process of readjusting the relation between the two
will be slow and, in a conflict between different kinds of knowl-
edge, it may often be hard to decide which view is the most
worthwhile.
We hope that these three issues have sufficiently informedthe
arguments in this book. We believe that there is no politically
correct, off-the-peg answer to the questions raised by these
debates, and that the issues can only be decided on a case-by-
case, trial-and-error basis. There is no substitute for discussion
and consultation, nor for judgement and balance.
 STARTING RESEARCH: SOME BASIC
-L ISSUES

Identifying a research need

Research never 'just happens' it is always started by some

person or interest group, with a problem or policy goal.
Sometimes, the very decision to call for research will be hotly
debated within a community or an agency, for research will cer-
tainly involve energy, time and money. However, it is also com-
monly the case that there are differences of opinion within an
agency about whether a particular programme should be under-
taken or not, or if it is going well or badly; and such divergent
views may result in disagreements over whether research is
needed, or what its focus should be. The initial definition of a
research 'problem' is therefore a matter to be carefully thought
through.
Sometimes the most important question may go unasked: has
the problem initially identified by the funding agency the same
degree of importance to the people the programme or policy is
intended to benefit? What preconceptions are at work which
might be rejected as meaningless by the people concerned? For
example, some approaches to street children have assumed that
their problems are that they have been abandoned by their par-
ents and by society but the children themselves often define
their problems differently, in less passive terms, stressing their
wish to make better livings for themselves. In another instance,
well-meaning visitors to Cairo were outraged by the working
conditions of the people known as Zabballeen, who live by sift-
ing through the rubbish dumps of the city. The result was an
inappropriate aid programme to change the living conditions for
this group without any understanding of the economic and
social relations within it. The programme was a failure, because
the basic research had not been done, and the primary assump-
tions of the outsiders had not been questioned.
Specialists of various kinds tend to view groups of people
through the lenses of their own specialisms. Water engineers
worry particularly about people using dirty water and having
poor sanitation, and health workers become preoccupied with
disease and nutritional inadequacies. In such cases, the major
problem may be defined in terms of the 'specialty', although a
sensitive researcher, asking the people themselves to name their
biggest problems, might well hear not about disease or water
but about land tenure or loss of income. Thus, the questions
must always be asked who says this problem is the major
problem facing the community, and what priority do the people
themselves give it?
Traditionally, one group has dominated the identification of
development issues to be researched: white, middle-class, mid-
dle-aged males. Although this situation is changing, it explains
why some areas of society and the economy have received sig-
nificant attention over the years whilst other areas have been
consistently neglected. Typically, this has often meant that
women's interests have been disregarded. To take the simplest
example: a community may be unanimous about the desire for
an improved well to be dug. If only a few senior men are con-
sulted about the siting of the well, it may be dug in a place
which inflicts a burden on women as the people who most com-
monly carry the water from the well to the home. There is now
more research being done which is initiated and directed by
women, but it is still rare for lower-class groups to initiate devel-
opment research, and children almost never do so.
There are a number of phases of a development project or
programme at which the need for systematic research might be
identified. First, the feasibility study: if a community or group
request an input from an agency, the potential flinders are faced
with the question, should we try to meet this request? Is this a
problem which our policies allow us to address? What is to be
done about this, by what means, on what scale? If a major disas-
ter or emergency is in progress, some assessment of basic needs
will be needed to inform the intervention. The most important
things to find out will be the numbers of people affected, and
their social profile how many men, women, and children, at
what ages, with what skills, and resources, and with what cultur-
al requirements concerning, for example, diet. The feasibility
study is perhaps the most important single piece of research
which project planners can do, because the basic issues about
the appropriateness and the future viability of the project should
be tackled at this stage, some agencies are increasing the time-
depth of their involvement. Action Aid, for example, now seeks
to work in a defined geographical area for not less than ten
years, and requires its staff to examine three separate areas in
detail before committing the agency to one of them.
Research might be required at the project planning stage.
Assuming that it has now been established what service or input
a local group desire, this phase focuses on how it might best be
delivered and administered. The research might be directed to
finding out who owns which resources, and under what terms a
necessary input such as a plot of land might be made available
to the project. Questions of responsibility for the local running
of the project would also need to be raised and decided. In such
a situation, research, consultation, and planning of project
administration are a single, continuous process, with the
agency's representative inquiring as effectively as possible into
how the beneficiaries conceive of their participation, and mak-
ing clear to them the policy considerations which will guide the
agency in deciding whether or not to fund the proposed project.
At the planning stage, future monitoring and evaluation research
could be designed to ensure periodically that the project was
continuing to meet its stated goals. Best practice should seek to
specify how the alleviation of poverty will be demonstrated.
That is, the measurement of changes of income, consumption,
and well-being should be planned at an early stage.
If the project has been in operation for some time, research as
part of the process of monitoring and evaluation will be valu-
able in making clear to all concerned just how far the original
needs are being met. Supposing a revolving fund for small loans
to assist in income-generation projects for marginal groups has
been in operation. Since the viability of the scheme in the mid-
dle and long term will depend on the loan repayments of the
initial recipients, some analysis of the records will be essential.
Such an analysis is research with a small 'r'. It can obviously be
carried out in two quite different ways. On the one hand, the
senior agency staff manager could look at the records and draw
out their implications. Or, if the scheme is very much under the
control of a local group, the recipients of the loans could be
invited to participate in the data-analysis phase, so that they all
learned together just how well or badly the project was meeting
its goals.
Finally, there is retrospective evaluation research, which is
intended to reveal exactly why a project worked or did not work
in a particular way. This type of research could involve analysis
of project documents, interviews with all relevant participants,
and comparisons with projects in other places which had started
with similar goals and faced similar problems.

The level of participation

In recent years the development research community has
become concerned about the possibility that traditional research
procedures are perceived by local groups as intrusive, aggres-
sive, and distanced, and that too much development planning
has been a top-down affair, carried out in the name of local wel-
fare but over the heads of local people. To counter this, an idea
of development planning and linked research has arisen which
is altogether more consultative and collaborative.
For these and other reasons approaches have been developed
which have a completely different style and rationale. Hence,
the idea of 'participatory' research, in which researcher and local
people collaborate closely to explore the issues which the locals
think require research and action. The research is not something
initiated by the outside researchers, directed and controlled by
them, but rather something which grows up more naturally and
organically from the perceptions of the community, and their co-
option of the researchers. And even where an outside researcher
makes an initial overture, the subsequent course of the relation-
ship need not be outsider-led. But all kinds of compromises are
possible: if outsiders propose a project, and the local people
show enthusiasm, the subsequent direction of the work can still
be largely determined by their ideas and interests.

8
 Although some writers make it sound as though there is a sep-
arate 'participatory' research method, this is misleading. The idea
of participation is more an overall guiding philosophy of how to
proceed, than a selection of specific methods. So when people
talk about participatory research, participatory monitoring and
participatory evaluation, on the whole they are not discussing a
self-contained set of methodologies, but a situation whereby the
methods being used have included an element of strong
involvement and consultation on the part of the subjects of the
research. Not all methods or groups are equally amenable to
participation. For example, some of the technical methods
described later can only be carried out and interpreted after a
lengthy training, and although anyone might undergo such a
training, it is usually the case that the research is needed now,
rather than in several years' time.
We must retain a critical and sometimes sceptical view of
processes described as participatory, particularly because many
national governments compel their citizens to take part in devel-
opment programmes. The word 'participation' has been usedby
enthusiastic practitioners to include anything from obligatory,
through to genuinely democratic and enthusiastic, involvement
in a research project. Participation should be empowering, not
merely a process whereby people co-operate without complaint.
A sense of the extent of participation can be had by reviewing
the number of stages at which local people are involved in the
research process. Does participation occur at all stages of the
research (design, collection of data, interpretation, discussion
and presentation) or only at certain points in the process? And if
it does not occur at all stages, is the local community really dis-
advantaged by this? How busy are they? Do they really want to
be involved at every phase of research, given that they may
have crops to weed, children to feed, fuel and water to find, and
dozens of other daily tasks? Perhaps their participation should be
strategic, rather than total? About the big issues, rather than the
nuts-and-bolts?
Where there are limitations on the degree of participation, is
this for unavoidable technical reasons, or because the
researchers want to maintain their exclusive control over the
process and data collected? That is one of the more important
questions to be considered. Imagine how you would react to a
group of people who arrived in your neighbourhood saying that
they had a scheme for improving the quality of your life, but
requesting you to stop whatever you normally do, and take part
in a week-long exercise reviewing the neighbourhood's
resources through a large-scale house-to-house census. You
might well suppose that you know enough about these matters
not to wish to give so much of your time to such a project.
It is important to establish at an early stage the nature of the
research participation expected and desired. Unrealistic expecta-
tions should not be encouraged. It is important to establish
whether participation will go as far as allowing the subjects
overall decision-making power or whether you intend to reserve
some decisions. By involving people in genuine power-sharing
exercises they can be indirectly empowered through widening
their access to information, and by being encouraged to express
their views.
Recent experience has shown that, where it has been well-
managed, participatory research can in itself be an empowering
process. It can assist a group to work through questions, issues,
and future possibilities. It can help them to improve their under-
standing of the society and culture in which they live and the
alternatives open to them. At a more mundane level it can help
a group establish priorities for action in very practical ways.
Perhaps the most serious question to ask before initiating a
particular style of research is in what ways the community will
benefit from the exercise. Suppose you are trying to decide
between full continuous power-sharing participation, or a form
which involves the beneficiaries in less intensive involvement. If
the target group are people with time on their hands, little self-
confidence, and a need to feel that they can act effectively (e.g.
a group of unemployed young people), then a fully consultative,
fully participatory research project might do something signifi-
cant to enhance their self-esteem and skills, and thus become
genuinely empowering. But to ask a group of women, who are
already struggling with child care and the management of farms
because their men are away working as migrant labourers, to
devote precious time to step-by-step research participation might
be to ask too much of them. They will be able to advise you on
this, and you need to be sensitive to nuances here. They may
wish to show enthusiasm for the project and, in hope of your

10
agency's financial support, may be ready to say that they can
devote time and energy to research. It will be the way they say
this whether eagerly, or with a marked lack of enthusiasm
which will suggest how they perceive the task ahead. This will
need to be discussed sensitively and in detail with them, so that
a realistic assessment can be jointly agreed.

Community self-diagnosis: One simple system for community

diagnosis entailed a community meeting together and listing
major health problems in their area (respiratory disease, TB,
skin infections, intestinal problems). The group then listed
separately possible solutions (build a health post, introduce
piped water, organise rubbish collection). The two lists were
then the focus of group discussions about how the solutions
might or might not improve the health of the community. The
results showed that often the most obvious or commonly stat-
ed solution, in this case 'build a health post', was not neces-
sarily the most effective, or the least expensive, way of
improving health in the community. Instead, to improve water
supplies, sanitation, and rubbish collection were more likely
to improve the residents' health. The community learned a lot
about the major health problems and a great deal more about
the range of solutions available to them, as well as the reali-
ties of costs and available resources within the community to
meet these needs.

The issue of participation is also dealt with in Chapter 4, when

different research methods are described in detail.

Expectations aroused by research

Traditional 'top-down' research by outsiders tends to raise local
people's expectations, and they may expect it to solve long-
standing problems. People may assume that basic research will
inevitably lead to money being spent locally in a development
programme. The different interests of the people involved may
create very different expectations of the research findings and
the action these may lead to. The best way of preventing unreal-
istic expectations is, of course, to involve the local group fully in

11
the planning and execution of the research itself, whenever this
is practicable.
Not only is a local group likely to get unrealistic expectations
from a research project, but the agency staff themselves may fall
victims to the same mystique. The pressure to move automatical-
ly from research to commitment must be resisted. You should
never allow the research to become a prescription for action if
this is not fully justified.

Agricultural programme in western Sudan: The local represen-

tatives of two agencies identified possible action related to
irrigation and seed banks as a priority for rural development
during a short visit to an area in Sudan. They came to these
conclusions without serious and systematic research; and this
was their first mistake. They then started a base-line survey of
the area which assumed that both irrigation and seed banks
would be supported. The second mistake was to use this
biased survey as the basis for project intervention instead of
treating it as the base-line document for beginning to think
about a project intervention. Various programmes were then
started and technical staff came and went, but much of this
activity did not relate to the initial survey, although it was this
survey which acted as the justification for the development
interventions. After nearly two years a second consultant was
hired who was obliged to reformulate the programme com-
pletely by referring to the original survey, cross-checking the
initial assumptions upon which the programmes were based,
as set by the two agencies, and then redesigning the whole
programme. The premature pressure for some sort of devel-
opment intervention led in this case to a lack of correlation
between the original consultant's survey reports and the actu-
al implementation of the programme, which was haphazard
and did not take into account the information supplied in the
reports.

To avoid raising local expectations take care to explain who

you are, and what your agency stands for. Always try to make it
clear that the research might n6t lead to action.

Action research in Malawi: An Oxfam-sponsored research

12
 exercise in rural Malawi was designed to identify the major
problems and priorities of some of the most vulnerable peo-
ple in a district. It was realised that a major problem would be
that the research would raise people's expectations. The
researchers were specifically trained to ensure that they made
it as clear as possible that they were there to learn from peo-
ple, not to provide immediate assistance with their problems.
The researchers had to practise their explanations in front of
colleagues to ensure that they could avoid anything which
might raise false expectations. The research team explained to
the villagers that the research was intended to assist the gov-
ernment and other agencies to improve development plan-
ning in the country and district rather than to provide immedi-
ate help for the six villages covered by the research. But in
spite of all these precautions, people still hoped the research
would lead to immediate benefits. (See also Appendix 6.)

Constraints on conducting research

This section deals with some of the main constraints on
research, which is never carried out under perfect conditions;
there are always difficulties to be overcome.

Government attitude
The right of an agency to operate in another country depends
upon the will of the host country. Permission to do research
should normally have been part of a Country Agreement which
set out the conditions under which the development agency
operates. If for some reason research has been omitted from the
agreement, or had conditions attached to it, then negotiations
might have to be undertaken with an appropriate Ministry to get
permission to do research. If political conditions in the country
generally, or in the area you are working in, are good, and your
agency has a positive image with the relevant Ministry, research
permission should be readily forthcoming. But if the country is
riven by bitter internal conflicts, or your agency is operating in a
climate of mistrust and suspicion, then a relatively simple
research project may present itself to official minds as a major
diplomatic issue, and you will need all your patience, tact and

13
persuasiveness to get the project accepted. Under such difficult
conditions, research sometimes simply 'isn't on.'
Alternatively, you may take the view that anything which is not
explicitly forbidden by your Country Agreement, and which is
covered by your general statement of agency aims, is clearly
defensible, and go ahead with your research without seeking
official clearance. This strategy might initially save you time and
trouble. But be prepared for misunderstandings, and be ready to
explain your research in a calm, and diplomatic way if an anx-
ious and ill-informed official from the police, for example
decides that you are engaged in some kind of spying activity, or
something else harmful to the interests of the government. Social
researchers are used to having their inquiries misunderstood;
and the idea of the spy has diffused all over the world.
It is a better tactic, if you want to make a rapid start and cut
through the red tape, to notify your most friendly senior contacts
in an appropriate Ministry informally that you wish to do some
research on a specific topic agricultural productivity, or
mother-child health issues in a particular locality, so that in
the event of misunderstandings you can say to inquirers, 'Mr X
and Mrs Y. at the Ministry know what we are doing here, and
can tell you about our work.'
In short, government, whether through policy directives or the
attitude of particular officials, can often be a significant source of
constraints on research. It is inevitable that some of the activities
of foreign development agencies, in particular, should arouse a
certain amount of misunderstanding and mistrust; and one of the
best ways to counter this is by the greatest possible openness.
Try to collaborate with sympathetic officials wherever possible.
The style of research can also be a factor: a major formal sur-
vey with a large number of interviewers will have a high profile
and inevitably attract official attention, whereas a research
method such as participant observation, or informal group dis-
cussions will normally attract far less attention. Try to be aware
of issues or questions most likely to be controversial, or politi-
cised. Ensure that good local contacts are maintained. Try and
carry out as much of the research as possible in a low-profile
manner.

14
Conflicts and emergencies
Civil wars, famines, earthquakes and other highly disruptive situ-
ations create special problems for research. There is a tendency
among those directly involved to produce highly-coloured, parti-
san accounts of what has been happening and, as has been said
many times, in a war the first casualty is the truth.
It is only possible to get a useful account of conditions in a
war-zone by patient informal interviewing of refugees, but this
will need careful scrutiny and cross-checking to compensate for
any elements of exaggeration which may very understandably
have crept into the accounts given by traumatised people. Such
information may be very important if you need to gauge the
extent of food shortages which may be threatening, or the pro-
portions of local populations who have fled from their homes.
By careful cross-checking about dates and times at which people
left their communities, and asking how many people had not yet
left, you can form a picture of the extent of the flight. By use of
regional census information, and with the help of sympathetic
and knowledgeable local officials, you can build up a picture of
the total numbers of people leaving an area, for whom some
kind of relief provision may be needed.
In situations of major disruption, relief workers are likely to
make serious over-estimates of the scale of disaster if they do
not carry out normal cross-checking procedures. In general, the
more serious the emergency, the greater the need for cool,
thoughtful appraisal of the extent of the threats to people, food,
crops, and water supplies. Nothing is gained by taking a few
unconfirmed and sketchy reports at face value, and calling for
major, costly relief programmes. It is better to take a little more
time, research more thoroughly, reduce the margins of error by
a variety of cross-checks, and then call for a more realistic level
of outside assistance.
The most important information and analysis, however, will
not be based on the refugees' immediate experiences at all, but
will be concerned with that has been happening and will be
happening politically in the zone from which they have fled.
The agency will need to understand why they have fled, what
caused the underlying problems, and what are the implications
in the short, medium, and longer terms. The two basic questions
will be, is there a likelihood of their being able to return to their

15
homes at any foreseeable point, and how that likelihood or
the lack of it will affect the planning of programmes of provi-
sion.

In 1982 when 50,000 displaced people from Uganda arrived in

neighbouring Rwanda, the Red Cross reported that 20 per
cent of children under five were in a state of serious malnutri-
tion. This figure was accepted by UNHCR and by the govern-
ment, neither of which had reason to doubt the Red Cross
findings. The data were transmitted to NGOs, and plans were
made for a massive response based on emergency feeding
procedures.
An Oxfam doctor managed to reach the camp within a cou-
ple of days and pointed out that the 20 per cent figure was
unusually high and, if it were true, was an indicator that the
whole refugee population was in a very bad way indeed.
Further enquiries revealed that the Red Cross data had been
collected at the emergency health post in the camp. In other
words, the sample was severely biased because the only chil-
dren to be taken to the health post were children who were
already ill, many of them with intestinal complaints. A random
nutritional survey was put into operation and within half a
day reliable data were available which put the nutritional sta-
tus of under-fives at a figure which was normal for southern
Uganda at that time of the year. The nutritional status of the
whole population was therefore acceptable. Plans for the
emergency feeding were scrapped.
This case history illustrates two points: that in an emergency
people can accept data uncritically and appear to want to
believe in a worst-case situation; that in a crisis,fieldworkerscan
set aside normal and sound research techniques and in this case
it took two days before anyone pointed out that the data were
skewed because of the poor sample which was measured.
In situations of social disruption, you must take particular care
not to put people at risk by asking them to collect information
for you. Information which in peace-time is innocuous, such as
the whereabouts of a village, or the number of people resident
there, may have alarming implications if there are guerrillas or
counter-insurgency forces operating in the area.

16
Cultural factors: gender relations and
power structures
In many societies there is a structural inequality which militates
against the participation of certain groups in the research
process. There is often, to use Freire's term, a 'culture of silence'
a passive resistance to outside intervention which can easily
manifest itself in a lack of co-operation with researchers. This
lack of co-operation may well be based on bitter past experi-
ences and it might be necessary to identify the origins of the
problem before progress can be made and useful research car-
ried out.
Access to certain groups is often restricted, for example, non-
local researchers' access to women in some cultures; and the
physically or mentally handicapped are likely to be kept hidden
from outsiders in their own homes. The silent members of soci-
ety are often the very people you need to hear from, and
include in your work. (See Appendix 5) It is important to identi-
fy who these people are and where they are. Such people need
to be given the opportunity to express themselves, which means
that the researcher needs to develop a style which permits this.

A woman evaluator from India, who was used to working

with women in a semi-closed society, was able to achieve in
Sudan what she was told would be impossible. She succeeded
in holding meetings with large numbers of poor women, and
getting them to speak openly about their views on their eco-
nomic role. Previously the women project staff had accepted
the traditional view that women would not and could not
meet for such discussions, because inevitably the project staff
were themselves a part of that same tradition, despite their
professional roles.

For people who do not usually have the opportunity to voice

their concerns, research can be very positive and enabling in
itself because it can encourage such people to articulate their
needs. Sensitive research can help give the voiceless a voice.
Participatory research can sometimes lead to actions which
break with tradition, and in doing so it may be empowering for
groups involved. This in turn may lead to hostile reactions from
those losing privilege or power. The researchers need to foresee

17
such tensions, and think carefully about the balance of benefits
which will result.
Where access to a group is more than usually difficult, extra
time will be needed to develop the appropriate contacts, build
trust, and overcome inhibitions at both a group and individual
level. The methods used in the research need to take account of
cultural differences. It is a matter of judgement how far it is pos-
sible or desirable to work against the grain of local customs and
perceptions regarding certain gender issues. A group working in
Maharashtra divided the health team into a male group and a
female group in order to discuss the same issues with separate
groups of men and women in the villages where they worked.
This proved to be an effective strategy. In a different context,
where women might have something to gain by discussions
being held jointly with men, a more challenging and innovative
strategy might be tried.
In some cultures the whole idea of asking direct questions may
be unacceptable or even the height of rudeness. A special
approach will be needed to cope with such cultural constraints.
(See later section on Asking Questions, p. 53)
There is no escape from the problem of power in development
research, and it has two distinct aspects: the power structure of
the society being investigated, and secondly, the subculture of
the development agency itself.
The national and local structures you are working with will
involve differences of wealth, security, and influence between
groups. This will usually be most noticeable in terms of owner-
ship of crucial resources land, water, animals, vehicles, hous-
es, and businesses. The 'ownership' of government and military
jobs may also be crucial. Leaders, chiefs, notables, professionals,
or simply people of high rank, or particular ethnic or tribal ori-
gins, will be making decisions which fit their own needs and
interests, and affect the people you wish to help in all kinds of
ways.
You will usually be working to help the people who lack
power and influence, and who are subject to the decisions of
the power-holders. Many of the problems faced in a local area
are made worse by the pervasive differences in political and
economic power e.g. the system of land ownership, the low
wages paid to day labourers, the use of thugs to intimidate those

18
attempting to organise labourers. You will frequently find that
attractive development initiatives will run up against the power
structure, or cause friction with it. Local people will be the real
experts on how the system of power and influence really works
at grassroots level, and therefore which initiatives will be feasi-
ble, and which will not.
Any development agency has its own cultural and political val-
ues, which are partly external to the local society and partly
involved with it, since the agency inevitably becomes a player, if
usually a minor one, in the local power structure. An agency
tends to recruit as staff people who are often relatively highly
trained, and middle-class in the national system. Both expatriates
and locals may have a commitment to helping the least advan-
taged groups, but when viewed by the local poor, all theagency
staff may appear as wealthy, powerful and secure, even though
in their own eyes they feel none of these things.
It is sometimes argued that middle-class people cannot do
effective and sensitive research on the problems faced, by the
poor because their own backgrounds mean they have not expe-
rienced poverty. We reject this argument. Membership of a par-
ticular class is at best only a rough-and-ready guide to how peo-
ple think, and to their moral values. The most important ques-
tion to ask about a particular researcher regarding their appro-
priateness and competence to work with poorer people is not,
what is their class origin, but what are their human qualities? Do
they have empathy? Are they sensitive to the needs of others?
Do they have the imagination to grasp what life is like when, for
example, all fuelwood and drinking water have to be carried for
half a day by the family who uses them? Above all, good
researchers have to be good listeners, and be willing to control
their own egos sufficiently to grasp what other people can tell
them, and draw appropriate conclusions. These skills are not
confined to any particular class or subculture.

Resources
Time, money, and energy will all be consumed in a research
project. For any research you need to ask if staff can be spared
and if money will be available and energy forthcoming, and you
need to ask if the community being investigated can be called
on to assist, at the particular time envisaged. This requires

19
detailed knowledge of the seasonal patterns of community
work. In agricultural communities, the two busiest times of the
year are the planting and harvesting seasons, and it is a safe
assumption that no-one in such a community will have any time
for you during these seasons. It is far better to plan your
research for one of the slack periods of the year, and community
members will be able to tell you when this is; and, as we shall
argue later, you need to study a community at several points in
the productive year, to avoid distortions of various kinds.
There are a wide range of different methods available to assist
in the collection of information. Some of these are very simple
and inexpensive, whilst others entail complex planning and
some statistical knowledge and can be very expensive and time-
consuming. In the choice of an appropriate method it will be
important to take account of the needs, the expense, and the
time available before deciding. The large-scale sample survey,
where hundreds of people need to be contacted and inter-
viewed, is one of the slowest and most energy-costly methods
available, not only because of the data collection phase, but also
because the analysis of the data contained in several hundred
questionnaires may take months rather than weeks. Supposing
the essential information could have been obtained for one-tenth
of the cost, and in a quarter of the time what 'price' the com-
prehensiveness and rigour of the survey? It would not be sensi-
ble to spend 2,000 on researching a project which will distrib-
ute another 5,000-worth of help to 25 people. But on the other
hand, if very large sums of money might be committed to a pro-
ject, the extra time and money needed for a well-designed sur-
vey might well justify themselves. It might be better to spend
5,000 on a survey and learn that it would be unwise to commit
100,000, than to 'save' the initial 5,000, carry out more impres-
sionistic research, get the wrong answers, and then see 100,000
wasted on a failing project.
So, to summarise the resources issue: the more the project will
cost, and the greater the impact it will have on other people, the
more ready you should be to spend what is needed by way of
feasibility research. If in doubt, keep the research costs to below
10 per cent of the projected project costs.

20
 STRATEGIC ISSUES IN PLANNING
2 SOUND RESEARCH

Suppose you have decided that research is needed, and some

issues have been clarified with the help of local people. Before
committing yourself to one or several of the possible methods to
be discussed in Chapter 3, there are some preliminary matters to
be decided, each of which will have a bearing on what you do
and how you do it. These involve deciding whom to focus on,
your 'unit of analysis'; thinking about when and for how long
your research will be conducted, and the implications of the
time limits; controlling for bias and making the research properly
representative; avoiding certain common pitfalls; choosing the
researchers; and styles of research.

The unit of analysis

There are two distinct issues to be considered. The first issue is
easy: it involves having a clear sense of where the study begins
and ends, the cut-off points. This is largely a matter of being cer-
tain that you are studying the five villages on the east side of the
mountain, but not the ones on the north, south and west sides.
Or that you are concerned with children from one to five. Or
disabled persons, defined by some specific criterion of disability.
This is because research must end somewhere, and if you are
not careful, local pressures may make your research net expand
to cover ever larger populations, because of the 'arousal of
expectations' issue mentioned in the preceding section.
The second issue is more difficult: it involves the strategic level
at which you will study the research question individual,
household, community, or institutions. The decision to make a
particular unit of analysis the focus of the research is all-impor-

21
tant. The research questions very often point directly to specific
units: if you are concerned with mother-and-child health, then
mothers will obviously be primary informants, and so indeed
might children. But mothers are not all identical. First-time moth-
ers may have a different set of problems from mothers who
have had several children; and grandmothers may have different
views on child care from their own daughters, themselves now
mothers but who have been exposed to more schooling. The
health-care workers who will be dealing with the mothers and
children will also have information, and may be divided by age,
specialisation or experience. So, although 'mothers' will have to
be key figures in your study, they will need to be separated into
several types of mothers, and the health-care workers will also
need to be interviewed.
The household is far from being a straightforward unit of
analysis. In the past, researchers were happy to talk about 'the
household' as if it were a unit which pooled its members' inter-
ests, and met their needs in a fair-minded way. But recent
research shows this assumption is often questionable, for two
different kinds of reasons. First, households come in many dif-
ferent forms, composed of a variety of persons in all kinds of
roles. Some households may contain an unmarried brother and
sister of advanced years. Some households may have two gener-
ations, and three married couples. Some households may be a
man, and his wife, and one widowed parent of either husband
or wife. Many households consisting of a woman and her chil-
dren may be headed up by the woman for 10 months of the
year, while her husband if she still has one is away work-
ing as a migrant labourer. So it is unwise to assume any particu-
lar composition to 'the household', and any particular profile of
producers and consumers. Some households have land but not
enough adult labour; some households have labour, but no land.
Some households have few producers but many consumers.
Another major reason for not assuming too much about the
household in a research setting is to do with gender relations.
The income enjoyed by a household may sometimes be distrib-
uted between all members on the basis of need, but quite often
there are cultural rules or expectations about who gets what. In
some cultures, men get first choice of the available food, and
women and children eat later from what is left. In such cases,

22
men need not be consciously aware that they are taking at the
expense of the rest of the household; or the custom may be
rationalised by men and women as essential because men are
culturally defined as warriors, or farmers 'doing the heaviest
work'. A man may feel that cash earned by him through crop
sales or paid labour is his to keep, consume, or give out to his
wife as he sees fit. An increase in his income may not be shared
with his wife and children. He might decide to buy a bicycle
with it, to reduce his travel time when going to work. A woman
may have earnings she is unwilling to give to her husband,
because she uses the money to feed and clothe her children, or
pay for their schooling. The researcher cannot, therefore,
assume, that sums of money which reach one member of the
household will necessarily be shared with others. A benefit tar-
geted at mothers-and-children should be paid directly to the
mother.
It is a commonplace finding that, although households may
benefit from irrigated agricultural schemes, in that their cash
incomes are higher than before they joined the scheme, there
may at the same time be a decline in the nutritional status of
their children. This is for a complex of reasons: such as less food
being grown by the family, the range of locally available foods
for purchase becoming smaller, and the women having less time
to prepare food. Yet, from the perspective of national economic
planners, such a 'household' has now become 'better off.
. The identification of the poorest categories, households, or
individuals presents particular problems for research, and these
will be discussed again in later sections. First, simple visibility:
are the poor people in the research locality physically present
when the researchers are active, or are they away, as labour
migrants, or spending their days and nights far from where the
research team are, watching over browsing animals, or gathering
natural resources in marginal areas? Secondly, the poorest are
often forgotten by their more fortunate neighbours, so relatively
casual inquiries may not identify them. Thirdly, people who
think a research group may be willing to help them, may
describe themselves as the poorest, although there are others
poorer than they are. Various methods can be used to overcome
these difficulties, and we discuss these under Rapid Rural
Appraisal, (p. 66) and the use of indicators (p. 78). At this point,

23
we will simply note that mapping a community, dwelling by
dwelling, and finding out the names of all the people who live
in each dwelling is a way of making sure people are not over-
looked, and this can be followed up with a wealth-ranking exer-
cise. Both the mapping and the ranking work can usually best
be done with help from local people. (See also Participatory
Rural Appraisal, p. 74.)
To take a different kind of example: suppose representatives
of a remote community were seeking a loan to buy a vehicle
which would allow the members of the community to visit a
market town, for economic and health reasons. It might be
important to know who visits the market town, how often, for
what purposes; and how people think this pattern might change
if there were a vehicle making the journey once a week. Then,
too, it might be important to know how often people would use
the vehicle at specific prices for the journey.
There are all kinds of possible pitfalls in this apparently simple
set of issues. If we assume too quickly that the unit of analysis
must self-evidently be adult villagers as individuals we may miss
the importance of the fact that the individuals usually live in
families and households, and that their travel decisions might be
altered by whether or not another member of their household is
willing or able to make the journey on the same day. In prac-
tice, to cut down on expenses, it would be probable that mem-
bers of the same households, or even of closely-related people
living in separate households, would 'run errands' for each
other. So a questionnaire based solely on a village of 300 adults,
treating them all as discrete individuals, would run the risk of
seriously overestimating the numbers of paying passengers the
vehicle might attract, and thus put the economics of its opera-
tion on shaky foundations. Yet, on the other hand, to assume
that there should be a single questionnaire answered by a repre-
sentative of each household would risk losing sight of gender
distinctions. A man might see himself as the person likely to use
the vehicle, and think of his wife as staying at home. The wife,
for her part, might have definite reasons for wanting to make
her own journeys independently, particularly if the market town
has a health clinic and she has children.
There could be various solutions to such a problem. One
course of action would be to ask half the village as individuals

24
(half male, half female) and the other half as households (asking
the most active man and woman in each household, but inter-
viewing them separately), and compare the differences in
responses and estimates, if any. A second solution would be to
interview individuals, but to include a question to everyone on
the lines of 'Would you still be likely to make this journey once
a week (month, quarter...) if another member of your household
were already going?' There could then be a comparison of the
pattern of answers between men and women informants.
But there is one serious problem about all these approaches,
and that is that, because they are a study of future activity, they
rely on statements of intentions by asking hypothetical
questions, such as 'How often would you be likely to visit the
market town?'. Experienced researchers try to avoid the use of
predictions and hypothetical questions (ones which use the
word 'would'). Sometimes such questions are unavoidable, but
they need to be used with caution, and you should not rely
solely on answers to them for your information, but always try
and cross-check.
One way out of this dilemma would be to supplement the
study of the particular community with a 'control study' of a
neighbouring community, as similar as possible in its social com-
position in terms of numbers of farmers, artisans, and landless
people, which already possessed a mini-bus service. Thedata on
actual observed usage would probably be a better guide than
hypothetical statements of intent. The researcher should, ideally,
travel on the vehicle a number of times, noting who was travel-
ling and asking them for the reasons for their journey. (We
deliberately put 'reasons' in the plural, because people very
often have several compatible reasons for doing something, and
a good question should never assume there is only one.)
Moving to larger units, such as a whole community, a co-oper-
ative or a cluster of neighbouring villages, brings new decisions
about units of data collection and subsequent analysis. If you are
dealing with a population of farmers, for example, then class,
here understood as the size of land-holding, will be of signifi-
cance, and you might wish to think about the different con-
straints which affect the landless, small, medium, and larger
farmers. Your unit of analysis will need to take these differences
into account.

25
 In a study of the needs of rural people in Botswana for fuel-
wood, an area was chosen in which wood around the village
had been used up. One of the objects of inquiry was to find
out how people obtained fuelwood, and who was most likely
to buy it.
The results were surprising. If the researchers had relied on
a few casual conversations with some of the more prominent
members of the community, they might have heard the views
of the richer farmers only. These people own tractors, and
they never purchase wood on the market. Instead, once every
few months, they send one of their sons off with a tractor,
who drives a long distance and hauls back a large supply.
Families without tractors tend to gather wood more frequent-
ly, and from closer at hand. The poorer families have to send
someone to forage for wood every 24 hours. But the category
most likely to purchase wood turned out to be labour-poor
families, who simply could not spare anyone from the most
pressing food-producing and income-generating activities, and
in particular, old women living alone. They could not make
the long journeys needed, and had no-one to send. If they did
not buy wood, they could not cook their food.
In this study the unit of analysis was firstly, the household
but secondly, households reclassified by income and by
amount of labour power. A small selection of households, not
subdivided in this way, might have missed either the rich or
the very poor, and in each case their actual behaviour went
rather against the grain of common-sense expectations, which
would be that the rich were most likely to buy fuelwood, and
the poor, least likely.
Choosing the right unit of analysis is crucial to the success of
any research exercise, and it must be defined at an early stage of
the research. You may decide that you need to carry out a spe-
cial pre-research exercise, a 'pilot study', to establish the best
unit of analysis for your purposes.

Questions of time
An obvious limitation on any data-collection effort is the time
available in which to carry out the work. The first question to

26
 answer therefore is how much time have you got? It will often
be less than you would like. The time available may be limited
by the contracting agency, which needs information in a hurry
to enable it to continue with its programme, or other constraints
such as access, weather, and, of course, money; so the next
important question is how much research and research time
can you afford? The answer to this question is also likely to be
pragmatic and have more to do with the priority other people
(funders, government officials, senior managers) put on your
research. The resources at your disposal are unlikely to be ade-
quate for the job in hand; but this is the reality for fieldworkers
the world over, and what distinguishes them from people carry-
ing out individually-determined academic research.
Another time-related question is at what point research should
be done. Experience has shown that it is never too late to start a
research programme, even in a project that is already function-
ing. Existing staff can be re-trained and directed towards a more
thoughtful collection of information, monitoring, and reassess-
ment of their activities. If it is never too late to start research, it
is also much better to start early. Relief and development pro-
grammes may begin in a hurry, especially in an emergency, and
if adequate information is not collected, misunderstandings may
develop.

In southern Sudan an emergency water programme for

refugees was started on the assumption that the need was for
three months only. After nearly two years the programme was
evaluated, and the evaluation report highlighted many impor-
tant points that had previously never been understood, such
as local beliefs regarding the siting of water sources, and
revealed a much higher participation by local people, as
opposed to refugees, in the water programme than had been
originally envisaged. The moral of the story is that, although
the initial intervention was to meet emergency demands, the
process of research and investigation should have started with
the start of the water programme, which might then have
operated more effectively.

Some research entails a close look at seasonal changes. A sur-

vey carried out at one point in the year will miss a great deal

27
which is of importance. Data collected shortly after a harvest will
be very different from data collected after a period of shortage,
particularly regarding nutritional status, disposable income or
assets, and social relations. So research will have to be planned
to report on at least two seasons in a year if it is to be reliable
and comprehensive.
Migrations of labourers, nomads, and others may be seasonal,
which means that it is only too easy for interviewers to arrive in
a community and be unaware that many people are missing
who are working elsewhere at that time of year. A static survey
which does not take production and residence differences over
time into account is of limited value. But it is not necessary for
the research to be spun out over a full calendar year, since local
informants will always have a very lively and clear perception of
the major changes which occur cyclically. It will help the
researchers to understand these changes if they sit down with
the local people and work out the year's cycle, and note what
happens during each month, the local names for the different
seasons and the climatic and residential changes which occur.
The research timetable
Time is required to establish trust and to build up a fuller
knowledge of a community. If you are interested in attitudes it is
of particular importance to allow ample time for relationships to
develop to the extent that attitudes can emerge and local people
allow the researchers a sufficiently accurate and intimate knowl-
edge of the community, rather than a 'public relations' image of
the community as problem-free.
When you are working with certain types of group you will
need to allow more time for the research programme; it may
take longer to recruit and train research assistants, to win a
group's trust and gain access to its members. Language may cre-
ate problems: the secondary materials may be more difficult to
locate.
Many researchers underestimate the 'set-up time' needed for
preparation, training assistants, and obtaining research permis-
sion; they then find that at the end of the exercise they have run
out of time to do the all-important analysis, so are obliged to
write a report full of undigested material, of little use to anyone.
It might well take a couple of months to draw up job descrip-

28
tions, advertise, recruit and have people ready to start work. It
will then take a couple of weeks to train them in the approach
and methods being used.
It is possible to estimate the time required for some proce-
dures with a little more certainty. For example, suppose you find
out by running a pilot, or trial, that a straightforward question-
naire or survey takes half an hour; if you require a survey of
2,000 people to be carried out within ten days, you can calculate
that you will need 33 fieldworkers to cover the 2,000 people
selected as a sample, assuming that each fieldworker manages
six interviews a day, and works ten days in succession without
an official rest day. (33 interviewers x 6 interviews per day x 10
days = 198. The remaining 20 interviews can be done by the
research supervisor as one of a number of 'quality checks'.) If it
is easy to find interview subjects because they are in large, com-
pact communities, and ready to co-operate, then the estimate of
six interviews per day might be raised to eight or nine, and the
number of interviewers reduced accordingly. If surveys are
being carried out at village level it should be possible to esti-
mate the time required for one village, and then multiply this by
the number of villages to be covered. You then need to assess
how accessible people are likely to be and when they will be
available for interviews. It is important not just to include the
most easily accessible people, those at home during daylight
hours or who live closer to the tentre of the village. So you
must calculate the time required to get to the houses further
from the village centre, and to meet a cross-section of the com-
munity. Remember that men and women are likely to be free or
available at different times of the day or in different places.
Having made your estimates of the time your survey will take, it
is probably wise to add some extra time, in case things do not
go quite as smoothly as you expect.
The interviewer is not the only person whose time is valuable:
the people being interviewed also value their time. You should
be aware of the opportunity cost to those being interviewed
what would they have been doing if they weren't answering
your questions? Try to ensure that you ask for the minimum time
possible from your respondents compatible with obtaining the
information required.
Try to assess which stages of the research process could be

29
speeded up: could the writing be done faster? Is time being
wasted by using unnecessarily long questionnaires?
When selecting staff, make sure that appropriate staff are cho-
sen for the different stages of the work: designing and planning,
the actual data collection, and the analysis of results. Some staff
will be quick thinking, and keen to be involved in more direct
action; others are more suited to a slower, sensitive listening
role; others will be better at the report writing, and so on.

Continuous research or a snapshot?

A key question which will relate to the time required or avail-
able for research is whether the research will take place over a
period of time through a prolonged process of data collection or
be a one-off exercise. A continuous research process may occur
as part of normal monitoring, or a series of follow-up surveys,
regular updating of data through new samples, or result from
stringing together a series of 'snapshots' over a period of time.
The one-off or snapshot exercise will give an image of reality
and define certain social facts at a given point in time. Both
approaches have advantages, and sometimes only a one-off
exercise may be feasible due to a lack of pre-existing data, the
need to make short-term decisions, or the lack of resources for
continuous research.
The danger of taking a snapshot view of a group is that it can
often be extremely misleading or indicate a state which is nor-
mal to that group but is interpreted by outsiders as abnormal.
For example, many so-called emergencies are identified as such
by outsiders, when in fact the nutritional status of the group may
always be comparatively poor. Thus, longer-term solutions
rather than a short-term 'emergency' intervention are necessary if
the situation is to be changed for the better. The snapshot view
often does little to help analyse causes, although it can identify
symptoms of longer-term problems. Thus, while it helps to iden-
tify which group is destitute and how many people are in the
group, it may not assist in understanding why they are destitute,
for how long they have been in this state, and how long they
might remain destitute if no intervention were made by the
development agency.
Continuous research allows comparison over time and should
give a more accurate picture of changes and what is normal for

30
a given group. For example, nutritional data collected only once,
which cannot be compared to other information, are not very
helpful in indicating whether the group's nutritional status is
improving, static or deteriorating, and whether the present nutri-
tional status is normal, abnormal, or reflects a seasonal bias. A
project should possess base-line data for the area or group with
which it is working that is, descriptive data about the position
at a particular point in time, from which changes can be plotted.
It would then be possible to compare new information showing
social, economic and other changes, with the base-line informa-
tion and other follow-up information. The textbooks all say that
any programme must start with a base-line survey, and this
would certainly be desirable; but, in practice, data collection
often comes at a time when programmes are already function-
ing, as a part of the normal monitoring process. Although a tra-
ditional base-line survey may not exist, if the material collected
during a project cycle is consistent, in that similar information is
collected in similar ways, then at least this will provide for the
possibility of making comparisons over time and will enable
decisions to be made as to changes in the target group and the
functioning of the programme.
In addition to making comparisons over time, it is sometimes
possible to compare over 'space', by looking at information from
similar units. For example, one could compare literacy levels in
two villages, the first having joined a development programme
ten years previously, the second only recently. It might be found
that the level of literacy in the second village is similar to that
encountered in the first village at the time it first joined the pro-
gramme ten years ago. If the village with a well-established pro-
gramme now demonstrates much lower rates of illiteracy, it is
possible to make meaningful and positive comparisons to the
village which remained outside the scheme. Using such a
method, villages newly entering the programme act as 'control
groups' against which to measure the progress of other villages
in the scheme.
The processes of monitoring and ongoing research activities
should allow for an assessment of change over time. Included in
this monitoring should be an attempt to assess what the impact
of the programme really is, and separate this from other factors
both internal and external to the community. For example, an

31
evaluation of a programme designed to strengthen the food
security of a nomadic group should try to assess the impact of
the programme on the overall food availability, but will also
need to take into account internal factors, such as the decision
by more families to enter settled agriculture, and external fac-
tors, such as the supply of food aid by other agencies or govern-
ment control of grain prices in the area.
A balance can sometimes be found between continuous and
snapshot approaches to information gathering. Certain types of
information will lend themselves to a snapshot approach, but
these snapshots can then be incorporated into a longer-term
view. For example, a technical survey of the physical availability
of water in an area could be carried out and used in a longer-
term survey of changing water use in the same area.

Objectivity, subjectivity, and the control of

bias
All thoughtful research seeks to control for the biases of both
researchers and informants, and a great deal of the advice given
in this book is concerned with this issue in one way or another.
But professional researchers disagree about just how far in theo-
ry or practice it is possible to give an 'objective' account of what
is going on between groups of human beings. For some, such as
economists, economic historians, experimental psychologists, it
should in principle be possible to describe and analyse what
human beings do in such a way that errors of fact are eliminat-
ed, so that a panel of independent observers would agree that
they had seen a man bite a dog, a motorist knock down a
pedestrian, a policeman take a bribe, or any one of millions of
possible events.
But other researchers, particularly anthropologists, social histo-
rians, and sociologists interested in religious or political move-
ments, tend to be sceptical about the reliability and validity of
the 'objective' accounts just discussed. They point to the difficul-
ties often experienced by the judges and juries in a court of law
in deciding what happened in a legal case, to various philosoph-
ical and perceptual problems to do with how far two individuals
ever really share experiences of the same 'real world' and to

32
what extent anyone can succeed in communicating such experi-
ences to another person. The headache which apparently inca-
pacitates one person might be shrugged off by another; and a
downwardly mobile person might experience as real deprivation
conditions which to an upwardly mobile person appear almost
luxurious. Consider the loading on the phrases 'slum'; 'shanty-
town'; 'informal low-income accommodation'; 'squatter settle-
ments'; 'our home'. All are phrases being used to classify the
same physical structures, which are objectively present for
description. But each classification bears its own weight of sub-
jective values.
Let us return for a moment to the observers who believed they
saw a policeman take a bribe. If they saw a policeman stop a
motorist, and they saw that the motorist handed over a sum of
money before driving away, they might feel sure that what they
had seen was a bribe being offered and accepted. They might
have overheard the policeman say to the motorist 'I want a con-
tribution from you before I let you go!' and this would have only
confirmed them in their conclusions. And 99 times out of 100,
their conclusion would have been valid.
But supposing that the motorist is also a policeman, but not in
uniform, and that he is in fact being asked for his subscription to
the annual police dinner, then the whole event takes on a com-
pletely different meaning. The observed events really happened,
but their wider context, their true significance, was unknown to
the observers, and they reached a false conclusion.
In the case of the 'bribe', the final piece of evidence which
would have helped decide the matter would have been the
policeman's intentions if he agreed to give an honest account
of them. In this kind of case, we could be clear about his inten-
tions, and there would be a lot of supporting evidence to con-
sider. But even if people are co-operating with researchers, and
telling the truth as they see it, it is not always the case that they
are clear about their intentions.
Imagine three boys playing truant from school. They are clear
about not wanting to be bored in the classroom, but they do not
know how their day will end up, or what their specific inten-
tions are, apart from 'having fun'. Sometimes the participants in
a political or religious movement are protesting about something
they know what they do not want but are not at all clear

33
about the future, or their own intentions. At this point, the
researchers who seek 'objective knowledge' would be in some
difficulties, but those who see their task as the interpretation of
the actor's meanings might be able to offer a provisional attempt
at understanding what the movement seeks to achieve.
The debate between those who are more optimistic about the
possibility of objective research, and those who are more scepti-
cal, will never be resolved, and we do not intend to take sides
here. In principle, all researchers should try to be as objective as
possible in the way they carry out the research, that is, they
should seek for precision and accuracy, and to control bias. But
if they are wise, they will never suppose that they have com-
pletely succeeded, and they will remain open, right down to the
last stages of their research, to the possibility of their having
made interpretive errors of all kinds.
But the critics of objectivity are not therefore free to say, 'Since
real objectivity is impossible, and since I have a committed polit-
ical view, I am free to conduct my research in any way I choose,
and I shall choose a way which directly furthers my ethical and
political values.' They must be just as committed to accuracy,
precision, and the control of bias as the seekers after objective
reporting. The difference is, they may allow themselves to be
more aware of different and conflicting interpretations of the
social world, and prefer interpretations of research data which
present the findings as tentative, and provisional, rather than
conclusive, and written in stone.

Representative results
Good research seeks to give a representative account of the
group of people studied, such that all their views are represent-
ed, and not merely those views which the researchers approve
of or to which they are politically sympathetic. This is for two
distinct reasons. Firstly, it is never good research practice to take
short-cuts, or to leave views out because you think you know
what they are likely to be and so wish to discount them politi-
cally. Secondly, it is necessary to represent all views in a com-
munity because the success or failure of a development project
may depend crucially on the attitudes and interests of people

34
who are not the target beneficiaries of the project. If large land-
lords are employers of casual labour, and your project is
designed to help landless labourers, the landlords' interests may
need to be understood before any scheme to increase the
income of the labourers can succeed. This does not mean that
the landlords will be given any kind of veto over the project
design, but it does mean the researchers need to hear from the
landlords, just as clearly as they hear from the labourers.
There is no easy way for either identifying or eradicating bias,
in interviewer or interviewee. It is up to each researcher to think
about the similarities and differences between himself, or herself
and the people with whose lives the research is concerned, and
to attempt to become sensitive to how far social differences pro-
mote different interpretations about what is going on.
Take time to think about such basic conflicts of interest, per-
ception and reality-definition as are likely to arise between land-
lords and tenants (issues of fair rents, and good amenities),
between pedestrians and car-drivers (issues about taking the
other's speed and needs into account), between employers and
employees (wages, quality of work, working conditions), teach-
ers and students (time and quality of teaching), doctors and
patients (time, and the wish to get better). In all these pairs of
roles, there are possibilities of conflicting definitions of what has
been transacted, but there is also the possibility that a careful
observer could give a description and analysis of the transac-
tions which would satisfy both sides as being accurate, and un-
biased.
One important source of bias involves gender, where men may
be blind to the interests, preoccupations, and working lives of
women. To begin to come to grips with this issue involves con-
scious awareness of the need to have women fully involved
from the first inception of the research project. By 'fully
involved' we mean that women are equal partners in policy
making, research design, and all the other phases of the
research. Superficial and ad hoc consuftation is not enough. (See
Appendix 5.)
Other sources of bias arise through cultural differences
between researchers and informants, which may cluster around
issues of rank, age, social origins, skills, religious differences
indeed, an open-ended set of possible issues.

35
Some pitfalls to be avoided
In an earlier section we argued that it is best to treat the house-
hold as a neutral and unknown unit rather than to assume in
advance that you know how its members will behave towards
each other in matters of economics and domestic budgeting.
Similar care is needed with some other loaded terms class,
solidarity, and community.

Avoid false certainty about people's class: Another kind of

naive assumption involves the labels given to particular classes.
For example, in Latin America people often talk about the 'pop-
ular classes', which may be a convenient short-hand description
but is extremely difficult to define accurately, because it is not
based on any set of agreed, objective criteria. It would not be
possible to decide who is a member of the popular classes by
measuring income or property rights, for example. It would,
however, be possible to ask people whether or not they consid-
ered that they were a member of the popular classes. The results
would be interesting but totally subjective, in that you would
find a range of people who might consider themselves members
of such a class irrespective of their actual economic status. It is a
label which many people are inclined to attach to themselves,
but to withhold from those of whom they disapprove.
Do not assume solidarity: To take another example: you might
decide you want to investigate the income levels of female-
headed households. It would not be wise to assume that there is
any set of common interests shared by such women in a com-
munity simply because they share a common sociological char-
acteristic. It is possible that the women in question might not
even know of other women who are also household heads, or
you might discover marked differences n their income-levels
and economic interests. To sum up: to share a common charac-
teristic is no guarantee of a sense of having 'something in com-
mon', nor of solidarity, nor common purposes; and you should
not make such an assumption in planning research. Indeed, this
is precisely the sort of issue which research might help to clarify:
if the research suggests a sense of solidarity among women
household heads, then something useful might be built upon

36
this bond. But if the solidarity is assumed in some way in the
research design, when it is not in fact present, a project may
founder on its absence.

Do not assume community: Development workers would

often like a physical community a village, a township to
meet their expectations that it is a natural unit of social co-oper-
ation. We see a cluster of dwellings, and allow a romantic fanta-
sy that the physical clustering implies a convergence of interests.
All too often small-scale communities turn out to be full of
groups with conflicting interests, and histories of competition
and even hostility between families and individuals. This does
not mean that they are unable to co-operate over the administra-
tion of some public good, such as a clinic, a well, or warehouse;
but it should signal the dangers of assuming that simply living
near to each other automatically leads to something called 'com-
munity spirit'. Such an assumption may well lead to problems if
certain common interests and characteristics are projected on to
a community before there is any real evidence that they exist.
If research seeks to target a group, it is even more important
that the different levels of group and individual interests are
understood; and that the linkages between them are identified.
To take a simple example: in some cultures brothers (married or
unmarried) find it easy to co-operate in economic matters, and
in others they find it so difficult that nothing positive necessarily
follows from two men being brothers. In another culture, men
may be part of a clan of several hundred members or more, and
may recognise obligations to fellow clansmen, while finding it
difficult to trust men from other clans. In times of civil war and
famine, the 'unit of trust' tends to shrink, and men will fight for
resources against those they co-operated with in better times.
The assumptions about the nature of the target group and how
the group is being defined need to be made clear and explicit in
the research brief and subsequent reports. A system of validation
may be required to cross-check that false assumptions have not
been made about a group's identity. There are many constraints
operating to hide or confuse the perceptions of all actors in this
process; in certain countries, governments may claim to repre-
sent all views and to have organised people into interest groups.

37
Do not read rural patterns into urban contexts: One major
failing of development planners and researchers alike is that
they are tempted to transfer rural models to urban contexts.
Some development workers base their experience and ideas on
rural work and arrive in urban areas with preconceptions about
the degree to which ideas of community and family operate. In
a rural context, a neighbour may be someone who will
exchange help over agricultural task, because both households
have small land-holdings and face common problems. But in an
urban context, it is much less likely that two neighbours are
involved in the same kind of economic activity, and there may
be nothing which predisposes them to help each other.
However, schemes for small loans, which require high rates of
repayment to remain viable, sometimes work better in urban
shanty towns than in rural communities, because the money
economy may be unfamiliar to a land-based society, and has not
become part of the system of moral evaluations.

Avoid over-reliance on particular informants: Strictly speak-

ing,anyone who gives you information can be thought of as an
'informant'. But this section is about a special problem: many
researchers use people they think of as 'key informants' for much
of their information about the community they are researching,
because these are people who have shown willingness to help
them from early on in their first contacts with the community, and
seem confident, knowledgeable, and articulate.
Inevitably, most key informants will have their own biases and
their own perspectives, based on their social standing in the com-
munity, and any information from them needs to be cross-
checked or validated against the views of other people. Some
development workers are strongly critical of any reliance upon
key informants because they believe that this can provide
researchers with an excuse to avoid talking directly to a wider
range of people. It has even been argued that many of the failures
of development planning can be traced to the ill-considered but
widespread use of such informants. It is true that some academic
studies have relied on information from a single informant, when,
for example, a life-history is elicited at length and published by
itself. But such a single focus will not do for a development pro-
gramme where money is to be spent for wider purposes.

38
 Some of the risks of using key informants are evident; if they
are all men, or all women, the views of the other sex may go
unheard. If they are all senior people, then the often sharply dif-
fering views of their juniors may go unconsidered. Secondly,
there is a tendency for certain people to become 'professional
informants', people who make a speciality of talking authorita-
tively to outsiders. They may well feel they know all the answers
before you have started asking the questions, because they have
already worked for every student, aid agency, and survey team
for the past ten years and are certain they know what you really
need to know! If your informant is also your interpreter you are
even more likely to be given a heavily edited version of what is
being said: what the interpreter thinks you need to know instead
of an accurate translation. Interpreters need to be carefully
trained to stop them abbreviating, and 'cleaning up' what is said
by an interviewee. In addition, the professional informant may
often not be from the community itself but someone marginal to
it, from a different tribe, ethnic group or caste, so may not in
fact be well integrated. In conclusion, be wary of key infor-
mants; only use them when you can cross-check what they are
telling you, and try to be aware of their personal biases.
The informant should be appropriate to the information you
are seeking. A male government official posted from another
area is likely to be a good source of information on those things
which affect his job but is unlikely to know much about child-
rearing; whereas the local traditional birth attendant will know
about childbirth and women's health, but less about how gov-
ernment officials check the co-operative's accounts.

Choosing the researchers

The principal questions are to what extent and in what ways can
local people most usefully participate in the research, and how
far will experienced or expert researchers be needed? Let us
assume that a research problem has been identified, with the
helpof the local people or on their initiative. There are three
conventional phases to research: design, data collection, and
analysis. Each stage may require slightly different sorts of people
to carry it out.

39
 For research design it may be advisable to get some initial
expert guidance, whether this means consulting a handbook,
having a discussion with an academic or specialised civil ser-
vant, or writing for advice to an individual or organisation. If the
research is properly designed, it may then be possible to carry
out most or all of it using local people, who may be willing to
participate because they see themselves as having a strong inter-
est in the outcome.

Research on the effects of structural adjustment: In one

African country it was felt that NGOs should monitor the
effects of an IMF adjustment plan on public services and the
lives of poor people. Unfortunately, the instigators of the
research started with an assumption of what the conclusions
would be (i.e. that the plan would be detrimental), and then
sought to find a way of proving it. They were probably right
in their assumption, but it clouded discussions of the method-
ology because they were seeking to provide the proof rather
than to provide information from which conclusions could be
drawn. Secondly, clear terms of reference based on advice
from professional researchers were not produced at an early
stage. Thirdly, it was not recognised that to obtain statistically
significant results very large-scale surveys would be required.
Alternatively, shorter, more impressionistic studies would
require a standardised format and need to be carried out over
a period of years. They would have to include comparative
material collected before the effects of the adjustment plan
were being felt.
The failure to seek professional advice on the research at
the outset and to appreciate the likely scale of the research
required for such monitoring led to several false starts and
lost opportunities. Nearly five years passed with little useful
information being produced and the chance for monitoring
was lost.

In data collection, it is important that the people collecting the

information understand the research design, and that, if not
themselves drawn from the local group, they are acceptable to
them. The local community may be able to supply people to
help, particularly secondary school pupils, teachers, and others.

40
Such people may need to be properly paid even though the pro-
ject is for the long-term benefit of their community. Never
assume that significant free time can or should be given, espe-
cially during busy periods. The issue should be sensitively
explored and negotiated, without preconceptions on the
agency's side.

Local participation and experts

Data collection may need expertise if, for example, it is neces-
sary to carry out testing of blood or water samples. This exper-
tise should be obtained within the region whenever possible.
In a drought-recovery programme supported by Oxfam, a
research project was undertaken to monitor and analyse the
changes in nutritional levels of the population, and the recov-
ery of vegetation and livestock after a drought. Existing food-
aid monitors were integrated into the research programme,
through their collecting information on the communities they
visited. The data these teams collected was collated and com-
pared with other data collected from aerial surveys, drought
and rainfall monitoring stations. The findings were then dis-
cussed in a series of seminars prior to publishing the results
of a computer-assisted analysis carried out by professional
researchers.

There may be a need for specific expert skills if computer pro-

cessing or laboratory analysis are required. During the analysis
stage there may be an opportunity to test out some of the appar-
ent causal links with the local community. If, for example, the
researchers were trying to find out why attendance at health
clinics was low, then the suggested reasons thrown up by the
data-collection phase could be further validated by being
referred back to local people, to check on their plausibility.
If you decide to employ professional staff for some phase of
your research, make sure that the people concerned have a
proven track record in research in less-developed countries, or
have carried out research on behalf of agencies with appropriate
priorities. Sometimes, experts 'travel badly' and cannot improvise
or adjust to a field situation, and to the need to consult ordinary
rural people. The professional should be willing to explain the

41
 research method being used and why that method is the method
of choice, necessary for the needs of the research. Any reluc-
tance to explain, or failure to explain in a clear and understand-
able way, is a bad sign.
Such professionals will expect specific and explicit informa-
tion from you, with clearly written 'terms of reference' compre-
hensively detailing what is required of the research.
Disappointment with the results of research can often be traced
to confused and mixed messages from the commissioning
agency and lack of clarity in the brief provided for the research
team.
Various factors will be important in choosing staff to carry out
the research. These may include their ethnic background, gen-
der, languages, level of education, and experience. Before
recruiting it is important to prioritise these factors in terms of the
jobs to be done. Thus you might decide that it is essential that at
least one interviewer is a Spanish-speaking woman, and then
hire accordingly.
It is always important to consider the impact of using different
kinds of people as researchers. For example, in many societies it
is almost impossible for a man to interview women effectively,
or for women to interview men. Social origins may also be a
sensitive matter: a local researcher with a high status may be
seen as intimidating by low-status informants. Seniority may be
an additional complication: if your field researchers are young
secondary school pupils, because they are literate and have
breaks from the school programme which make them available,
they may nevertheless need special instruction in how to talk
sufficiently respectfully to older men and women who are not
educated.
In certain societies it may be useful to use 'cultural brokers',
people of mixed descent who cut across normal cultural/ethnic
lines and therefore may have better access than usual to the
whole group. In one research programme a tribal woman was
employed who spoke a tribal language but had been taken
away from her community whilst young and had worked as a
maid in different countries, where she learned both Spanish and
English. On her return to her community she was able to com-
municate with government officials in Spanish, American aid
workers in English, and with her community in her mother

42
tongue. She was also more confident than other women in the
community and had no problems talking to or interpreting for
men.

Training the field team

It is often possible for research to be carried out using a high
proportion of local people after they have been given some
basic training. If a survey is being conducted, then the prospec-
tive field staff can perform role-playing exercises, with one per-
son acting as questioner and the other as questioned, in front of
the trainer and the rest of the group, to get them familiar with
introducing themselves, explaining the survey, and taking their
respondents gently through the questions, thanking them for
their participation, and informing them about how they can
keep in touch with the research and its outcomes. If oral history
information is needed about some highly specific issue, such as
changes in local cropping patterns and crop rotations over a
lengthy period, then a 'demonstration' interview can be carried
out by the trainer, and helpers can each be asked to select five
neighbours from the right age-group, interview them, and report
their findings back to a person responsible for writing up the
results. It would not be necessary for the helpers themselves to
be literate, because they will usually have far better memories
than literate people who rely on note-taking, and will have no
trouble recalling, a set of specific responses on specific issues.
If the research involves tasks where amounts consumed need
to be quantified household woodfuel consumption, for exam-
ple then the use of a suitably strong spring balance to weigh
wood-loads can be demonstrated, and the training can include
taking the group through a typical day, with an emphasis on
how, when and where the relevant observations and weighings
will need to be made.
Training a field team for survey, interview or observational
work requires all the steps in the research procedure to be made
explicit, the reasoning behind each step explained, and an
account of how the information will contribute to the over-all
research task. If a questionnaire is to be used, then the trainer
should take the team through it slowly, question by question,

43
discussing the wording of each question, the sorts of answers to
be expected, and the ways in which the answers should be writ-
ten down. Examples of good and bad data recording should be
compared. The best practice is to involve the trainees in the very
earliest stages of formulating the questionnaire, and testing or
'piloting' it with a few respondents, before committing the team
to the final wording. In this way, any problems can be readily
identified and corrected: questions which do not produce the
intended responses will need to be scrapped, and re-phrased. In
any research involving observations and data recording, the
team will need to have untypical or problematic patterns and
responses discussed.
We cannot leave this topic without again stressing one of our
main contentions the need for a sensitive awareness of gen-
der issues. There are crucial questions which must always be
asked about any research: are women involved in identification
of research needs? Are they involved in the control of the
research? Are they involved in the data collection, processing,
and use of the material?
Sometimes, the choice of men or women for a field task will
have some important practical issues which revolve around gen-
der differences.
In a survey of woodfuel consumption in a Southern African
country, high-school graduates from the capital city were
selected to carry out the survey because they had the neces-
sary skills, and were available for hire. Both men and women
were included. The study required several weeks of observa-
tion and data recording in a series of villages. The villagers
were not willing to cook for the researchers, so the research
coordinator provided the research teams with wood for cook-
ing, utensils, and basic foods. In the event, the young men
proved unwilling and unable to cook for themselves, having
never done it before, and tended to drink beer and smoke
instead. In order to ensure the survey, which had a tight
deadline, was completed, the research coordinator was
reduced to cooking food and transporting it to the men. The
women researchers proved willing and able to cook for them-
selves, and also proved much more conscientious as field
researchers. They seemed to have a greater stake in showing

44
 how well they could do the work. The research coordinator, a
European woman, resolved that in future projects of a similar
nature she would hire women in preference to men, because
of their greater independence and reliability.

Styles of research
'Styles' is not quite the right word, but it is difficult to find a bet-
ter one. Briefly, there are two sets of contrasts in the ways peo-
ple organise research: the quantitative/qualitative split; and the
formal/informal split. Instead of trying to be exhaustively sys-
tematic in discussing these, we shall simply suggest how the
issues are related by looking at a single example.
Policy makers who must commit large sums of public money
to social policy issues are aware that mistakes may mean both
wasted money and social disruption or blighted lives. Consider
the problem of working out accurately how many primary-
school teachers will be needed for a particular region in ten
years' time. Since family sizes are liable to change, a serious
over-estimate would mean that teachers are trained and schools
are built but with no children to occupy the places or the teach-
ers. Result: wasted money, and frustrated, unemployed teachers.
But a serious underestimate would mean shortages of school
places, and of qualified teachers, with bad effects on the quality
of primary-school education for the unlucky children. Therefore,
precise predictive, quantitative research is needed to get thecal-
culations right. This must also be research employing formal
research methods particularly the sample survey as a data-
gathering device, and statistical analysis via computer software
programmes.
But in the earlier, exploratory phase of this research, it will be
necessary to ask married couples some searching questions
about how many children they plan to have. This is a notorious-
ly difficult area to research, but it is safe to say that, in order to
get some understanding of the factors which.make people
decide such personal matters, some lengthy, relaxed, open-
ended conversations between sensitive researchers of the same
gender as the respondents will be required. This means infor-
mal, qualitative research. Only when issues have been explored
at leisure, in depth, and in an informal, conversational setting,

45
will the researchers be able to formulate precise questions for
the essential quantitative phase of the work.
Note that in this case, informal and qualitative work is the
essential pre-cursor for the formal, quantified phase. Each
depends upon the other for over-all predictive accuracy. The
methods are not in competition they are appropriate for dif-
ferent phases of the total research exercise. How does this relate
to development project research? Scaling down from country to
province or district, the satisfactory expenditure of development
funds (even more scarce than tax revenue because raised by
voluntary contributions from people who have already paid
taxes), requires equal sensitivity and rigour. Development pro-
grammes are guided by the notion of equity, or fairness.
Programmes are aimed at the most vulnerable and needy social
groups, and can only reach those targets if such groups are cor-
rectly identified. Development funds which miss the poorest and
help make the rich richer are seriously wasted. They are not
only failing to do what they were meant to do, but they may
also be strengthening local power structures. Therefore, the
accuracy of targeting involving the correct identification of
the categories of people in need is of essential importance.
As in the previous example, research on development issues
has both quantitative and qualitative aspects. The same prob-
lems face NGOs as face national planners, but on a small can-
vas, and with the difference that NGO development researchers,
who give priority to close consultation with and participation of
local people, stand a better chance of designing projects which
will work well and lead to local empowerment.
Development agencies, and large projects supported by these
agencies, are increasingly employing programme staff from the
countries in which they are operating. Involving these staff in
research has two useful spin-offs. First, the local staff become
better informed about the great variations in social and econom-
ic conditions, and the survival strategies of the poor, in their
own countries. Secondly, the research experience gives them
specific roles in contributing to the setting of objectives for the
agency's annual workplan. Because they obtain direct experi-
ence of.collecting and analysing the data and feeding research
into the plan, they develop enhanced commitment to the
achievement of programme objectives.

46
BIBLIOGRAPHY
Beaudoux E. et al (1990), Supporting Development Action at
Community Level: a methodological guide, Belgium: COTA.
Casley, D.J. and Lury, D.A. (1981), Data Collection in Developing
Countries, Oxford: Clarendon Press
Cole, R. (ed.) (1989), Measuring Drought and Drought Impacts
in Red Sea Province, Oxford: Oxfam.
(1988) 'Participation in Evaluation', Community Development
Journal, Oxford: Oxford University Press.
Epstein, T.S. (1988), A Manual for Culturally Adapted Market
Research in the Development Process, Sussex: RWAL Publications.
Fernandes, W. and Tandon, R. (1981), Participatory Research
and Evaluation: Experiments in research as a process of libera-
tion, New Delhi: ISI.
Feurstein, M.T., (1986), Partners in Evaluation, London:
Macmillan.
Gardner, Katy (199D Songs at the River's Edge: stories from a
Bangladeshi Village. (Sensitive evocation by a young European
woman of living in a Muslim community as a researcher.)
Virago.
International Institute for Environment and Development [IIED]
Newsletter on Rapid Rural Appraisal.
Marsden D and Oakley P. (1990), Evaluating Social Development
Projects, Development Guidelines 5, Oxford: Oxfam.
Nichols P. (199D, Social Survey Methods: A fieldguide for devel-
opment workers, Development Guidelines 6, Oxford: Oxfam.
Oakley, P. (199D, Projects with People, Geneva: ILO.
Robertson, A.F. (1978), Community of Strangers (An informal
description of anthropological research in Uganda.), London:
Scolar Press
Trivedy, R. (1990), Action Research in Southern Malawi, Oxford:
Oxfam.

85
Tina Wallace with Candida March (eds) (1991) Changing
Perceptions: writings on Gender and Development. Oxfam.
UNICEF (1988), Methodological Guide on Situation Analysis of
Children in Especially Difficult Circumstances, UNICEF: Bogota.

86
APPENDIX 1 CHECK-LIST OF QUESTIONS

Do you really need to do some research?

Who needs to know what, and why? For policy formation? To
assist a decision on a grant application? To monitor or evaluate a
project? To assess needs in an emergency?
Could it be that someone has already done the research you
think you need?
How much time have you got? As long as you like? A year?
Three months? A week? A day?
How much will it cost to find out? Will a specific level of accu-
rate research justify a relatively large expenditure? If you are
being asked to spend 5,000, is it worth 250 of research to
make a reasoned decision?
Have you the staff you need? Do you need expertise? Can local
people undertake some or all of the research? Can you train up
some helpers in the time available?
Have you consulted local people? How far does the proposed
research take account of local perceptions of real-world prob-
lems? Is this the research the local community thinks it needs?
Will you check with them at each main phase?
Will your research arouse local expectations? Is it clear to the
community why this research is being done? Are people expect-
ing miracles, or major commitments, just because some ques-
tions are being asked? Will their expectations skew the findings?
Have you thought about possible bias? Gender bias? Class bias?
Locational bias? Ethnicity? Religion?
What is the research unit? A region? A village? A cluster of settle-
ments? A category of people, such as landless labourers?
Individuals? Households? How is the unit defined and bounded?
Are there any 'missing persons' (labour migrants, herders)?
What about time? Will you take a snapshot, or must you follow a
process over a period of months? Will the research be affected
by the data having to be gathered over a period of several

87
months, or in a particular season?
Will you need to sample a large population, or is it practical to
deal with everyone? What kind of sample makes most sense?
One method or several? Is there an ideal research method to
meet your needs, or do you need to combine several? What mat-
ters most - predictive accuracy, because it will affect expendi-
ture, or the quality of the insights? The need to maintain good
rapport with a vulnerable group?
Can existing sources meet your needs? Is there a national census
with data on your region? Has a department of local government
tried to answer these particular questions?
Is your research vulnerable to informant bias? Have you become
the intellectual prisoner of any particular section of local opin-
ion? Have you any cross-checking procedures to help in valida-
tion?
Can you, should you, use a sample survey or questionnaire?
What about participant observation? Have you the time to do
this? Has someone already done it for you? Is there an experi-
enced anthropologist or rural sociologist who knows this region,
speaks the language, and can get your questions answered in a
few weeks' intensive work?
Rapid Rural Appraisal? Should you field a team of specialists,
who will consult relevant literature, then work flexibly with local
people and present their provisional findings to them, and to
you, in a series of workshops?
Have you really taken advantage of local knowledge, local
memories, local skills? They have had a lot of experience in sur-
viving and adapting - is your research giving that experience full
weight?
Has the research got back to the people it was meant to help?

88
APPENDIX 2 QUESTIONNAIRE DESIGN

We assume that the purpose is to assess community interest in a

local person running a small mini-bus service to surrounding
market towns. If community interest is strong enough, there
might be a case for making a loan to the future operator. The
two questionnaires are obviously not complete - merely initial
sections tackling obvious issues.

A poorly designed questionnaire

Ql Do you travel a lot outside the village?
(The interviewer has jumped straight into the task without a
word of explanation or introduction. This is bound to seem
abrupt, if not rude. The question itself is a very vague ques
tion. What is 'a lot1, and what will be learned from the
answer, whether it is 'yes' or 'no'?)
Q2 Would you travel more if there were cheap transport?
(Another vague question, and it is hypothetical, as well. It
invites the answer 'yes' and little more.)
Q3 How do you travel?
(This question will encourage a very general answer, and
possibly some confusion. If the listener is thinking about the
next village, one kind of answer will emerge; but if they are
thinking of a market town 50 kilometres away, then another
will apply.)

Q4 When do you travel?

(Although this question looks specific, it can elicit all kinds
of answers: 'When I have to.' 'When someone offers me a
lift.' 'When I have something to sell.' 'When I must buy
something.' 'If a child is sick.' The question merges times,
and motives in the word 'when' and gives no clue as to
which kind of answer is intended.)

Q5 Who else in your household also travels outside the village?

89
 (As we suggested in the main text, it is much better to ask
each member of the household about their actual behaviour,
separately. The only excuse for a question in this form
would be a drastic shortage of time and money, and it is
possible that a male household head would give answers
about his wife or children which carried implications about
his status and authority, and could not necessarily be a reli
able guide to what really happens, or what other family
members would like.)

A more thoughtfully designed questionnaire

Introduction: Hello. My name is X and I have been invit-
ed to your community by the mayor. I am trying to under-
stand about how much the people here travel outside the vil-
lage, and some related things. Could you please spare me a
little time - about 15 minutes? I am asking people from
every house in the village. There are no right or wrong
answers to the questions, and you should not find it at all
difficult to reply because they are all things you know about
from your daily life.
(These introductory remarks inform the interviewee exactly
what is going on and should make them feel relaxed and
unthreatened.)
Ql In the last 12 months, you may have travelled outside the vil
lage. I want to write down all the times you did this, and for
what purposes, starting with the most recent trip.
A 1 the place the reasons
A 2 the place the reasons
A3 the place the reasons
A 4 the place the reasons
A 5 the place the reasons
(continue on separate sheet if necessary)
(By choosing actual, recent behaviour, and by eliciting an
open-ended number of examples, this question and the fol-
lowing questions, avoid the vague generalities which might
have resulted from the first questionnaire.)

90
Q2 When you travelled to (place, from Al) how did you travel?
Tick one of the following:
on foot Q on an animal
by bicycle by motorbike/moped Q
in a lorry in a mini-bus
in a large bus in a car
other [specify]
(Repeat with all the answers from Q 1.)
Q3 How much did it cost to travel to (place from Al)
(Repeat for all answers to Ql.)
Q4 How much is it worth to you to be able to go once a week
to the town of XYZ, and come back the same day?
Al not more than 50 pesos
A2 from 55-75 pesos?
A3 from 80-100 pesos?
A4 from 105-120 pesos?
A5 from 125-150 pesos?
(This question is hypothetical, and so suspect. It would need
to be checked against actual prices being paid, either in this
community or in a similar one.)

91
APPENDIX 3 SOME HIGH-TECH, HIGH-
COST RESEARCH METHODS
You are unlikely to employ these methods directly in a 'hands-
on' sense in project research because they require special skills,
or costly equipment. But it is useful to know they exist, and
how their results might be exploited by you if you can buy them
in 'off the shelf.
Aerial/satellite photographs: the use of aerial photos and
satellite photos and images can help give a view on a large
scale, not only for the more obvious physical features of an area,
but can also allow estimates to be made of population and live-
stock densities. They can also be used for city mapping; this
may be of particular importance given the delay in incorporating
new slum/squatter settlements in many city maps. Of course,
such large-scale pictures will not help you to piece together
social relationships but it will give you access to crude numbers
which can then be supplemented by on the ground validation.
Off-the-shelf data: A great deal of material is often available
off the shelf from large agencies and can be obtained free of
charge or at low cost. Although one needs to be aware that
sometimes there may be security sensitivities about access to
material (and for this reason many countries locate maps and
aerial photos in the military institutes rather than civilian).
Time comparisons: Geographic information may enable com-
parisons across time as even in countries with a weak adminis-
trative infrastructure on the ground it may be possible to find
photographic and other information. For example, material was
found from the 1960s which allowed comparisons of wadis in
Sudan with the present day. Other land use patterns can be
established for which oral traditions may be inaccurate and
administrative records unavailable, especially in 'marginal areas'
such as arid zones or rain forests.
Mapping changes: mapping can also be useful in tracing the
development of cities especially where the administrative
boundaries failed to recognise or keep abreast with growth, a
problem commonly encountered in fast-growing Third World

92
cities, whose administrative capabilities are over-stretched.
You can use this material for historical comparison of settle-
ments, land use, and land use changes, true rates of desertifica-
tion, etc., and to resolve issues otherwise difficult to verify.
Around Picchis Palcazu in the central Peruvian jungle, the gov-
ernment claimed there were no settlements but overflying con-
firmed a considerable number (3,500 colonist and 3,500 tribal
people) of people settled in the valley and involved in agricul-
ture; in Sudan, maps were used to establish whether certain
areas were suffering drought or not, and whether this was
exceptional or normal to the area.
There is a danger with all technology that it can mystify rather
than illuminate and for this reason it is not used even when it
could supply crucial information. However, we would again
stress the need for validation on the ground to cross-check the
material gathered at a macro level. For example, a settlement
may be identified through aerial photos, but on a follow-up visit
we may find that many of the dwellings have been abandoned.
Geographic systems are not a substitute for social research.
They may identify how many animals are grazing in an area, but
will not show who owns them, how they are used, or where
they are sold. You need also to be aware of seasonal differ-
ences, especially as some techniques may only be utilised at cer-
tain times of the year (aerial photos are taken when flying con-
ditions are good and visibility clear).
Technology is a good servant but a poor master. Its sales staff
and enthusiasts will hard-sell it, because they stand to profit
from it, or are caught up in its exciting capabilities. You should
not be seduced into thinking hi-tech systems can resolve all your
problems, but conversely there is little gain in making life diffi-
cult by refusing to use computers, aerial photography or any
other technical process if they can help you do your work bet-
ter, or with less effort for the same quality of results.
Geographic information systems: This is the generic name
for satellite photography used in conjunction with computers
and their graphic display systems in the understanding of phe-
nomena distributed on the earth's surface i.e. physical rather
than social characteristics. The GIS is a computerised system for
the specification, acquisition, storage, retrieval and manipulation
of data in a defined physical area, for example a county or dis-

93
trict. The principle task of a GIS is to make many and diverse
forms of data comprehensible and to elicit patterns and relation-
ships between variables where they exist. Typical data sources
that are used are soil surveys, air photos, landsat images, topo-
graphic and thematic maps, tabular data, and almost any form of
survey of other features in the area. A vast range of material
could be incorporated such as land use, land cover analysis,
farm size, field size, livestock population, agricultural produc-
tion, vegetation type and density, farm buildings, grain stores,
human population, malnutrition rates, and many others. There is
no limit in theory to the number of variables that may be used
as long as they relate to the defined target area.
The advantage of the GIS over manual records is that all of the
above sources of data and different variables may be combined
to produce new classifications of the space or area and can be
examined in relationship to other variables. The results can be
expressed in a report, graphical, cartographical or statistical
form. The map is the most easily understood and common prod-
uct of a GIS. The GIS is composed of both hardware and soft-
ware. The hardware is usually a minicomputer, a digitising table,
a pen plotter, a dot matrix printer and a high resolution monitor
or screen. Sometimes a tape drive is included for reading tapes
of landsat satellite images for use as a base map and data source
for the GIS. A digitising table is used to do two things: enter the
area of the earth to be studied and identify and assign values to
discrete entities. The digitising table attached to the computer is
really an electronic table of defined space composed of thou-
sands of tiny sensors. Typically the base map is taped to the top
of the table and identified as occupying such and such a space.
A small digitising puck is used to locate points and areas on the
map on the digitising table and enter them onto the computer.
The whole thing is like an electronic tracing paper. There are
many software packages available for the analysis of spatial data.
A fully equipped system is likely to cost in the region of more
than 20.000 sterling, and will require skilled staff to use it.

94
APPENDIX 4 LOGICAL FRAMEWORK
ANALYSIS
Logical framework analysis (LFA) is also known as Project
Framework Analysis. LFA was first developed in the US but has
been adopted and modified by a number of governmental agen-
cies, and increasingly by NGOs. Many governmental co-funding
donor agencies now encourage NGOs to use LFA in preparing
their project proposals and reports. As noted in the Introduction
to this book, LFA is not really a research method but a way of
ordering information. The names may be different but the basic
approaches are similar in that they clearly relate to 'projects' and
all assume that these are designed to effect a positive change in
the lives of people. LFA is based on a relatively simple matrix
whereby objectives are matched to project activities and
assessed in terms of inputs (resources utilised) and outputs
(benefits). These are then measured by previously agreed indica-
tors in the light of the risks and assumptions made by the pro-
ject planners. An illustrative sketch of a LFA matrix is given
below.
The proponents of LFA argue that it brings together key state-
ments and information in one document and ensures that pro-
ject objectives are clearly spelt out. It enables people to relate
project actions to project purposes and goals, as well as the
resources utilised and outputs attained. It provides a basis for
further monitoring and evaluation and is a flexible system,
which can be adapted to suit various purposes.
However, critics are unhappy about the dominance of LFA in
many agencies because they feel that the system can be very
inflexible in that it can force people to think in terms of easily
quantifiable and rigidly defined categories. They argue that
development workers find themselves being obliged to sum-
marise complex ideas and social relationships into meaningless
phrases. In addition, they are expected to quantify the unquan-
tifiable, or measure major social changes and movements.
A further criticism is that LFA is too project-centred, and forces
research, reporting, and evaluations unnecessarily into the
framework of projects. Many people argue that projects in them-
selves are merely short, artificially created episodes imposed on

95
the social processes through which people are living.
ZOPP: ZOPP (Target Oriented Project Planning) was developed
in Germany in the early 1980s. It is based on LFA, with certain
additional elements. These are, an emphasis on the team
approach to project planning, and a recognition of the impor-
tance of the pre-planning research and survey work required to
ensure that problems re properly identified and that their causes
and effects are related to the real experiences and needs of indi-
viduals and groups of people. These problems, it is said, can
then become the basis for agreeing project objectives.
The advocates of ZOPP have also relied on the use of visual
aids such as wall charts for group meetings involved in activities
of planning or monitoring activities. People are encouraged to
contribute by writing their ideas on cards which are added to
the wall-chart, which may be based on a LFA matrix.

Typical UFA Matrix (based on one used by the British ODA)

Indicators Measurement Assumptions and
(Means) Risks
Main
Objectives
Short-term
Objectives
Outputs
Inputs

96
APPENDIX 5 WOMEN AND
EVALUATION
(This paper, by Frances Rubin, was originally published in
Changing Perceptions: Writings on gender and Development,
published by Oxfam in 1991. It gives a first-hand account of
some of the problems that may be encountered in attemping to
involve women in research projects.)

It is now nearly two years since I first coordinated an Oxfam

Country Programme Review. The exercise gave plenty of food
for thought about the difficulties that women can face in devel-
opment work, and particularly in the process of evaluation. This
note is anecdotal, but may illustrate some of the problems and
give pointers to solutions, or at least steps that need to be taken
to address the danger of excluding women from important
moments of decision making.

Terms of reference
One of the first tasks to be completed was the drafting of the
terms of reference (TOR) for the review in collaboration with the
Country Representative. The programme staff had voiced con-
cern about how to tackle gender questions within their work,
and were trying to develop a country strategy where 'gender'
was seen as an important issue. Following consultation with a
number of colleagues, 'gender questions' were written into many
aspects of the TOR. There was quite a strong reaction to this.
Many people felt that such explicit mention of gender issues
suggested that nobody, including consultants who might be con-
tracted, were aware of gender issues. Because of these strong
feeling the TOR were modified but remainedexplicit in terms of
briefing the evaluation team members that gender issues were to
be treated in an integral way throughout the evaluation. Above
all, gender was not to be treated as a discrete issue in a separate
chapter.

97
Contracting of consultants
At one time in the course of looking for members of the evalua-
tion team it seemed possible that they might all be women. This
raised reactions of alarm among some who suggested that an all
women team might not be balanced. For how many years has
Oxfam used all male teams without feeling concerned? At the
end of the day the evaluation team was composed of two men
and three women. At times they split up into different groups
but these were always mixed.

Field visits
Even though gender was recognised as a central issue and there
were always women in the visiting teams, we still encountered
considerable difficulties in actually meeting women and hearing
their views.
In a number of situations the male members of the team
explained with confidence what women thought on a number of
issues. To give one example, they explained that women could
not participate in training courses because they needed to stay at
home to look after the domestic affairs of the household.
However, when we spoke directly with women, the women out-
lined various strategies that would enable them to leave their
families and community, and attend meetings they saw as rele-
vant.
On one particular day we were carrying out a programme of
visits that had been organised by the village leaders. We asked if
we (the visiting Oxfam team) could split up into two groups, so
that the women on the team could talk separately to the women
in the village. Since this was not on the agenda it took quite an
effort to organise, and we were continually interrupted by peo-
ple who wanted us to hurry so that we could rejoin the'main'
delegation.
Another time, I was reading the notes in the file, as we were
on our way to meet a women's group. In one of the tour reports
it was clearly stated on which days it was inappropriate to visit if

98
you wanted to talk to the women because the women were
preparing for, and selling in, the market. Yet we found ourselves
en route to this town precisely on market day! This happened
again in another small community, where there was a very low
turn out of women because they were all busy preparing man-
ioc loaves for the market on the following day.
On our last project visit, committee members of a health post
began to assemble for the customary discussion. One by one the
com-mittee members were introduced. There was one women
member. Her name was read out 'Citoyenne (treasurer) pre-
sefit.' 'Present?' I asked, 'but where?' The treasurer then popped
her head into the inner circle of the crowd, bowed and disap-
peared.
After the round of introductions was complete I asked why
the citoyenne who was present in the village was not participat-
ing in the meeting? The reason for her absence was that she was
busy preparing lunch for the delegation. We politely suggested
that she should be present at the discussions. She appeared, and
in the course of the debates raised a number of issues of impor-
tance: issues related to the management and safekeeping of the
finances that had been entrusted to her, as well as health issues
that were of specific importance to the female members of the
community, and which had not been raised by the other mem-
bers present.
A plate of cold peanuts was passed round instead of a lovely
meal: we had passed up on some culinary delights but we had
enabled the treasurer to raise some important questions for the
community; had she not been in the meeting these issues would
not have been brought to Oxfam's notice.

Conclusion
It must be a rule that the TOR of all evaluations include gender
as an integral part; that, except in very specific circumstances,
there should be mixed evaluation teams and that, however diffi-
cult it continues to be, women evaluators must ensure that they
get the opportunity to talk to women on their own.

99
APPENDIX 6 ACTION RESEARCH
PROGRAMME THE METHODS AND
ARPROACH
(This is an extract from a paper by Roy Trivedy, Action Research
in Southern Malawi, published by Oxfam in 1990. It describes
the methods employed in research undertaken for Oxfam in
Mulanje District, during 1987 and 1988. The aims of the
research were threefold. Firstly, to identify and explore in detail
the development problems and priorities of the rural communi-
ties in Mulanje District. In particular, the research sought to dis-
cover the problems and priorities of some of the most vulnera-
ble groups of people. Secondly, to identify and explore possible
solutions to these problems and priorities, and thirdly, to make
recommendations to Oxfam on its future role in Malawi.)

The approach of the ARP can be broadly divided into three main
stages. The preliminary stage, lasting from mid-August until the
end of October, included the formulation of the basic approach,
selection of villages, and selection and initial training of the
research assistants. The second stage, 'fieldwork', lasted for four
months from the beginning of November 1987 until the end of
February 1988. Finally, the third stage 'post fieldwork', has
included circulating a summary of our findings as well as con-
sulting with various groups of people about the validity of our
findings and the question of how Oxfam could follow up and
build on the work of the ARP. In this section we will provide a
more detailed account of each of these stages.

The preliminary stage

This began with the Development Adviser conducting a litera-
ture review tofindout:
what had been written about rural poverty in Malawi;
what methods had been used in conducting research on simi-
lar themes; and

100
 specific details about Mulanje District which could be useful
for the ARP.

The literature review revealed a number of interesting things.

Firstly, we found that there was very little written specifically on
rural poverty in Malawi. Although a number of writers had dealt
with many closely related subjects, including malnutrition and
food availability, the effectiveness of agricultural extension,
women's and smallholders' participation in agricultural develop-
ment, and the role of income-generating activities in rural areas,
we found very little literature which had attempted to deal with
the issue of rural poverty in any comprehensive manner.
Secondly, we found that, with few exceptions, almost all of the
research which had been carried out in Malawi on the above
mentioned themes had utilised formal, structured questionnaires
as an essential part of their methods. Whilst formal structured
questionnaires do have certain advantages over more
'open-ended' research methods, it was felt that if we were gen-
uinely to try to find out from the rural poor themselves how they
define their needs and priorities, it would be more appropriate to
use other research methods. Apart from the literature survey,
another factor which assisted us in formulating the research
methods was our initial discussions with several key people in
the area including staff at the District Commissioner's office and
at the Rural Development Projects at Mulanje and Phalombe.
These discussions confirmed our view that if structured question-
naires were used as part of the research methods then it would
prove extremely difficult for us to gain any further insights
beyond what was already known on the problems and priorities
of the poor, let alone to increase people's awareness ofthese.
By mid-September we had begun to formulate an outline of
the research methods which we felt would be appropriate. We
proposed to have six Malawian research assistants staying in six
villages for a period of approximately four months from
November to February 1988. Although it would have been inter-
esting to employ more research assistants and thereby work in
more than six villages in the District, we felt that this would

101
have been detrimental in terms of the quality of support and
supervision that the Development Adviser would be able to pro-
vide. We hoped to employ an equal number of female and male
researchers and wanted to try to ensure that the researchers
were of different ages. The six villages which would be chosen
were to be well spread geographically so that the findings
would provide a fair indication of the problems and priorities of
people in the District as a whole (see below on village selec-
tion). A variety of factors influenced the decision to do the field-
work for a period of four months. Firstly, if we were to be able
to make recommendations to Oxfam by July/August of 1988, we
felt it would be problematic to extend the field work much
beyond March. Although the initial plans envisaged a
four-month period of fieldwork, the decision about whether it
might be appropriate to extend the fieldwork for a further
month, until the end of March, was left open. Secondly, it was
felt that the fourmonth period November to February would
roughly coincide with the peak agricultural period for farmers
and also the period of 'annual maximum stress' for people, in
terms of food availability and health-related problems. Much of
the research in Malawi, as elsewhere, has been done during the
dry season, and by doing the fieldwork during the rainy season,
we were aware of this being a 'seasonal reversal'. Although we
were aware that there was a danger that many farmers would be
'too busy' in their fields to talk with the researchers, we felt that
the advantages of doing the research during this period out-
weighed this. We felt that the main advantage was that seasonal
factors would enable the researchers to identify the most vulner-
able groups of people much more easily. In addition the
researchers themselves would learn a lot about rural life just as a
result of having experienced life in the villages during the rainy
season, when many of the problems faced by rural people are
most acute.
During their stay in the villages the researchers were to con-
duct a 'listening survey'. The idea was that the researchers
would begin by establishing trust and equality with villagers.
Thereafter they would be engaged in listening to people in dif-

102
ferent situations, at work, at the market, during leisure activities,
in their homes, etc. At the same time as listening to people, the
researchers were to make observations about village life and
also informally talk with people. By using these methods it was
felt that we were more likely to be able to find out from people
themselves about their problems and priorities. A crucial ele-
ment of the exercise was that the researchers were to make it
clear from the beginning that they were there to learn from the
villagers about rural life, what problems confront people, how
people cope with these and what people's own priorities are.
One issue which we were very aware of, concerned the pos-
sibility that our work might raise people's expectations. Having
explained their problems/priorities to the researchers, it was felt
that people might then expect the researchers or the
Government or perhaps even Oxfam, to provide 'solutions'. We
discussed this issue at length in our initial training session and
also in many of the ongoing training sessions. We felt that it was
crucial to explain the purpose of the research to villagers clearly
from the beginning so that no false expectations were created.
This was felt to be of such importance that part of the initial
training session specifically involved each researcher practising
explaining the purpose of the research in Chichewa in front of
other people, so that any ideas which could raise expectations
would be omitted. It was also felt that we should explain to
people that ultimately the research aimed to assist the
Government of Malawi and other agencies to improve 'develop-
ment planning' in the country. Hence, although the villagers
themselves would not immediately gain anything as a result of
the research, it was hoped that in the longer term the District as
a whole would benefit through better development planning.
(See also fieldwork part of this Section.)
To assist the researchers a set of broad 'guidelines' were pre-
pared. These were in the form of a set of questions covering a
number of specific areas of interest for the study, including
social organisation, land constraints, agricultural
extension/advice and training, credit/loans, labour supply, irriga-
tion, soil conservation, livestock, forestry, off-farm income and

103
ideas for income generation, health/illness, water and sanitation,
nutrition, patterns of work, exchanges and transfers, and
income/expenditure. It was pointed out to the researchers that
the guidelines were not exhaustive but merely sought to outline
some of the types of information which it would be useful to
find out, particularly when villagers themselves raised specific
subjects as priority/problem areas. Moreover, it was made clear
that the guidelines were not to preclude the researchers from
collecting information on other areas which villagers themselves
prioritised. The researchers were informed that the guidelines
were not to be interpreted as a rigid list of information which
had to be collected. The purpose and aims of the guidelines
were discussed extensively during the initial training session and
the guidelines were revised and reformulated in subsequent
meetings of the team.
A final point concerning the research methods was that the
field work was divided into two parts. In the first part, lasting
approximately six to eight weeks, the researchers were to
attempt to build up a general picture of the problems and priori-
ties of the village as a whole and the extent to which existing
service provisions met or failed to meet these needs. In addition,
during this period, the researchers were to identify some of the
most vulnerable groups of people in their village, and this meant
that they had to ensure that they met every person in their vil-
lage. The main emphasis of the first part of the research was on
the researchers establishing a relationship of equality and trust
with villagers; without this we felt that the second stage of the
research would be extremely difficult and probably unsuccess-
ful. In the second part of the fieldwork, we focused much more
closely on the more disadvantaged members of the communities
in order to try to identify their problems and priorities. By using
this method we felt that we were more likely to be able to find
out the priorities of the poorest and most vulnerable groups
without stigmatising these groups. [Note: We are indebted to
PHAM (Private Hospital Association of Malawi) Primary Health
Care experiences for making us aware of this factor. We also felt

104
that it would have been impossible for us to gain any significant
insights into the issues of intra-household and gender inequali-
ties without the researchers first building up a sufficient level of
trust with each of the communities as a whole. The other advan-
tage of this approach concerns the question of the unit of analy-
sis in this type of research. By focusing initially on wider social
relations and not only on 'the household' we felt that we were
more likely to be able to distinguish between the 'idealised fami-
ly forms and the reality of family structures'.]
Having decided on the methods to be used, the next stage
involved selecting the six villages. We used three main criteria
for the selection of the villages. Firstly, we felt that since only
one researcher would be based in each village the size of the
village, particularly in terms of population, had to be one deter-
mining factor in our choice. In this respect we felt that it would
be unrealistic to expect the researchers to do the work effective-
ly if the population of the village exceeded 250-300 people.
Unfortunately, the only data available at the time on the popula-
tion of villages in the District was from the Census of 1977. Since
this was out of date, we estimated that the population of each
village would probably have grown by approximately 3 per cent
each year since 1977. On the basis of these very rough figures
we shortlisted 28 possible villages for the research. A second cri-
terion used was that the villages should be geographically well
spread. In this respect the literature survey and our discussions
with people in the District proved useful. Since we were already
aware of some locally specific problems in the District, we felt
that by having a good geographical spread of villages we were
more likely to be able to build up a picture of the problems and
priorities of people in Mulanje as a whole. A third set of criteria
which was used was what may be described as essential practi-
calities. These included:
that the selected villages had to be accessible with a
four-wheel-drive vehicle throughout the rainy season
(although we were also conscious of trying to avoid villages
which were unrepresentatively near to main roads or other

105
 main service provisions); and
that some accommodation was available which could be rent-
ed for the four-month period in the selected villages. In this
respect we were conscious that the hired accommodation
should be representative of housing conditions in the village
generally, and that no special favours were made available
which would make the researchers' experiences unrepresenta-
tive of life in the village.

During late September and early October, 18 short-listedvil-

lages were visited in order to try to assess which villages fitted
the criteria and would be suitable for the work. The visits were
made with accompanying staff from either the Rural
Development Projects or Community Development or other ser-
vice providers. The visits enabled the Development Adviser to
learn more, through the discussions with accompanying staff, as
well as people in each of the villages especially the 'Village
Headmen'. The final selection of the six villages was done after
consultation with the DCs office in Mulanje, Blantyre ADD offi-
cials, the RDP staff and Louis Msukwa. The DCs office agreed to
notify the Village Headmen in each of the six villages, as well as
local party officials, so that we could start the fieldwork at the
beginning of November.
The next stage of the work involved the selection and training
of the research assistants. In mid-October, Louis Msukwa and
the Development Adviser interviewed seven candidates for the
research assistant posts. The candidates, three women and four
men, had all been educated to degree or diploma level. Five of
them had previous research experience, although this had main-
ly involved questionnaire-based research. Two of the candidates
were recent graduates. Of the seven, four (the three women and
one man) had agriculturerelated qualifications, the remaining
three had social science backgrounds.
In terms of their ages we were not wholly successful in being
able to attract people in different age ranges. Six of the candi-
dates were in their mid-twenties, the seventh was in his early
forties. We invited all seven candidates to attend the initial train-

106
ing sessions during which we hoped to assess the suitability of
each candidate and choose six people to do the village-based
work. It was hoped that the remaining candidate would perhaps
be able to assist the Development Adviser either on a temporary
or part-time basis.
The initial training session prior to the fieldwork lasted just
over a week and was led by Louis Msukwa, Percy Kantunda of
PHAM (one of Oxfam's partners in Malawi), and myself. Several
factors were instrumental in determining the length of the initial
training course. It was recognised that the validity of our find-
ings ultimately depended on the level of awareness and sensitiv-
ity of the researchers, and that they had little experience of
using the research methods outlined above. At the same time,
however, we were keen to try to ensure that the researchers
were settled in the villages prior to the start of the rains. After
much thought we decided that having just over a week of initial
training, followed by regular ongoing training/discussion meet-
ings every two to three weeks during the fieldwork, would pos-
sibly be the best solution. In addition to our regular two- to
three-weekly meetings, a two-day refresher training course was
arranged mid-way through the fieldwork period in January.
As part of the background to the initial training session some
reading material was distributed which the researchers were
expected to read and which the group as a whole later dis-
cussed. The reading material included the Oxfam Country
Review as well as material focusing on rural poverty and the
problems of investigating this. The aim of distributing this mater-
ial and discussing it was both to enable the Development
Adviser to assess the researchers' awareness of rural poverty and
also to increase the group's awareness of some of the problems
we were likely to face during the fieldwork.
Although there were a number of themes which we had to
cover in the initial training session, we did not attempt to follow
a rigid pre-determined agenda. The session began with a very
open discussion about development policy in Malawi in the
post-independence period. This discussion was led by Percy
Kantunda and the Development Adviser. By using some of the

107
techniques developed by PHAM, namely by raising a series of
questions and issues for discussion, we were able to begin
thinking about how certain groups of people had been
by-passed in the Country's development efforts. The discussion
then moved on to the issue of which groups could be consid-
ered to be 'at risk' and how these could be identified. Part of
this discussion centred on the crucial issues of the accessibility
of services for the most vulnerable groups and the relevance of
these services to the needs of these groups. At this stage the
group discussed how services could be made more accessible
and relevant to the needs of the most vulnerable groups. The
group then began to realise that a prerequisite for this to take
place was that service providers may often need to change and
be willing to listen and learn from the most vulnerable groups.
For this to become a reality, the group concluded that a relation-
ship of trust and equality had to be established between 'service
providers' and those for whom the services were meant. We
then concretised these discussions by brain-storming on some
techniques and methods which could be used to build up trust
and equality. For this part of the course, we were able to use
some of the material from the Training for Transformation man-
uals, as well as adding our own life experiences. This part of the
training took up almost the whole of the first two days.
On the third day we discussed the aims of the research and
the methods (including the guidelines) that we would be using
in the context of what we had learnt earlier. This was followed
by Louis Msukwa leading a discussion on development planning
and policy in Malawi which tied together many of the themes
which we had discussed. This discussion drew together the links
between different sectors of the economy such as health, educa-
tion and agriculture. Louis Msukwa also provided us with a lot
of background information about the District, as well as drawing
our attention to some of the problems we were likely to face in
the fieldwork. [Note: This included discussion about village
power structures, the role of local party officials, Village
Headmen, traditional healers, etc.] We discussed at some length
how we should introduce ourselves to villagers and how to

108
explain the purpose and aims of the research without raising
people's expectations.
The final one-and-a-half days were spent discussing ways in
which our 'listening skills' could be improved, as well as talking
about some of the practical arrangements, including support and
supervision arrangements and recording procedures.
At the end of the training week a decision had to be made
about which of the seven candidates would be left out of the
work in the villages. This decision was made easier by the fact
that one of the women candidates informed us that there was a
strong possibility that she would not be able to work for the full
four-month period. As a result we were left with four men
(Mike Chibwana, Jackson Chimowa, William Chinkhadze and
Henry Mchenga) and two women (Grace Kamwendo and
Chimango Mkandawire). Each researcher was assigned to one of
the selected villages. By the end of October we were ready to
begin the field work.

Fieldwork
Despite all the preparations, our entry to some of the villages
was not without problems. In the Phalombe part of the District,
in two of the three selected villages (Nantapo and Ligola) there
were no problems. The Village Headman and party officials had
received notification of our arrival from the DCs office and were
awaiting us. In the third village, where Mike Chibwana was to
stay, we encountered problems in that the village leaders were
unable to find suitable accommodation. We were eventually able
to place Mike at Mmora village, a few miles away from the vil-
lage which had originally been selected. In the Mulanje RDP
area the situation was even more complicated. We found that
only in one of the three villages (Misomali) were local officials
expecting us. In the other two villages the authorities had not
received notification from the DCs office and so we decided
that it would be advisable to go back to the DCs office, ensure
that local officials were notified, and return the following week.
Again, lack of accommodation proved to be a problem in one of
the villages. Fortunately, there was another suitable village

109
which had originally been visited during the village selection.
Hence, after making appropriate arrangements, it was agreed
that we should work in that village. By the second week in
November we had begun the research in the following villages;
Ligola (Grace Kamwendo), Nantapo (Jackson Chimowa), Mmora
(Mike Chibwana), Murotho (Henry Mchenga), Chilembwe
(Chimango Mkandawire) and Misomali (William Chinkhadze).
During the first month of fieldwork, the researchers were
engaged in explaining the purpose and aims of the research and
introducing themselves to all the members of their village. This
could perhaps have been made easier if the researchers had
been able to talk to the village members in a general meeting
called by the Village Headman. However, during the planting
season public meetings in the villages are not encouraged by
party officials because it is felt that this draws people away from
their fields. In any case, in our discussions during the training
we had agreed that the researchers should not ask for any spe-
cial meetings to be called on their behalf. This meant that in
order to meet everybody in their village the researchers had to
frequent the places where groups of villagers often met. This
included meetings at the water-collection points, at the market,
in tea rooms and occasionally bars, accompanying people to
their fields and on their way to collect firewood. At the same
time the researchers visited each household in their village to
ensure that no-one was left out and that every person in the vil-
lage knew who the researcher was and his/her purpose in the
village. We were conscious that all of this should be done with
as little disruption of 'normal village life' as possible and that the
researchers should not draw unnecessary attention to them-
selves: At this stage the researchers were mainly engaged in
observing village life and listening to people.
Throughout the fieldwork period one thing which we tried to
do was to cross-check a lot of the information that was collect-
ed. Whenever possible the researchers were asked to
cross-check information in three different ways, a process
known as 'triangulation'. For example, we were often told by
individuals about the size of their landholdings. Whenever possi-

110
ble the researchers themselves would visit the garden and try to
estimate the size themselves (the researchers with an agricultural
background were generally better at this); in addition, compar-
isons were made with estimates of the landholding sizes of other
villagers, and in some cases where Farm Assistants visited the
village, the researcher would ask their opinion of the size of a
particular field. By using the triangulation method, we felt that
we were likely to get reliable information, particularly on things
like prices of products bought and sold by private traders,
ganyu rates, and farmers' yields of crops for the previous years.
[Note: This method was also used to check information given by
particular members of households, for example, initially we
found that in some cases the male household head claimed that
the household had enough food for their requirements through-
out the year; later, through observation and by talking with the
women household members, we established that in fact this was
not the case.]
Throughout the fieldwork, part of the Development Adviser's
role was to provide support and supervision for the researchers.
During the first six to seven weeks of the work this meant visit-
ing each of the researchers twice a week, although later the fre-
quency of the visits was reduced to once a week. During these
visits we discussed what had been happening in the village as
well as any problems encountered in the research work. In
some cases we were joined at these meetings by the Village
Headman or other members of the village, so that if any prob-
lems had emerged from their side, these could be discussed.
During many of these visits the Development Adviser was
accompanied by some of the service providers from the RDP
offices, the Community Development Office or from the District
Social Welfare office.
In addition to these meetings, throughout the fieldwork peri-
od we met as a group, with all the researchers as well as Louis
Msukwa (on several occasions), every two to three weeks for
one day, when we discussed our findings as well as preparing a
rough work schedule for the next few weeks. These meetings
were particularly useful because they enabled each researcher to

111
 talk about his or her experiences and findings, any problems
encountered and how they had sought to deal with these, and
to listen to the experiences of the others and compare their
own.
Despite all our efforts to try to ensure that people's expecta-
tions were not raised, just by our presence in their villages we
found that in some cases people approached the researchers
thinking they could somehow provide 'solutions' to their prob-
lems. Whenever we were faced with these situations our
response was always to make it clear that we were unable to
provide such 'solutions' but that we were willing and committed
to work with people in searching for possible solutions.
Throughout the research we were careful to encourage people
to think about solutions to their problems which they could
manage themselves. In this way we felt that we could enable
people to take more control over their own lives.
One area in which it was felt the researchers needed addition-
al training was gender awareness. On the recommendation of
Oxfam's Regional Representative we therefore arranged a
two-day refresher/training course in January in which we con-
centrated on gender as a development issue. For this training
session we used materials from Oxfam's Gender and
Development Unit as a basis for discussion. We also invited the
Women's Officers from Blantyre ADD and Mulanje RDP to these
sessions.
During the second stage of the fieldwork, once each of the
researchers had identified some of the most vulnerable house-
holds in their village, the researchers spent much more of their
time with these households. It was at this stage that we were
able to find out much more about intra-household inequalities,
since the researchers spent a lot of their time talking with differ-
ent members of the households.
The decision about whether we should extend the fieldwork
beyond the four months which had initially been envisaged was
made at the beginning of February. The two women researchers
had been offered full-time employment which they were due to
start in early March and so it would not have been possible for

112
us to continue working in Ligola or Chilembwe. The other factor
which was instrumental in our decision not to extend the field-
work was that we felt we would already have a good indication
of how much was likely to be harvested by most vulnerable
groups and the proportion of the produce which was likely to
be consumed prior to the harvest. In view of this we decided
not to extend the fieldwork beyond the end of February.

Postfieldwork
Throughout the four-month fieldwork period we made a con-
scious effort to keep people in the villages, including Village
Headmen and party officials, informed about our findings in
their own villages. In addition, officials at the DCs office, the
RDP's, Blantyre ADD, and other local officials in Mulanje were
regularly informed about our overall findings. At the beginning
of February we arranged a meeting with the DDC officials in
Mulanje to inform them of our findings, to find out from them if
there was any additional information we should try to ascertain,
and to get DDC officials thinking about some of the ways in
which we could build on the work of the ARP.
In March the first summary report of our overall findings was
produced and this was circulated to OPC, various Ministries,
Blantyre ADD, officials in Mulanje, as well as other interested
parties. The main aim of this was to find out whether the find-
ings were valid and to stimulate discussion and thought about
how we could seek to address the problems and priorities which
had been identified. During this process of consultation we were
able to get comments, ideas and suggestions from a variety of
sources and much of April and May was spent following these
up.
In addition to this, we felt that we owed a duty to go back to
each of the villages and inform people about our findings for
the District as a whole. We also tried to find out from them
whether they felt the findings were accurate (were there things
that we had missed out or things which had been given too little
or too much emphasis). This process of consultation was seen as
another opportunity for us to learn from the villagers but also to

113
 raise people's awareness of their own situation and in this way
to enable them to take more control of their own lives. Ideally,
we would have liked each of the researchers to go back to their
villages and spend a few days there during which time they
could inform/consult with everyone. Unfortunately, we were
unable to do this because four of the researchers had started
work with other organisations and were unable to take time off
work. Instead, each researcher went back to their village, and
spent three to four hours there, during which they briefed a
meeting of the village as a whole, as well as following this up
by talking with smaller groups of people. In many respects this
was not ideal, since the most vulnerable groups of people gen-
erally do not attend meetings, and when they do they are often
unable to speak in public. In most cases, however, we found
that because the researchers had worked particularly closely
with the most vulnerable groups these were well-represented at
the meetings and the researchers were able to talk with these
members of the community informally after the meeting. The
numbers of people attending each of the meetings varied
between 50 and 90-100 adults, plus children. Women were
well-represented at the meetings and in several villages outnum-
bered the men at the meetings. In one of the villages (Misomali)
we were able to involve the Theatre for Development Group.
They accompanied us to the village and performed a short series
of sketches involving the community, based on our findings on
the theme of apathy and its links with agriculture, health and
education in the village. We found this to be an extremely useful
medium of getting many more people actively involved in the
process of consultation. Unfortunately, the 'Theatre for
Development Group were unable to accompany us to the other
villages. Nevertheless, the scope for 'animation' work in the vil-
lages through a group such as the Theatre for Development was
very apparent.
The process of consultation at the village level was particular-
ly useful in terms of our findings. It enabled us to incorporate
some additional factors in the draft of the final report. Another
part of the consultation process before the final report was pre-

114
pared was a workshop held in late July. The workshop was
attended by Government Officials, local service providers from
Mulanje, as well as NGOs and other interested parties, and we
were able to discuss the findings and recommendations.

115
INDEX

Action research 99
Aerial photos 51, 91
Analysis 80
Anthropologist 65
Archives 50
Attitudes to development 2
Base-line data 31, 69
Base-line survey 12
Bias 35
gender 66
of key informants 38
personal 67
seasonal 64
control of 34
Botswana 26
Buid 53
Census 48, 80, 104
Check lists 68
Choice of researchers 42, 105
Class 19,36
Community 24, 37, 63, 83
Community self-diagnosis 11
Conflicts 15
Continuous research 30
Control of bias 34
Control study 25, 31
Cross-checking 15, 37-8, 48, 52, 71, 79, 109
Cultural brokers 42
Cultural indicators 78
Culture 37
Development
rural 12
attitudes to 2
Dissertations 49
Dry-season bias 66
Elite-bias 66

116
Emergencies 15, 27, 30
Empowerment 10
Ethnographic studies 49
Evaluation 8, 32, 83, 97
Expectations 11-13, 102
Feasibility study 6, 20
Fiji 72
Food shortages 15
Fuelwood 26
Gender 24, 35, 111
issues 44, 96
relations 22
Gender and Development Unit, Oxfam 111
Geographic information systems: 92
Government attitude 13
Group interviews 55, 70
Health problems 11
Household 22, 24, 26
Hypothetical questions 25
Income-generation 7
Indexes 50, 52
Indicators 69,78
Information networks 50
Information
sources of 47
Interpreter 39
Interviews 29, 54, 69
semi-structured 69
Key informants 38
Libraries 50,82
Life-histories 38, 55, 57
Literacy 31
Literature review 99
Logical framework analysis 94
Maharashtra 18
Malawi: 12, 99
Mapping 24,9
Maps 51
Mini-surveys 62
Monitoring 7,31

117
Mulanje District 99
Newspapers 51
Objectivity 32,34
Opportunity cost 29
Oral testimonies 56, 58
Participant observation 63
Participatory research 8-9, 10, 17
for evaluation 9
for monitoring 9
Participatory Rural Appraisal (PRA) 74
Pastoralists 57
People, as information source 51
Pilot study 26, 80
Poor people, identification of 23
Popular classes 36
Poverty, alleviation of 7
Power 18
Project Framework Analysis 94
Project planning 7
Publications 50
Qualitative 59
Qualitative research 45, 59
Quantitative research 45
Questionnaire 24, 29,44, 54, 58, 62, 88, 100
design of 88
testing of 44
Quota sampling 6l
Random sample 60
Ranking 68, 70, 78
Rapid Rural Appraisal (RRA) 66
limitations of 71
Raw data 47
Red Cross 16
Refugees, 15, 57
Registries 51
Report writing 82,112
Representative results 34
Research
access to 82
continuous 30

118
 costs of20
design 40
participatory 8,39
stages of 30,39
style of 14, 45
timetable 28
use of 84
Revolving fund 7
Role-playing 70
Rural development 12
Rwanda 16
Sample 16, 60
random 60
stratified random 6l
systematic 61
Sampling methods 60
Satellite photography 51,92
Seasonal biases 64
Seasonal patterns 20,27
Secondary sources 47
Semi-structured interviews 72
Set-up time 28
Shuar Federation of Ecuador 58
Site-bias66
Sketch-maps 70
Small-loans schemes 38
Snapshot exercise 30
Solidarity 36
SOS Sahel 57
Statistics47
Strategic level 21
Stratified random sample6l
Street children 5
Structural adjustment 40
Structural inequality 17
Subjectivity 32,36
Sudan 12, 17, 27
Survey 20, 27, 29, 43, 58, 62, 80
pros and cons of 62
Systematic sample 61

119
Target Oriented Project Planning 95
Tarmac-bias 66
Terms of reference 42, 96
Theatre for Development Group 113
Training43, 106
Triangulation 71, 110
UNHCR 16
Unit of analysis 21, 22, 24, 25, 26
Unit of trust 37
Universities 51,82
Validation 37,79
Water 6, 27
Wealth-ranking 24
Women, access to 17
Workshops 70, 72, 114
Zabballeen 5

120