THE NATURE OF EPISTEMOLOGY AND METHODOLOGY IN

DEVELOPMENT STUDIES: WHAT DO WE MEAN BY ‘RIGOUR’?

Andrew Sumner* and Michael Tribe#

Paper prepared for:

‘The nature of Development Studies’,

DSA Annual Conference, ‘Bridging research and policy’,
Church House, London, 6 November 2004

‘Exploring the frontiers in Development Studies epistemology and methodology’,

ESRC DSA Postgraduate Training Workshop,
Abbey Centre, London, 14 December 2004.

Abstract: Epistemology and Methodology in Development Studies are a special case.

Due to the nature of the subject matter, the simple transposition of generic social
science concepts and methods onto developing countries is often problematic. This
paper is concerned with the distinctness of development studies and what this implies
for ‘rigour’ in development studies research. The paper takes a journey, stage-by-
stage, through the research process: specification of problem and research question(s);
epistemological and ontological stance; research design; collection of data; and
analysis and findings. At each stage, epistemological and methodological issues are
noted that make development studies distinct, focusing the discussion throughout on
what this means for ‘rigour’ in Development Studies research. It is argued that many
generic concerns in social science concepts and methods are amplified in a developing
country context and that this creates a heightened need for attention to ‘rigour’ in
terms of research design and epistemological stance, replicability, interpretation and
conclusions.

*
Correspondence to Andrew Sumner, International Development Studies, School of Social Sciences,
Media and Cultural Studies, University of East London, Docklands Campus, 4-6 University Way,
London E16 2RD, UK. E-mail: a.p.sumner@uel.ac.uk Fax: +44 208 223 2898
# Michael Tribe, Bradford Centre for International Development, University of Bradford, Bradford,
West Yorkshire BD7 1DP, UK. E-mail: m.a.tribe@bradford.ac.uk Fax: +44 1274 235280

1
1. INTRODUCTION

Over the last few years, Development Studies (DS) has entered a period of soul-searching to
identify its distinctive and defining characteristics.1 In the UK this process has been
stimulated by the recent announcement of a stand-alone Development Studies sub-panel for
the 2008 Higher Education Research Assessment Exercise and the likely benchmarking of
Development Studies degrees next year by the UK higher education Quality Assurance
Agency.2 In light of these developments in teaching and research it would seem an opportune
time to discuss the nature of DS both conceptually and epistemologically/methodologically.
The former is the focus of Tribe and Sumner (2004) and the latter is the focus of this paper.
The concern of this paper is with the two central components of DS. Firstly, epistemology
in DS – how do we know what we know in DS? Secondly, methodology in DS – what is a
‘rigorous’ process for knowledge creation in DS? The answers to these questions play an
important part in research and teaching and identify the distinctive nature of DS.
The paper takes a journey through the research process in DS identifying at each stage the
distinctiveness of DS and what this means for ‘rigour’ in DS research. In the opening section,
the epistemological point of departure in DS is addressed. This is then followed by a stage-
by-stage journey through the research process in DS: the problem identification and research
question specification; the ontological stance; the research design; the collection of data; and
the analysis and findings.

2. THE EPISTEMOLOGICAL POINT OF DEPARTURE IN DS

Development Studies is often thought of as falling firmly within the area of social sciences.
However, given that DS is centrally concerned with the poor, the overwhelming number of
whom reside in rural areas, work in agriculture, and rely on an interaction with natural
resources, clearly natural science is also of importance (Molteberg and Bergstrøm, 2002a: 25;
Morton and Martin, 2004: 4). Taking an even wider perspective other technical areas are also
of relevance, such as the engineering logistics of service delivery and its distributional

1
This can be seen in recent special issues of journals covering the matter (see in particular, the articles by Shaw
and by Loxley in the Canadian Journal of Development Studies, 2004, 25:1; Kanbur, Harriss, Jackson and/or
White in World Development, 2002, 30:12; Maxwell in the European Journal of Development Research, 2003,
15:1; and a number of ‘progress reviews’ including Developments, 2000, 43:2, 43:4; and the Journal of
International Development, 2004, 15:7).
2
World Development has recently been established at A/AS level and significant interest already exists in
secondary school teaching (QCA 2003).

2
impacts (such as engineering for water, sanitation and electricity provision). How DS studies
deals with phenomena with differing ontological status is a central question to its distinctive
nature and defining characteristics, as are the implications for epistemology and
methodology.
Epistemology is the branch of philosophy that is concerned with the nature, origin, scope of
knowledge and ‘how we know what we know’. From Plato’s Theaetetus dialogue (Plato, 360
BCE) onwards the question of what is knowledge, or what constitutes knowledge, has been a
major area of philosophical inquiry. DS draws on both the empiricist basis of knowledge
(knowledge as human experience) as well an instrumentalist basis (our perceptions do not
necessarily reflect the real world but are useful instruments to explain our experiences).3 The
former, empiricism lends itself to a positivist epistemology for knowledge creation and meta-
narrative modernism while the latter, instrumentalism, lends itself to constructivist
epistemology for knowledge creation and the post-modern critique.4
Is DS an objective, natural science-like search for one ‘truth’ or is it a subjective,
constructivist search for many ‘truths’, or is it perhaps both? What constitutes legitimate
research goals and research practices in DS? Epistemology provides the philosophical
underpinning – the credibility – which legitimises knowledge and the framework for a
process that will produce, through a ‘rigorous’ methodology (consisting of the full range of
research methods), answers that can be believed to be valid, reliable/replicable and
representative/typical. Credibility is, of course, important in any academic discipline, but
even more so in DS because of the relative infancy of DS as a distinct field of study (by
comparison with more established fields of study) emerging recognisably only in the last fifty
years.
DS has a wide range of epistemological perspectives to choose from in both the natural and
the social sciences. At one end of the spectrum lies positivism and at the other end lies
constructivism: sitting in opposition as differing perceptions of the objectives of academic
inquiry and of the creation and generation of knowledge. Both are concerned with what
constitutes legitimate intellectual goals and practices (see Table 1). Across every conceivable
fault-line the two approaches sit diametrically opposed, raising numerous issues for DS such

3
This is somewhat different to the use of ‘instrumentalism’ in economics which refers to the arguments put
forward by Milton Friedman that theories should be judged not by their assumptions but by their predictions and
even if assumptions are known to be false they can still be useful as a stepping stone to more valid theories.
4
For a useful reference source see the on-line Stanford University Encyclopaedia of Philosophy at
http://plato.stanford.edu/

3
as triangulation, agnosticism, and determinism (closure) which are taken up below and in
Cameron (2004).
Table 1: Points of contrast: Positivist and constructivist approaches

A positivist approach A constructivist approach

What is ‘reality’? A definable ‘reality’ or ‘truth’ There is no ‘reality’ or ‘truth’
exists and is observable beyond our experiences
What is the goal of academic Acquisition of the ‘truth’ A more informed construction of
enquiry? the world
How are the researcher and the The researcher is independent of The researcher is not independent
‘researched’ related? the ‘researched’ of the ‘researched’
What should be the role for None - objectivity sought Part of ‘reality’ - subjectivity
values? celebrated
What kind of approach? Predominantly based on Predominantly based on discourse
observability or measurability and and meaning with the aim of
with the aim of seeking ‘evidence’ seeking a more informed
understanding of the world
What kind of data is preferred? Predominantly quantitative Traditionally associated with a
predominantly qualitative approach
Examples of such studies in Dollar and Kraay (2002) Growth is Narayan et al., (2002) Voices of the
Development Studies Good for the Poor Poor
Note: The characteristics set out in this table are regarded as tendencies rather than as absolutes. The two
examples have been abbreviated as Growth is Good for the Poor or “Dollar and Kraay” and Voices of the Poor
or “Narayan” throughout the remainder of the paper.

Positivism is ontologically posited on the contention that reality and universal ‘truths’ are
observable. Academic inquiry is the search for these ‘truths’. These ‘truths’ can be observed
in an objective and independent manner by the researcher, uninfluenced by the researcher’s
values and assumptions. This appeals to the natural science element of DS, a scientific search
for ‘truth’ through objective, experiments usually based on a quantitative (i.e. measurable)
approach. For example, econometrics is one technique derived from such a positivist
approach, of which Dollar and Kraay’s (2002) Growth is Good for the Poor is a good
illustration (see Box 1). This study was based on the logical positivism credentials of
econometrics: a ‘view from above’ of one, quantitative and objective ‘reality’, with
verification through quantitative testing of hypotheses.
The philosophy of (logical) positivism is associated with the 1920/30s work of Moritz
Schlick, Rudolf Carnap, Otto Neurath and Hans Reichenbach amongst others.5 Within the
positivist epistemology, a quantitative approach is often taken to be more ‘rigorous’ (this
point is developed further in the following section – The Research Process in DS). Within the
logical positivism framework there are only two forms of knowledge: logical reasoning and
empirical evidence (i.e. refutable/testable propositions). The former is a priori and the latter a
posteriori. The former is synthetic and the latter analytic. Advocates argue that research must

5
For discussion of their work see Giere and Richardson (1997). In a later generation Karl Popper would fall into
the same category (1968).

4
satisfy the standard natural science model of observable phenomena and the ‘verification
principle’: If a statement cannot be tested it is of no consequence, and is not of any research
interest, and a statement is only meaningful if it can verified by reference to logic or
evidence.6 Knowledge is only produced through scientific method. It is assumed that an
objective reality can be described through measurement and quantification (i.e. ‘observed’)
and that the ‘reality’ is both independent of the researcher and of the instruments of research,
reflecting the importance placed upon the replicability of experiments and the objective
position of the researcher in the natural sciences.
A prominent concern to DS from this approach is the role of measurability of social
phenomena and the widespread use in DS of proxies. Proxies may be imperfect not only in
terms of whether they conceptually capture the phenomenon in question but also empirically,
in terms of the quality and accuracy of what is captured.7 Dollar and Kraay use an ‘income’
definition of poverty derived from data in household surveys. Such surveys tend to record
‘expenditure’ and consumption rather than income and under-reporting is often a problem.
Constructivism is posited ontologically on the premise that reality, the world, as the subject
of research does not exist independently from our experiences. In addition, constructivism
argues that multiple realities exist which are intangible, local and specific in nature. The
concept of a single ‘truth’ is meaningless, as is any project to accurately describe the world.
All claims to ‘closure’ are suspect. Academic research should strive towards ever more
sophisticated, informed and inclusive constructions of the world through the interaction of the
researcher and the researched (Molteberg and Bergstrom, 2002a: 21). The constructivist
approach appeals to the social research component of DS – a subjective search for meaning
and understanding of the human condition. A recent and well known example is the Voices of
the Poor study by Narayan et al. (2002). This study is based on the epistemological
credentials of a social constructivist, anthropological approach to research with a ‘view from
below’, of many subjective and generally qualitative ‘realities’. The philosophy of social
constructivism is associated with the 1960s work of Peter Berger and Thomas Luckmann and
later, Karin Knorr Cetina and Bruno Latour.8

6
According to the Oxford English Dictionary, the word ‘evidence’ originates from the Latin evidentia, from
evidens meaning ‘obvious to the mind or eye’.
7
‘Dummy variables in regression analysis come into a similar category to ‘proxies’ in this context.
8
For further discussion see Berger and Luckmann (1966) and Hacking (1999).

5
 Box 1. Contrasts in Development Studies: “Dollar and Kraay” and “Narayan”.

David Dollar and Aart Kraay are two World Bank economists, who have published a number of studies using
cross-country quantitative analysis based on econometrics. Growth is Good for the Poor was originally released
as a World Bank Working Paper (Dollar and Kraay, 2000). The Working Paper was a background document for
the World Bank’s World Development Report 2000-2001 on Poverty (World Bank, 2000). Growth is Good for
the Poor (2002) argues that it has provided ‘evidence’ (the word was used 37 times in the study) to support the
proposition that not only is growth good for the poor, but that the policies pursued by the IMF and World Bank
have been good for the poor because they have led to growth. Although the paper has faced sustained
methodological criticism (see for example, Amann et al., 2002; Nye et al., 2002; Rodrik, 2000) few have
disagreed with the main finding that the income of the lowest quintile in the income distribution rises
proportionally with average per capita income growth.

The Narayan et al. (2002) participatory poverty assessment was also carried out by the World Bank, and much
of the analysis was included in the World Development Report 2000-2001 (World Bank, 2000). It was one of the
World Bank’s most ambitious studies to date covering 60,000 poor people in more than 60 countries. The study
had two components, and was carried out with help from NGOs and independent research centres. The first part
is a literature review of 75 participatory poverty studies from the late 1990s with coverage of 40,000 people in
41 countries. The second part consists of new studies in 23 countries covering 20,000 people. The studies sought
to elicit poor households’ perspectives on well-being.

In contrast to Dollar and Kraay, the poverty definition used in the Narayan study is much
broader definition of poverty including non-economic dimensions of security, vulnerability
and empowerment to name but a few. This difference in focus between the two studies
represents a fundamental methodological division between purely economic concepts of
poverty (‘income’ poverty) and broader concepts as used by the Narayan study. In essence,
the possibility of finding an objective definition of the concept of poverty is questionable, and
the objectivity of data may also be questioned.
The contrast of the Growth is Good for the Poor and Voices of the Poor illustrates how
broad a church DS is epistemologically and methodologically (see Table 2). The Dollar and
Kraay study was based on quantitative analysis of secondary data and on an experimental,
natural science type method/technique, while the Narayan study was based on qualitative
analysis of both secondary data and primary data from specially conducted household
surveys. Dollar and Kraay’s study was a descriptive-explanatory study (what is the
relationship between x and y?), while Narayan’s study is an interpretative study (what are x
and y?). In sum, within the same academic field and sub-field (poverty) two quite different
ends of the epistemological and methodological spectrums are counterposed.
Table 2. Epistemological and methodological stances in
Growth is Good for the Poor and Voices of the Poor

Growth is Good for the Poor Voices of the Poor

Epistemological stance Positivist Constructivist
Methodological stance Quantitative Qualitative

6
For DS the epistemological battleground is much concerned with what is ‘known’ (and what
‘known’ means) and the relationship between the researcher and the researched.9 Much of DS
is normative: it is not concerned with knowledge creation for its own sake, but as an
instrumental means of contributing to the improvement of natural and social conditions for
the sake of improved welfare (Molteberg and Bergstrom, 2002a: 7). In this respect a middle
ground of realism has been proposed by Molteberg and Bergstrom (2002a: 21). This
epistemological stance distinguishes between physical reality (positivist) and human
cognition (constructivist). The ontological basis of realism is that there is a physical reality
that exists independently of our cognition but that we cannot appraise it – we can only
describe it due to the fact that we are ‘dependent’ observers – not independent of events.
Thus knowledge is a social construct but one which aims to explain a physical reality. The
question of subjectivity-objectivity amounts to accepting that DS research is subjective and
acknowledging that there is ‘bias’, some of which is acceptable and some of which is not.
‘Rigour’ in DS research means that it is necessary to distinguish between misinterpretation or
omission which arises deliberately to bias the results in a direction which serves the purposes
of the researcher (this being absolutely unacceptable), and acceptable, or acknowledged bias
in values (such as recognising that poverty is a multi-dimensional concept) or data (where
biases can be systematically identified). The research process is now reviewed stage by stage
with a focus on ‘rigour’.

3. THE RESEARCH PROCESS IN DS

The research process, not least in DS, can be seen as a cycle (see Figure 1). Each stage leads
to the next, drawing upon Galtung’s (1967) approach to social research.10 The first stage
identifies the research problem and specifies the research question(s). The second stage
involves the ontological choice of theoretical/conceptual framework and then the research
design. The next stage is that of data collection, followed by analysis and findings. The final
stage involving assessment of priorities for future research leads back to stage one in a
complete cycle. Each stage is now discussed in turn.

9
This battleground relates most to those types of research which involve the collection of primary data,
although similar questions can be asked about secondary data – as for example in the case of the Dollar and
Kraay study.
10
DS has generally utilised generic social science methodology books (for example, Burns, 2000; Bryman,
2001; De Vaus, 2001; Denscombe, 2003; Fink and Kosecoff, 1998; Galtung, 1967; Gilbert, 2001; May, 2001;
Punch, 1998;) arguably because there are few dedicated DS methodology books (other than Laws et al., 2002;
Mikkelsen, 1995; Scheyvens and Storey, 2003).

7
 Figure 1. The Research Process in Development Studies

1. Problem identification
and specification of
research question

6. Future
research
priorities 2. Ontological choice of
theoretical / conceptual
framework

5. Analysis and
findings

3. Research design

4. Data Collection

i. Problem identification and research question specification

The first stage of the research process is the identification of a problem. This is a search for a
problem which is researchable. This starting point is rarely as ‘rigorous’ as one might think:
as Cameron (2003:3) notes:

[t]he start of the [research] process is socially haphazard. The individual researcher takes an
unsolved problem from individual experience, from the existing literature, from a promising
commercial opportunity, [or] from a sense of patriotism.

The research problem is often a choice of personal, organisational and/or funding priorities as
to what is worthy of study. What constitutes a ‘good’ or ‘rigorous’ research problem? This
might be considered from the point of view of what constitutes the PhD’s ‘contribution to
knowledge’ (see Box 2). The contribution has to be real and tangible and will be incremental
to the existing body of literature. Given that there are relatively few totally unresearched
areas in DS, the focus has turned to taking existing problems and then developing the current

8
level of understanding further by reframing the research issue, by connecting hitherto
unconnected phenomena, by collecting new up-to-date data (or reinterpreting old data), or by
challenging orthodox beliefs with new or reinterpreted data.
Box 2. What Constitutes Originality in Development Studies PhD research?

A PhD course of study is distinguished from other study (notably an MPhil) because it is an ‘original
contribution to knowledge’ and is identified as such in university postgraduate handbooks and in assessment
criteria. It is also increasingly a period of research training. What is an ‘original contribution to knowledge’?
The Dearing (England, Wales and Northern Ireland) and the Garrick (Scotland) reports both recommended a
unified framework for higher education qualifications in order to maintain qualification standards. The HEFCE
QAA Framework for Higher Education Qualifications identifies a PhD award as follows:

‘Doctorates are awarded for the creation and interpretation of knowledge, which extends the forefront of a
discipline, usually through original research. Holders of doctorates will be able to conceptualise, design and
implement projects for the generation of significant new knowledge and/or understanding’.

Refer to the Higher Education Funding Council for England – Quality Assurance Agency: website at

http://www.qaa.ac.uk/crntwork/nqf/nqf.htm

What does this mean in practice? A PhD research programme could relate to one of the following on a non-
exhaustive list:
1. A new research problem
2. A new way of looking at an old research problem
3. An old research problem revisited with new data or theory/concepts
4. A challenge to orthodox beliefs

While the research problem relates to a general definition of an area, the aims and objectives
of the research are specific and the research questions/hypotheses attempt to specify a
feasible research project that effectively addresses the problem identified. Research is about
setting and answering questions, and this especially so in DS. For example, what is poverty?
Or is economic growth beneficial to the poor?
How are the aims and objectives and the research questions formulated? Aims tend to be a
general focusing of the research problem into a sub-area of interest, and then objectives make
the overall aims specific through research questions and hypotheses which identify the
questions and statements which are to be answered and tested.
Research questions may be interpretive (what is x?); descriptive (how does x vary with y?)
or explanatory (does x cause y?). What constitutes a ‘rigorous’ research question? Qualities
that might be highlighted here are the alignment of the question with the problem; whether
similar research has been done already; building on current understanding; the clarity of the
question (is it phrased in a clear, operational form?); the scope of the question (is it too much
to answer within the study’s constraints or is it too broad to be meaningful?); the feasibility of

9
being able to answer the question or hypotheses (time, resources, accessibility of data); the
implications of the question for the independence or otherwise of the researcher in answering
the question; and the nature of the relationship between the subject and the researcher.
Additionally, there is emphasis in social science (and particularly in DS) on the policy
relevance of research. If we take a research question such as “is growth is good for the poor?”
there are clearly strong policy implications. Finally, the epistemological stance may be highly
deterministic – is the research question verifiable – can it be proved true or false – or is it
irrelevant? Table 3 indicates the research problems identified and research question specified
in Growth is Good for the Poor and Voices of the Poor. Both have a clear problem identified,
and have a focused, relevant, feasible and epistemologically derived research question.
Table 3. Research problems and questions in Growth is Good for the Poor and Voices of the Poor

Growth is Good for the Poor Voices of the Poor

Research problem If economic growth is shared then Attempts at universal definitions of
there is no need to depart from an poverty have not included the
unmitigated growth strategy, but if meanings poor people place on
this is not the case then different their own situation
policies are required
Research question Does the income of the poor What does being poor mean to
(bottom expenditure quintile) rise poor people?
faster than average incomes in the
process of economic growth?

How does the identification of the problem and specification of the research question differ in
DS compared to social or natural sciences? This is largely a matter of researchability. What is
researchable in an industrialised country can be quite different to what is researchable in a
developing country – notably in two areas: i) accessibility and reliability of data and ii) the
nature of the researcher–researched relationship. The accessibility of both primary and
secondary data in developing countries is often problematic: there are issues of gate-keepers
and concerns over which data is and is not possible to collect and the validity, reliability and
the extent to which the data collected is representative. Furthermore, the researcher–
researched relationship is far more problematic if those involved in the research are at a
completely different level of economic and political power. In sum, at this stage of the
research process, ‘rigour’ refers to a focused problem and to the clarity, feasibility and
researchability of the research question or hypotheses. From this starting point ontological
choices over theory/conceptual frameworks can be made and the research design can be
planned.

10
ii. Ontology, theory and conceptual frameworks in DS

Ontology refers to a set of basic assumptions about inter-relationships between phenomena

within the world which is the subject of the research. It is an explicit specification of how
phenomena are assumed to relate to each other. In DS ontology is an area of some ferment
(see Appendix 1). The best known theories and frameworks – as taught to students of DS –
Modernisation theory; Underdevelopment theory; Structuralism and Dependency theory, to
name a few, are no longer widely utilised in contemporary research. Those largely positivistic
(in terms of one ‘reality’) approaches have been replaced with a range of constructivist,
conceptual frameworks to guide research in DS. Each with a unique purpose. The choice is
between such frameworks as Sustainable Livelihoods, Rights Approaches, Gender and
Development, and Participatory Approaches. Different theory is appropriate for different
contexts and purposes. DS also ‘imports’ theory from constituent disciplines. For example,
concerning the factors explaining changes in the income distribution in the process of
‘development’ from economics. Finally, perhaps surprisingly, some research in DS has been
theory-less, presenting a risk of excessive description replacing analytical rigour.

iii. Research design

The research design is the operationalisation of the research question. This includes the
specification of data required to ‘answer the question’ or ‘test the hypotheses’ and how this
data is to be collected. This is a deterministic stage in research, even more so in DS because
choices made at this stage shape what kind of questions and hypotheses can and cannot be
answered. For DS a range of issues present themselves: all are concerns in social science but
are amplified in DS research. They include how the focus of the research question is to be
observed or measured? What relationship will be formed between the researcher and the
‘researched’? and is the study to be objective or subjective?
Two questions can asked of the research question in order to inform a ‘rigorous’ research
design. The first is an epistemological choice and the second is a methodological and
methods choice. What kind of answer is the research question looking for? Are we looking
for one objective answer or many subjective answers? And are we looking for a qualitative or
for a quantitative answer, or for both? The type of research question will determine the
epistemological and methodological approaches to be selected. Indeed, the choice between
the two depends on what answers are sought. It is somewhat of an over generalisation, but

11
descriptive (how does x vary with y?) and explanatory (does x cause y?) type research
questions are associated with positivism and quantitative research due to perceptions of its
objective nature, while interpretive (what is x?) research questions tend to be associated with
social constructivism and qualitative research due to its more subjective nature (Mikkelsen,
1995:213).
A positivist approach tends to be more quantitative, and based on published secondary data
from official sources.11 It seeks one observable, verified, aggregate, objective, ‘truth’. In
contrast a constructivist approach has a tendency to be associated with a more qualitative, and
based on primary data collected specifically for the research in question. It seeks a ‘view
from below’ of many ‘truths’. The choice is objective truth (positivist) vs. subjective
meaning/interpretative (social constructivist) or some attempt to triangulate both. If the
research question demands generalisations then positivist epistemology and more quantitative
methodology may be more appropriate. On the other hand, if the research question demands
an attempt to understand and explain social phenomena then a social constructivist
epistemology and more qualitative methodology may be more appropriate.
The objective-subjective question is of interest to DS and relates to acceptable and
unacceptable bias in research. Table 4 outlines examples of bias in primary and secondary
research in DS. Bias can be introduced by the researcher, by the ‘researched’ or by the
research process.
Table 4: Types of Bias in Development Studies

Bias introduced by researcher Personal values

Design of research – specification of research problem
Rejecting evidence that does not support hypothesis
Bias introduced by the ‘researched’ Under-reporting
Imperfect recall
Responding on behalf of others
Bias introduced by the research process Availability of accurate sample frame
Imperfect access to data
‘Interviewer influence’

Are quantitative methods more ‘rigorous’? White (2002: 512-3) argued that different
techniques are appropriate for different settings and that there is a prevailing presumption (in
economics at least) that quantitative methods are more ‘rigorous’ than qualitative because
they are perceived as less subjective and more tangible, and qualitative methods are perceived
as being more subjective and thus less ‘rigorous’.12 However, White challenged this and

11
Some data – for example, national accounts data – cannot be collected on a primary basis by researchers, thus
some types of research are dependent upon official sources.
12
But, for example, see Olsen (2003) for an empirical view of how qualitative research can be made more
rigorous and systematic.

12
called for synthesis between qualitative and quantitative methods. He noted that the basis for
claims to ‘rigour’ relates to how the techniques are applied; that badly applied qualitative and
quantitative approaches could lead to inaccurate conclusions and different techniques suit
different purposes. Table 5 outlines the strengths and weaknesses of quantitative and
qualitative approaches.13
Table 5: Quantitative and Qualitative Approaches: Which is More ‘Rigorous’?

Quantitative Qualitative
Strengths Weaknesses Strengths Weaknesses
Representative Lack of validity Validity Unrepresentative
Reliable Shallow/‘thinner’ Depth/‘thicker’ Unreliable
Objective? Closure? Lack of closure? Subjective?

The methodology of the study relates to how methods combine to generate the research
data which forms the response to the research question. There is a wide range of approaches
utilised in DS data generation and collection including surveys, case studies, experiments,
action research and ethnographic approaches. Of course, none of these approaches is perfect.
Validity, reliability and the extent to which data is representative differ across approaches on
a depth vs. breadth/generalisability trade-off, or a ‘thicker’ versus ‘thinner’ description. One
might imagine a spectrum with ‘thinner’ descriptions at one end and ‘thicker’ at the other end
(see Figure 2).
Case studies have significant depth in the data created but little breadth or generalisability –
and are nearer the ‘thicker’ end of the spectrum. In contrast, surveys have wide breadth and
limited depth and are nearer the ‘thinner’ end of the spectrum. In a ‘rigorous’ research design
the research question should guide the choice of approach. Descriptive (how does x vary with
y?) and explanatory (does x cause y?) type research questions might tend to ‘thinner’
approaches such as surveys. In contrast, an interpretive (what is x?) research question would
tend towards ‘thicker’ approaches such as case studies.

13
A good example is the dominance of quantitative measures of poverty (especially income). It could be argued
that the dominance of economic measures is based on the preconception that economic measures are more
precise and objective because they are amenable to quantification. They are tangible – consumption of a certain
amount of rice in kg can be recorded (assuming there are no recall and respondent bias). In contrast, non-
economic measures are somewhat less amenable to quantification and rely on more tenuous and subjective
proxies – for example equating being ‘educated’ to the subjective concept of ‘literacy’. It is perhaps assumed
that what is more amenable to quantification is more objective (i.e. the same to all people). For example, one kg
of rice or one dollar-a-day is the same to everyone. However, it could be argued this would be a false
proposition because the gain or loss of a dollar or a kilogram of rice has a different welfare impact on a
poor/hungry person than someone else better off/not hungry – the principle of diminishing marginal utility of
consumption.

13
 Figure 2. Thinner and thicker descriptions

Case Study approach

Survey approach a) Few Observations
a) Many Observations b) Not Statistically
b) Statistically Significant Significant
c) Less Confidence in c) More Confidence in
Individual Responses Individual Responses

What determines choices over methods/techniques to be used to collect the data? DS uses a
range of methods: in primary data collection, interviews, questionnaires, and observation are
used, while in secondary data collection, official statistical data and documents tend to be
used. Typically, combinations or triangulation are proposed to overcome the validity
weakness in quantitative methods and in the reliability and representative weaknesses of
qualitative methods. Indeed, given DS claims to multidisciplinarity the mixing of methods is
common. Multiple methods are useful to corroborate and ensure validity, not providing proof
but improving consistency across methods in a process of triangulation.
In terms of choice of methods, the research question is again deterministic in ensuring a
‘rigorous’ approach. The questions to be addressed are: what kind of data are required? What
are the practical considerations (time, resources and access), how is the focus of the research
question to be observed/measured, and to what extent can proxies be used in place of direct
measurement of variables?.
Descriptive (how does x vary with y?) and explanatory (does x cause y?) types of research
question might tend to quantitative analysis – especially with secondary data. In contrast, an
interpretive (what is x?) research question might tend more to qualitative methods approaches
– especially with open-ended interviews.
The final methods decision is what unit of analysis – country level, regional, community,
household or intra-household level? Each has implications not only for the type of data to be
collected but also for aspects and nuances of the data that might not be captured, such as
intra-household inequality.

14
 Table 6 shows the research designs of Growth is Good for the Poor and Voices of the Poor.
In Growth is Good for the Poor, the epistemology was positivist, the methodology
quantitative and the data analysis involved econometric regression using 285 pieces of
secondary data from 92 countries. Dollar and Kraay’s research design has been criticized for
a lack of ‘rigour’ on the grounds of using of cross-country data (with little use of cross-
temporal data) to investigate a temporal relationship (of the 92 countries only India, Mexico
and Sri Lanka had some time series data); that the sample did not differentiate between
developing countries and developed countries, and that the lowest expenditure quintile is not
a good proxy for the poor (those below the poverty line).14 In contrast, in Voices of the Poor
the research design was within social constructivist epistemology because it involved
interpreting how poor people themselves understood the phenomenon of poverty. It will be
recalled that this study assembled data based on i) a literature review of participatory poverty
studies conducted in the late 1990s (secondary data), and ii) a series of original participatory
poverty studies in 1999 (primary data) using interviews and focus group discussions.
Table 6. Research design in Growth is Good for the Poor and Voices of the Poor

Growth is Good for the Poor Voices of the Poor

Research design - Econometric regression - Participatory poverty studies from
primary and secondary

- 285 pieces of secondary data - 60,000 people in 60 countries

from 92 countries covering four
decades

- Data on economic growth, - Data on perceptions of a good

income poverty and income life; most pressing problems and
inequality priorities; quality of interactions
with institutions; and changes in
gender and social relations.

In summary, ‘rigour’ in the context of research design consists of the following of a logical,
integrated process. The research question leads to a choice of epistemology, leading to choice
of methodology, leading to a choice of data collection/sources, leading to choice of data
analysis methods. Thus the appropriate epistemology, methodology, data sources, and
analytical methods/techniques needed to answer a particular research question have to be
selected with great care. Arguably research design in DS requires far more care than is usual
in the natural or social sciences because of the difficulties in accurately capturing the social
phenomena in question, the feasibility of the research, and how this constrains DS methods.
These issues are pursued further in the next section.

14
For a full critique see the studies noted in Box 1.

15
iv. Data Collection

The data collection stage is the operational stage when the methods chosen are used to create
the data that will be analysed in the final stage. As a point of departure, data collection
through either primary or secondary methods must relate directly to the research question or
hypotheses if it is to directly and unambiguously answer the objectives of the research.
Arguably it is at this data collection stage that ‘rigour’ is most crucial and most contentious.
The data collection stage is particularly problematic for DS because it is at this stage that
generic concerns of research methods in social science can be amplified.
The first issue is that of access to the data. For primary work this means access to the
sample, and for secondary work, access to data and documents which already exist. Primary
work creates a whole range of ethical considerations in terms of the relationship which will
be formed between the researcher and the ‘researched’. Questions to be addressed include
who does the actual primary data collection – community members or outsiders? Accurate
data collection might be based on well trained local researchers (if they are available) but
beyond accuracy (perhaps due to over-familiarity of enumerators with respondents – the
relationship can work both ways) a whole range of ethical issues present themselves. For
example, gaining ‘informed consent’; avoiding undue intrusion; safety of and risk to the
respondent or researcher; guaranteeing confidentiality and anonymity; fair returns and
reciprocity with respondents and addressing raised expectations that research (for example on
poverty) may bring; explaining the purpose; cultural influences and language differences;
attention to sensitive issues and legal issues.15 Because of this long list of ethical concerns in
primary research, many UK universities now have a research ethics committee that all
primary research involving personal data collection must go through.
It can be argued that ‘rigour’ at this data collection stage refers to: (a) the close alignment
of the data collection with the research question; (b) some reasonable precision in capturing
the subject phenomena through standardised techniques and (c) attention to objectivity and
subjectivity, to acceptable or unacceptable bias and influence or subjectivity.
The first issue is self explanatory. The data collection methods must generate data which is
directly relevant to the research question and irrelevant material needs to be removed. The
second is a particular issue in social science research and in any inquiry into human subjects.

15
For more detailed discussion see Laws et al., (2003: 233-247) and Mikkelsen (1995: 250-281) for example.

16
Proxies are often used for social phenomena that cannot be directly observed – the selection
of ‘indicators’ involves questions of the extent to which data is accessible and representative.
For example, the most commonly used (and arguably, the most controversial) indicator for
poverty is the dollar-a-day line, which many researchers would regard as a proxy rather than
as a direct indicator. There are various problems with this popular proxy, not least the fact
that we cannot be sure whether a human being can live on this daily expenditure in all
countries although we do know that an average human can survive on 2100 calories per day
(Ravallion, 1998; Redde and Pogge, 2002). Another common income poverty measure which
is used in Growth is Good for the Poor is the lowest expenditure quintile. This definition is
inconsistent with measures based on numbers living below a minimum standard (the more
common income/expenditure basis for the definition of poverty – based on ‘headcount’
measures). On the other hand, in Voices of the Poor, the capturing of poor people’s own
interpretation of poverty has a significant internal contradiction: to identify those who were
poor, in order to interview them, a (universal) definition of poverty and existing poverty data
were used in the construction of the sample frame and in the survey design. If the existing
poverty data was questionable then the results of the study may have been biased, for
example by asking the non-poor about the meaning of poverty.
The third issue is central to ‘rigour’: the question of acceptable/unacceptable bias. If the
researcher has the responsibility for selection of appropriate data they may find themselves
engaged in a range of decisions that are deterministic.16 To illustrate this point, White (2002:
513) described two researchers: the ‘data-analyst’ and the ‘data-miner’. The former lets the
data tell the story and draws conclusions from the data, while the latter digs for data to
support positions which have been decided before the ‘research’ started (see Table 7 for
examples). If this could be seen as a spectrum with the ‘data-analyst’ at one extreme and the
‘data-miner’ at the other then where would most researchers fall? White, an econometrician
himself, suggests that any reasonably competent econometrician could generate the desired
findings given sufficient variables through a process of ‘data-mining’ – an example of
unacceptable bias.

16
Arguably peer review of publications is the means by which academics ensure ‘rigour’ here and throughout
research.

17
 Table 7. Examples of ‘data-mining’ in primary and secondary research

‘Data-mining’ in primary methods ‘Data-mining’ in secondary methods

Sample chosen to support hypothesis Data source chosen to support hypothesis
Research instruments written to support Data inconsistent with hypothesis rejected
hypothesis
Data interpreted to support hypothesis Data ‘cleaning’ - data unacceptable in various
other ways is rejected

All of these issues apply to both primary and secondary research because both go through
the same process, the difference being that the researcher has more control over primary data.
In addition, there are a number of method-specific ‘rigour’ issues to contend with in DS that
social scientists in industrialised countries need to take account of but not to the same extent.
For example, in primary research, samples in DS are highly problematic due to incomplete
official records (such as national identity card or electoral register) for the generation of a
sample frame. Examples can be found where researchers have studiously constructed their
own sample frames, rather than relying on previously available frames (e.g. Edusah, 1999;
Mensah, 1999). Sampling errors are near certain and lead to under-representation of some
groups in socio-economic surveys. In particular those without full ‘legal status’ may be more
likely hidden. It is likely that a disproportionate number of ‘hidden’ households will be
poor.17
Additionally, in a developing country setting, primary methods such as interviews, focus
groups, questionnaires and observation may lack reliability and validity due to under-
reporting and recall difficulties or to concern over the use of information. ‘Interviewer
influence’ may be exacerbated by inhibitions and perceptions created by the interviewer-
researchers such as dialect/accent, sex, age, class, education, formality, appearance, and/or
ethnicity. Local socio-cultural factors may also influence responses: Household heads or
village leaders may answer on behalf of respondents. Responses may be formed by culture,
who is present, or what the respondent thinks the interviewer wants to hear.
In light of the above, what might a ‘rigorous’ approach to primary research in DS entail? In
particular, the following non-exhaustive list can be specified: i) ensuring that the purpose of
the research is clear (the instructions); ii) piloting of the research instrument to ensure that all
questions are relevant to data needed, have a high level of clarity, are not leading (bias), are
not vague and are well ordered; iii) that the questions are directly given to respondents, and
that a full written or taped recording of the interview/observation is made. The overall

17
For example, in Indonesia, Suryahadi and Sumarto (2001:15) noted: ‘It is apparent that many… …do not
possess Jakarta identity card and are not officially registered with the local authority as residents. Hence their
existence is not in the survey data’.

18
objective is to minimise bias and to raise reliability and validity. Indeed the main concern
with primary interviews is that they lack reliability – the interviewer may have influenced the
answers or the respondent, for whatever reason, did not answer truthfully. One illustrative
example for DS is the Voices of the Poor study, which might well have raised expectations
about future aid. The only way to corroborate the validity of the interviews is to compare
with other information. As Denscombe (2003:186) observed, ‘[u]ltimately there is no
absolute way of verifying what someone tells you’.
Secondary research is also popular as a method in DS. In most developing countries
secondary data and documents are cheap and readily available, and raw data tapes are often
accessible. With respect to the quality of secondary data, one problem is the consistency
across time and across changing definitions. The quality is open to question if it fails a basic
test of consistency, and using data from different sources is highly problematic (all data
would need to be standardised across all countries if complete consistency was to be
achieved). For example, data presented on maternal mortality in the UNDP Human
Development Report and World Bank World Development Report is not always consistent.18
A non-exhaustive list of salient questions for reflection might include: How has secondary
data been generated? Who collects them and for what purpose? How is the sample frame
created? Who is omitted? What definitions are used? What are the data used for? However,
whilst these points are important they may be regarded as being somewhat ‘academic’ when
data availability is limited and the choice of indicators may be dictated by what is in
existence.
Turning to the examples of Voices of the Poor and Growth is Good for the Poor, their
approaches to data collection differs (see Table 8). Voices of the Poor was based on interviews
and focus group discussions in two components, the first a literature review, the second a new
set of studies in 23 countries. As noted by Narayan et al., (2002: 19-20) the main problems in
data collection were bias in interpretation and coding, the fact that the studies used differing
sample frames, and that source studies were not necessarily intended for the purposes for which
they were used. In contrast, Growth is Good for the Poor was based on secondary data and on
econometric techniques. Dollar and Kraay’s methods have been criticised for the fact that much

18
Loup and Naudet (2000, p. 11) cite a comparison of maternal mortality rates in the Human Development
Report (HDR) and World Development Report (WDR) in the mid 1990s. The WDR listed 56 countries with data
and the HDR listed the same countries (minus one) and a further 48. Of the 55 listed in the WDR (and in the
HDR) only a quarter were within a similar range – and a half were significantly higher and a quarter
significantly lower as compared with the data in the HDR.

19
of the data used in the study’s regressions were not recent data but from the 1970s and 1980s
(Weisbrot et al., 2001).
Table 8. Data collection in Growth is Good for the Poor and Voices of the Poor.

Growth is Good for the Poor Voices of the Poor

Data collection - Secondary data (sources: World - Secondary data (source: literature
Bank data; UNU-WIDER database review)
and other published datasets).

- Technique - econometrics - Primary data (23 surveys of

regression 20,000 people.

- For details see Dollar and Kraay For details see Narayan et al.,
(2002: 14-19; 44). (2002:253-268)

‘Rigour’ in data collection requires a highly reflective and critical practitioner. Ideally, data is
available from a variety of different sources and methods so that it can be corroborated in
analysis. This is the focus of the following section.

v. Analysis and findings

With data in hand the DS researcher enters the final stage of research, data analysis and
findings – but what is data analysis and how does it form findings? Data analysis is the stage
when the researcher takes all data collected and considers, examines, categorises and
interprets. Then trends, regularities and patterns are identified and data tabulated in order to
answer the research question and to test the hypotheses, and to present the data in a clear
manner.
What constitutes ‘rigorous’ data analysis? The data which has been collected should be
directly linked to the research question/hypothesis, and to objectives. The analysis may face
criticism over subjectivity in analysis and interpretation (especially if qualitative data is
collected in the local language – the connotations and cultural understanding of language are
open to interpretation) or the strength of any findings made may be challenged.
Strictly speaking the distinction between qualitative and quantitative approaches is about
analysis and the treatment of data rather than the actual methods (Denscombe, 2003: 231).
Although in practice qualitative and quantitative analysis are not mutually exclusive, for the
sake of identifying differing issues with regard to ‘rigour’ in analysis and different
requirements for each, it is useful to consider them in turn.
For quantitative data the analysis can take the form of descriptive statistics or the
interpretation of the outcomes from complex regressions. Issues that may be contentious

20
include the results from significance tests and claims about the direction of causality. In more
complex modelling such as econometrics, these statistical issues of ‘rigour’ may extend to the
instrumental variables used for endogeneity – disentangling causality (e.g. a positive
correlation between capital flows and growth says nothing about the underlying causes and
the direction of causation, but it is difficult to find variables which are not correlated to both
capital flows and growth). Furthermore, there are differing ways of dealing with the
simultaneity bias – the isolation of effects. Several elements present at the same time as
capital flows may contribute to growth. Controlling for country specific effects and omitted
variables are yet more contentious issues.
For qualitative data, the first issue is that a form of analysis has to be selected. There is a
wide range of essentially post-structuralist analysis of language and meaning in qualitative
analysis including discourse analysis, ethno-methodology, phenomenology, grounded theory,
hermeneutic research, content analysis, life history studies, structural ethnography and
symbolic interactionism. Then, usually but not always, a software package (such as the
Ethnograph, NUD*IST, NVivo, winMax, ATLAS/it) is chosen and the research data is coded
at word, line, sentence and/or paragraph level for concepts, patterns, regularities, systems,
and themes. Clearly, this is a highly interpretative process and ‘rigour’ can only be taken to
mean that all data is treated in the same way (aided by computer entry) and differing
interpretations are recognised. Data analysis is not descriptive – ‘x said this’ – and should go
beyond the face value of the data in order to seek meaning and context for an understanding
and more complete answer to the research question.
In Voices of the Poor and Growth is Good for the Poor, data analysis and findings were as
follows: in Growth is Good for the Poor the data analysis involved the interpretation of
econometric output; in contrast the data analysis in Voices of the Poor was based on
qualitative analysis of transcripts of interviews and focus group discussions (see Table 9).

Table 9. Data analysis in Growth is Good for the Poor and Voices of the Poor.

Growth is Good for the Poor Voices of the Poor

Data analysis Interpretation of econometric Literature review and qualitative
output analysis of transcripts of interviews
and focus group discussions

In Growth is Good for the Poor, the analysis has been criticised because the regressions were
statistically insignificant with the exception of the one-for-one growth-poverty finding
(Weisbrot et al., 2001) and the study was misleading because few countries were near the

21
one-for-one average (Nye et al., 2002: 12). In Voices of the Poor, the study explicitly
recognised the likely criticisms over bias and subjectivity in analysis (Narayan et al. 2002:
19-20). This brings the debate through a full circle to epistemology in DS. Is DS a search for
one ‘reality’ as Dollar and Kraay suggest or does the heterogeneity of experience make
average values meaningless?

3. CONCLUDING DISCUSSION

Epistemology and methodology in DS are a special case. In epistemology, DS can range from
a natural science-like search for one ‘truth’ to a subjective, social constructivist search for
many ‘truths’ or both if the tenets of realism are accepted and the researcher is conscience of
their bias. In methodology, due to the nature of the subject matter, the simple transposition of
generic social science methodology and methods onto developing countries is often
problematic. Many of the generic concerns in social science methods are more sensitive in
development research. For example, concerns over the validity of research, the extent to
which the results are representative, the reliability of data, and the subjectivity and
interpretation of results are particularly problematical in developing countries, and ultimately
the replicability of the research may not be certain.
In the light of this, what does ‘rigour’ in research mean for DS? From the identification of
the problem and research question specification, through the research design and data
collection, analysis and findings several themes predominate in DS. These include firstly,
logical processes linking the research from start to finish with a coherent thread throughout.
Each stage informs the next. Secondly, what constitutes acceptable and unacceptable
bias/subjectivity. Having values is acceptable. Changing research to gain findings to suit
those values is unethical. Finally, there exists a set of techniques and methods. The strengths
and weaknesses are well known. Thus corroboration through mixing methods as appropriate
and a reflective practitioner approach to the imperfections of research and choices made are
signs of a ‘rigorous’ approach. Ultimately, if other researchers can see when and why
appropriate choices were made and of these choices can be intellectually defended, then in an
imperfect world the research journey will at least have been transparent and ‘rigour’ attended
to.

22
REFERENCES

Amann A, Aslanidis N, Nixson F, Walters B. 2002. Economic growth and poverty alleviation: A
reconsideration of Dollar and Kraay. Paper prepared for the Annual Conference of the Development Studies
Association. University of Greenwich, London, 9 November.
Berger P. Luckmann T. 1966. The Social Construction of Reality. Doubleday: New York.
Burns R. 2000. Introduction to Research Methods. Sage: London.
Cameron J. 2003 Background Notes: Foundational ideas of epistemology, development studies and science. Paper
prepared for Development Studies Association (DSA) Postgraduate Workshop. Overseas Development Institute.
London, 20-22 October.
Cameron J. 2004. The Epistemology of Post-Modernism and Post-Structuralism. Paper prepared for DSA
Annual Conference. Church House, London, 6 November.
Denscombe M. 2003. The Good Research Guide (2nd ed). Open University Press: Maidenhead.
De Vaus D. 2001. Research Design in Social Research. Sage: London.
Dollar D, Kraay A. 2002. Growth is Good for the Poor. Journal of Economic Growth, 7: 195-225.
Dollar D, Kraay A. 2000. Growth is Good for the Poor. World Bank Policy Research Working Paper Number
2587. World Bank: Washington, DC.
Edusah S. 1999. The Socio-Economic Significance of Rural Small-Scale Industries in Ghana. PhD Thesis,
University of Bradford, Development and Project Planning Centre.
Fink A, Kosecoff J. 1998. How to Conduct Surveys: A Step-by-Step Guide. Sage: London.
Galtung J. 1967. Theory and methods in social research. Allen and Unwin: London.
Giere R, Richardson A. (eds.) 1997. Origins of Logical Empiricism. University of Minnesota Press:
Minneapolis.
Gilbert G. 2001. Researching Social Life. Sage: London.
Hacking I. 1999. The Social Construction of What? Harvard University Press: Harvard.
Harriss J. 2002. The Case for Cross-Disciplinary Approaches in International Development. World
Development. 30(12): 487-496.
Hulme D. 2000. Impact Assessment Methodologies for Microfinance: Theory, Experience and Better Practice.
World Development. 28 (1):79-98.
Hunt D. 1989. Theories of economic development. Harvester Wheatsheaf: New York.
Jackson C. 2002. Disciplining Gender? World Development. 30(12): 497-509.
Kanbur R.2002. Economics, Social Science and Development. World Development. 30(12): 477-486.
Kiely R. 1999. The last refuge of the noble savage? A critical assessment of postdevelopment theory. The
European Journal of Development Research. 11(1): 30-55.
Laws S, Harper C, Marcus R. 2003. Research for Development: A Practical Guide. Sage: London.
Loup J, Naudet D. 2000. The state of human development data and statistical capacity building in developing
countries. Human Development Report Office Occasional Papers. March. UNDP: Geneva.
Loxley J. 2004 What is distinctive about International Development Studies? Canadian Journal of Development
Studies. 25(1): 25-38.

23
Maxwell S. 2003. Debate: Development research in Europe: Towards an (all)-star alliance? European Journal
of Development Research. 15(1): 194-198.
May T. 2001. Social Research: Issues, Methods and Process. Open University Press: Buckingham.
Mensah J. 1999. Small-Scale Industries in Ghana's Economic Development: Comparative Economic Efficiency of
Rural and Urban Small-Scale Industries in the Central Region. PhD Thesis, University of Bradford. Development
and Project Planning Centre.
Mikkelsen B. 2000. Methods for Development Work and Research: A Guide for Practitioners. Sage: London.
Molteberg E, Bergstrom C. 2002a. Our common discourse: Diversity and paradigms in Development Studies.
Centre for International Environment and Development Studies, Agricultural University of Norway
(NORAGRIC) Working Paper Number 20. NORAGRIC: Ås, Norway.
Molteberg E, Bergstrom C. 2002b. Our common discourse: Diversity and paradigms in Development Studies.
Centre for International Environment and Development Studies, Agricultural University of Norway
(NORAGRIC) Working Paper Number 21. NORAGRIC: Ås, Norway.
Morton J, Martin A. 2004. Natural Resources Research and Development Studies. Paper prepared for DSA
Annual Conference. Church House, London, 6 November.
Narayan N, Patel R, Schafft K, Rachemacher A, Koch-Schulte S. 2002. Voices of the Poor: Can anyone hear
us? World Bank: Washington, DC.
Nye H, Reddy S, Watkins K. 2002. Dollar and Kraay on ‘Trade, growth and poverty: A critique’. Mimeograph,
Columbia University.
Olsen, W. 2003. Pluralism, Tenancy and Poverty: Cultivating Open-mindedness in Poverty Studies. Paper
presented to the Development Studies Association Annual Conference, University of Strathclyde.
Oxfam 2002. Rigged Rules and Double Standards: Trade, Globalisation and the Fight Against Poverty. Oxfam
Publishing: Oxford.
Parfitt T. 2002. The End of Development? Modernity, Post-Modernity and Development. Pluto: London.
Plato 360 BCE, Theaetetus (translated by Benjamin Jowett). Downloadable from http://classics.mit.edu/
Plato/theatu.html (accessed 7th November 2004).
Popper K. 1968. The Logic of Scientific Discovery (2nd ed): Hutchinson, London.
Punch K. 1998 Introduction to Social Research: Quantitative and Qualitative Approaches: Sage: London.
QCA (Qualifications and Curriculum Authority) 2003. A Level Performance Descriptions: World Development
– downloadable from http://www.qca.org.uk/ages14-19/ downloads/wor_pd.pdf – accessed 5th November 2004.
Ravallion M. 1998. Poverty lines in theory and practice. World Bank Working Paper No. 133. World Bank:
Washington, DC.
Reddy S, Pogge T. 2002. How Not to Count the Poor. Department of Economics Working Paper. Colombia
University: New York.
Rodrik, D. 2000. Comments on Trade, Growth and Poverty by D. Dollar and A. Kraay’. Downloaded from
http://ksghome.harvard.edu/~.drodrik.academic.ksg
Scheyvens R, Storey D. (eds.) 2003. Development Fieldwork: A Practical Guide. Sage: London.
Schuurman F. (ed.) 1993. Beyond the Impasse: New Directions in Development Theory. Zed Books: London.
Shaw T. 2004. International Development Studies in the Era of Globalization… and Unilateralism. Canadian
Journal of Development Studies. 25(1): 17-24.

24
Suryahadi A, Sumarto S. 2001. The Chronic Poor, the Transient Poor and the Vulnerable in Indonesia, Before
and After the Crisis. Social Monitoring and Early and Early Response Unit (SMERU) Working Paper. SMERU:
Jakarta.
Sutcliffe B. 1999. The place of development theories in theories of imperialism and globalisation. In Munck R,
O’Hearn D. (eds.) (1999) Critical Development Theory: Contributions to a New Paradigm. Zed Books: London.
Tribe M. 2003. Primary and secondary research: What do we mean by rigour? Paper prepared for ESRC
Development Studies Association Postgraduate Workshop. Overseas Development Institute. London, October
20-22.
Tribe M, Sumner A. 2004. The Nature of Development Studies. Paper prepared for DSA Annual Conference,
Church House, London, 6 November.
Weisbrot M, Baker D, Naiman R, Neta G. 2001. ‘Growth may be good for the poor – but are the IMF and World
Bank policies good for growth?’ Centre for Economic Policy and Research Working Paper: Washington, DC.
White H. 2002. Combining Quantitative and Qualitative Approaches in Poverty Analysis. World Development,
2002, 30(12): 511-522.
World Bank. 2000. World Development Report 2000-2001. Oxford University Press, New York.

25
Appendix 1: What is ‘development theory’?

From the 1950s to the late 1980s there was a stand-off between capitalist and non-capitalist
development theory. On the one hand, capitalist modernisation theory (Rostow’s anti-
communist manifesto), was very influential in the 1950s/1960s. This was partly a by-product
of McCarthyism and was also a reaction to fast industrialisation in the USSR and to concern
that newly independent developing countries would follow a non-capitalist path. In contrast,
the anti-capitalist/anti-imperialist Latin American neo-Marxism in the form of
underdevelopment theory – structuralism and dependency theory – was very influential in the
late 1960s/1970s and was instrumental in the introduction of the developmental state and
import substitution industrialisation in many developing countries following independence
from former colonial powers.
In the late 1980s Development Studies theory entered an academic impasse from which it
has not been fully resuscitated. This was a stalemate between modernisation and
underdevelopment theories and a result of deficiencies in those frameworks. Theories were
criticised for their overriding belief in a linear, common path to development through
economic growth and industrialisation. This belief became increasingly challenged as the
homogeneity of the ‘Third World’ fragmented; as the meaning of development broadened
and as the environmental impacts of industrialisation became more evident (Schuurman
1993; Sutcliffe 1999). Development theory also faced sustained attack from the post-modern
critique of theory formation in social science (see Parfitt, 2002; Schuurman, 1993), for an
over emphasis on the macro to the neglect of the micro (especially intra-household) and
meso, and for the lack of empiricism and testability/refutability. Finally, the loss of the
socialist paradigm at the end of the Cold War ended the link between development theory and
practice. The impasse in development theory coincided with the resurgence of neo-liberalism
in what became known as the ‘Washington Consensus’ consisting of reconstituted
components of the ‘modernisation project’ synthesised with economic liberalism.
There is now an array of conceptual frameworks utilised in Development Studies. A
conceptual framework does not seek to understand and explain everything in the way an
overriding theory does, or necessarily seek prescriptions like a theory, but rather, as Hulme
(2000:81) argues, to specify what is to be assessed or measured and at what level it is to be
assessed and with some similarity to a theory, to model an impact chain. The chain explains
some or all of the contributory factors to that phenomena under analysis, but does not provide
an overriding meta-narrative to explain everything (see for discussion on diversity of

26
theories, Hunt, 1989). Post-impasse contributions have focused on broader definitions of
development and in particular emancipation. Numerous conceptual frameworks have
emerged in recent years including Sustainable Livelihoods, rights approaches, Gender and
Development and participatory approaches. The only attempt at theory has been the anti-
theory of Post-development – an application of post-modernism/post-structuralism to
development that rejects outright ‘development’ as an imperialist discourse (see for a critical
discussion, Kiely, 1999).

27