1
Methodological or epistemological issues in social research
Achin Chakraborty
1 Introduction
The courses in research methodology seem to have been driven by the widely held notion that carefully drawn methodological principles would tell us how to do research scientifically (or social-scientifically?). And once they are drawn, the next obvious step would be to appraise an actual piece of research or a research programme in terms of those principles. In other words, the logical sequence turns out to be from a set of prescriptive principles to the practice that is supposed to follow those principles. In economics, for example, philosophers of science were believed to hold the key to how to do âeconomic scienceâ, even though several groups outside mainstream economics (e.g. Marxists, Austrians, Institutionalists) have had their shared methodological approaches. What has been common amongst the practitioners of economic research â both within and outside the mainstream â is that both sides have viewed methodology as offering a set of prescriptions on what constitutes legitimate practice. In this view, the common concern of methodological discussions is essentially normative and is based on philosophersâ attempt to justify knowledge claims.
In this chapter, we take the opposite route. We argue that there is enough evidence to show that actual practice of research in social sciences is too rich in diversity and innovativeness to be disciplined by a few prescriptive norms. Therefore, in section 2 we begin with the invocation of the diversity of practices and a rough classification of different types of research inquiries, each of which is ostensibly driven by a different motivation. Research inquiries are not always explanatory â or as economists tend to suggest, predictive. Apart from explanation and prediction there are several other motivations that drive social research. A major area, for example, deals with the normative issues involved in assessing states of affairs or changes therein. For example, an issue like how development of a country or a region is to be assessed is evaluative in nature. Of course, the brief account of different types of social research presented here is far from exhaustive. In section 3 we discuss how an explanatory kind of research question is dealt with within the positivist-empiricist framework. In particular, the respective roles of theory (or explanatory framework), data and method are discussed. In section 4, we raise a few issues about normative-evaluative kind of research. In section 5, we briefly discuss the post-positivist approaches in social research, and in section 6 we conclude.
2 From practice to methodology
The commonplace view about social research is overwhelmingly explanation-oriented, where the central question is âwhyâ. Why is the labour force participation rate of women low in India? Why are some states better at human development than others? Why have so many farmers committed suicide in India in the recent past? Answers to these questions take a causal form, even though the method usually deployed to establish a causal explanation can accomplish the job only imperfectly. Nevertheless, most policy discussions are based on some understanding of the causes and their effects on various outcomes. In other words, the essential nature of inquiry here is explanatory. Inquiries of this kind end up indicating or âestablishingâ some causal connections between choices or actions of agents (individuals, groups, governments, corporations etc.) and outcomes. However, the self-conscious practitioners of statistical or econometric techniques know rather well that at best their techniques establish some association between variables, rather than a causal connection. One must take a big leap of faith to claim an associational observation between, say, x and y, as a causal one, even though certain econometric techniques, such as the Granger causality test, claim to establish causal connections between variables. Thus, we might self-consciously seek to establish some association between entities, in which case the nature of inquiry would be associational or relational, rather than explanatory.
Besides explanation and finding association, one can identify several other motivations which drive research inquiries. When a study is designed primarily to describe what is going on or what exists, without entering into the analysis of underlying relationships or causal connections that are not so apparent, it is descriptive. A question such as âhow has GDP of India grown in the post-reform periodâ falls in this category. To answer this question one has to describe the pattern of growth in Indiaâs GDP between, say, 1991 and the present. However, there is no such thing as âpure descriptionâ, as description involves conscious methodological choice.1 As in this apparently simple question, one has to decide on whether the average annual rate of growth or the trend rate of growth should be calculated, whether the period should be divided into sub-periods and the average or the trend growth rates in the sub-periods should be noted and so on.
Different underlying motivations seem to dominate different disciplines. In economics, for instance, prediction is considered to be the most important motivation behind theoretical and empirical inquiry. In mainstream economics, the standard methodological route is to set up a model of behaviour of agents (individuals, firms etc.). Starting from a set of axioms about behaviour of the agents, conclusions are derived using deductive mathematical logic. The methodological approach is therefore called hypothetico-deductive. Such models based on deductive logic are expected to predict future outcome. This dominant view was made explicit by Milton Friedman in his widely known paper âThe Methodology of Positive Economicsâ (Friedman [1953]). Friedman argued that the assumptions made by economists while modelling individual behaviour should be judged âby seeing whether the theory works, which means whether it yields sufficiently accurate predictionsâ, not by the ârealismâ of the assumptions. Amartya Sen, however, holds a different view:
Prediction is not the only exercise with which economics is concerned. Prescription has always been one of the major activities in economics, and it is natural that this should have been the case. Even the origin of the subject of political economy, of which economics is the modern version, was clearly related to the need for advice on what is to be done on economic matters. Any prescriptive activity must, of course, go well beyond pure prediction, because no prescription can be made without evaluation and an assessment of the good and the bad.
(Sen [1986, p. 3])
Thus, âevaluation and an assessment of the good and the badâ gives rise to yet another altogether different kind of inquiry, which is evaluative. For an evaluative inquiry one applies certain normative criteria to judge states of affairs. For example, a question such as, âIs gender inequality more in country A than in country B?â apparently falls in the descriptive category. But on closer scrutiny, it becomes clear that there is no obvious way of assessing gender inequality with a comparative perspective. Even if one restricts oneself to this question, ignoring such related questions as why gender inequality is more in one country than in another, it turns out to be non-trivial as explicit value judgements with moral philosophic underpinnings are deeply involved. Amartya Sen often makes a distinction between evaluative2 exercises and descriptive-analytic or predictive-prescriptive exercises, as in the earlier quotation, and emphatically points out that the motivation behind the evaluative type of inquiry is no less important than that behind others. The entire theoretical literature on measurement of inequality, poverty and human development falls in this category.
All these types of research inquiries briefly described here roughly fall in the paradigm which can be called positivist. In the next section we elaborate on the notion of paradigm and the epistemology of positivism.
3 Positivist-empiricist practice
There is no simple formula to establish any connection between specific âcausesâ and âeffectsâ. Three basic ingredients of social research are (1) some ideas about how things are or how change takes place, (2) data or observations on âfactsâ and (3) methods that integrate ideas and observations. By method we mean a set of tools or techniques informed by an approach which is applied in a research inquiry. But methodology is concerned with the framework within which particular methods are appraised. In other words, methodology deals with the broader question of âhow we know what we knowâ and is somewhat close in meaning to what we understand as epistemology. Ideas are obtained from various theories. They may often look like commonsense. But if they are part of a theoretical framework one can expect logical coherence in the ideas, which commonsense does not guarantee.
What is theory? Before we come up with an imprecise answer to this question, it would be helpful if we accept that theory can be defined only within a paradigm. Roughly speaking, a paradigm is a combination of a set of underlying beliefs about the ways things are and specific ways of inquiring about how things are, how they change, how they are connected with or influenced by each other and so on. In other words, a paradigm can be identified with specific ontological and epistemological positions. For many of us who work in what is loosely called âdevelopment researchâ, a kind of positivism seems to be the underlying paradigm. In this version of positivism the core belief is that reality is out there and that by gathering âfactsâ it is possible to find out what is happening in reality. The researcher is assumed to stand apart from the observed and produce objective knowledge. How does he or she go about it? First, the researcher identifies separate aspects of reality and expresses them as âvariablesâ. Then he or she goes on examining the relationships between variables. This involves both observation and reasoning based on arguments acceptable within the paradigm. Within the positivist paradigm, a theory is expected to answer our âhowâ and âwhyâ questions in the most generalized way with a coherent logical structure. Generalizability is at the core of theoretical statements.
Someone with an empirical bent of mind and relying less on theory for illumination often tends to say âfacts speak for themselvesâ. As a matter of fact, facts hardly speak for themselves. One has to sort out relevant from irrelevant facts at the outset. Without some prior idea about the nature of the phenomena, without some propositions, assumptions etc., there is no way this can be meaningfully done, according to a positivist. Deciding that observation X or Y is relevant marks the start of a theory. In this paradigm, theory means a logically valid chain of reasoning starting from certain premises called postulates. Postulates are taken as axiomatically given and contain certain terms that are representative of persons, organizations, things, actions, states etc. found in the world of experience. A meaningful analysis presupposes that the terms are unambiguously defined.
In this positivist-empiricist paradigm hypothesis testing seems to take the pride of place. It is a commonly held view that any proposed research in social science must specify at the outset the hypotheses to be tested. Admittedly, certain types of social research do require the use of hypotheses. They can be useful in helping to find answers to âwhyâ questions and therefore are developed at the outset to set the direction. However, precise specification of the hypotheses is neither necessary nor appropriate in many cases. In particular, when explanation is expected to come out in the form of a complex web of interconnections and mutual influences, a cut-and-dried kind of hypothesis testing may not give a better insight into a phenomenon. Hypotheses should ideally be derived from a theory of some kind. Hypotheses that are simply based on common sense or intuition without making any reference to the existing state of knowledge rarely make significant contributions to the development of knowledge.
Most mainstream economists believe that their methodology is positivist. The philosopher who has had the greatest influence on the methodology of economics is Karl Popper, as evident from the frequent invocation of Popper by economic methodologists such as Mark Blaug (1992). Popperâs philosophy even influenced a major introductory textbook â Richard Lipseyâs An Introduction to Positive Economics. Popperâs philosophy of âscientific knowledgeâ is concerned with what he calls âthe problem of demarcationâ, i.e. the problem of distinguishing science from non-science. Popper introduces falsification as the criterion to be applied for demarcation. A statement is in principle falsifiable if it is logically inconsistent with some finite set of true or false observation reports. Popper himself gives an example of a scientific statement: âAll swans are whiteâ. This is a falsifiable statement since the observation of a non-white swan would establish its falsity. There is an asymmetry between verifiability and falsifiability. A universal statement concerning an unbound domain, such as âall swans are whiteâ, may be falsifiable but not verifiable. For example, the observation âthis swan is blackâ falsifies the statement âall swans are whiteâ. To verify the statement, we need to observe all possible swans, but the set of all swans is unbounded as it includes, inter alia, swans yet to be born. In other words, it is not possible to verify any truly universal statement, but one can falsify it or verify its negation (Hausman, 1992].
Ironically, economists do not seem to practice what they preach. Hausman criticizes âthe methodological schizophrenia that is characteristic of contemporary economics, whereby methodological doctrine and practice regularly contradict one anotherâ. We take a different standpoint here, one which fundamentally disagrees with the view that economists should stick to falsificationism as the only criterion to assess knowledge claims in economics. In what follows, we present a glimpse of the variety of theoretical ideas and the corresponding methodological approaches that scholars have actually adopted in social sciences in general and in certain brunches of economics in particular.
4 Explanatory framework and varieties of theory: structure versus agency
A broad classification of theoretical approaches in social sciences could be made in terms of the basic unit of analysis. Traditionally, the dominant view in social science happened to be that of understanding the functioning of the social system. From Adam Smithâs An Inquiry into the Nature and Causes of the Wealth of Nations to Max Weberâs The Protestant Ethic and the Spirit of Capitalism, most of the classics in social sciences focused on some kind of social (or economic) system. However, much of contemporary social research focuses on explaining individual behaviour. With the development of quantitative methods of research, dependence on individual-level data has increased significantly. If one still feels that the functioning of the system should remain the central problem for research inquiry, how does one go about explaining the system? From both ontological and epistemological points of view there can be two broad ways of explaining the system â systemic (or structuralist) and agency-based3.
A structuralist mode of explanation generally rejects the view that the social, economic or political structure can be explained entirely as the aggregate of the actions of individual agents. The most well-known statem...
