The main objective of this paper is to make a critical assessment from the perspective of Third World studies of the methodology used for data collection and analysis in comparative macro-social science research within the framework of broad ontological and epistemological debates in the social sciences in general, and sociology in particular. The paper begins with a general discussion of the empiricist method of scientific enquiry and explanation, which in my assessment is the hegemonic approach to knowledge claims and validation in contemporary social science. The general strategy and foundation of knowledge claims of the empiricist method is briefly reviewed, with the basic assumptions of the method highlighted. That is followed by an analysis of the efforts made to ameliorate the problems associated with the application of statistical techniques of explanation and reasoning in comparative macro-social science research. I then discuss an alternative epistemological perspective to empiricism. I argue that the use of narrative as a method of producing and validating knowledge claims in comparative-historical studies in the social sciences could be theoretically grounded on a social realist ontological perspective, which underscores how time, space, and the complexity of social reality limits the validity of universal law-like generalizations. I review how macro-international processes are integrated into historical-comparative research as reflected in the works of Theda Skocpol, Charles Tilly and Immanuel Wallerstein. The limitations of the research strategies used by these scholars were identified by Philip McMichael who suggests an alternative method of incorporated comparison on a world historical scale. The paper concludes by discussing methodological lessons and insights from the discussion and analysis that are relevant for Third World studies and research.
It is pertinent to justify why ontological and epistemological debates in the social sciences are critically discussed in this paper. Critical reflection leads me to conclude that, in general, many courses in data collection techniques and analysis are silent on the deeper ontological and epistemological assumptions underpinning the use of such techniques and on their implications for the validity of knowledge claims produced by the use of such techniques. One consequence of this pedagogic orientation in methodology courses is that people use such techniques without reflecting on the ontological and epistemological assumptions and limitations underlying the techniques. To avoid the pitfalls characterizing other people's work, I argue that any technique used for data collection and analysis is based on certain fundamental ontological and epistemological assumptions. Unless one consciously acknowledges this relationship, one easily falls into the trap of treating techniques of data collection and analysis as existing sui generis (i.e., having an independent status of their own). Awareness of the epistemological assumptions of the techniques one uses results in both a recognition of the limitations inherently imposed on the validity of our knowledge claims and the need to be modest in making such claims.
By making a critical review of empiricist ontology, epistemology, and data collection method and analysis, I am not suggesting that such a method does not generate useful and valid knowledge. Rather, the aim is to strongly assert that when data collection and analysis techniques are treated as existing sui generis, they may result in wild knowledge claims that, on critical evaluation, cannot be thoroughly justified. Similarly, when one uses the narrative technique of data collection and analysis without being conscious of its ontological and epistemological assumptions and limitations, it can result in claims that cannot be justified on close scrutiny. I begin by examining the general strategy and foundations of empiricist knowledge claims.
EMPIRICIST METHOD OF SCIENTIFIC ENQUIRY AND EXPLANATION: GENERAL STRATEGY AND FOUNDATIONS OF KNOWLEDGE CLAIMS
Empiricist scholars assert that our claims to knowledge are constrained by our ability to justify the claims by experience only. It is wrong, according to them, for anyone to make statements on the existence of any phenomenon that cannot be justified by experience (a critique of social realism). In view of this, it is baseless, they argue, to assume that there are certain hidden structures, social or natural forces, instincts, or inherent contradictions because they cannot be observed, and therefore are outside the purview of the empiricist method of justification. In other words, any variable or concept that will be used in this version of scientific inquiry needs to be transformed from something abstract to something empirically observable and verifiable. (1) Our general beliefs could be considered knowledge in this perspective only in so far as the beliefs originate from systematic observations and can be verified through experience. The works of empiricists attacked all areas of knowledge based on claims that were never grounded on particular observations and could not be verified through experience. In particular, empiricists tried to undermine ethics, theology, and metaphysics, etc. as branches of knowledge that were based on empirically unjustifiable claims. (2)
Empiricists supplement their epistemology of sense and perception with the principle of induction, which makes inferences possible. Some empiricist scholars argue also that induction is necessary in scientific enquiry in order to extract general principles and relationships from many idiosyncratic and "noisy" empirical events that are observed. (3) Basic empiricism uses perception, frequencies, correlations, and probabilities to generalize, and then generalizations are projected forward and backward for the purpose of prediction and explanation, respectively. In my view, this leaves out something very important about the individual, event, or actor being explained. Such prediction and explanation tend to be based on macro-aggregate summary data for groups to which the individual belongs. But such probabilities tell us little about the meaning the individual makes of the world around his/her or the interconnected nature of how events unfolded. (4) In this way, it is possible for our probabilities to be right; but on closer examination, the meaning the individual consciously makes of the world around her may accidentally fit the probabilities or may not even fit at all. Indeed, the individual could have a distinctly different explanation for his/her behavior or action. (5) Similarly, such macro-aggregate probabilistic explanation and prediction tend to theoretically remove the sense of agency from the individual or group in deciding whether to behave one way or the other, within the opportunity structure available, given existing social structural constraints. Sometimes this kind of explanation discounts the actor's interpretation, motivation, and explanation. (6) We are furthermore not sure of the mechanisms by which the individual exercises his/her agency; neither do we know why the agency was exercised. Without knowledge of the aforementioned considerations, even with our predictions and explanation (which tell us only about regularities at the aggregate level), there is a certain portion of the individual's behavior that we cannot account for. Indeed, at one level one may argue that explanation generated through macroaggregate data on individual or group behavior raises serious ethical and methodological problems because by disregarding the actor's explanation, we are denying the individual's or group's ordinary right of being a responsible moral agent. (7)
The major deficiency of many empiricist explanations is that they do not provide good micro-foundations and analyses of the mechanism by which the causal regularities they describe emerge. Where this is done, it is not the preoccupation of the research, but just an attempt to contextualize the already established statistical regularities. Unfortunately, a set of statistical regularities can theoretically be well explained by several different mechanisms. (8) This means that the statistical regularities established are theoretically overdetermined.
If empiricists want to avoid this problem, they have to develop a theory that while using macro-aggregate data on statistical regularities in human behavior, also integrates micro-foundations of such explanation in the form of a story of causal mechanism. With such integration, the empirical regularities identified by empiricists via statistical descriptions will not necessarily entail the assumption that social actors lack moral bases for decisions and choices. Doing this will allow empiricist explanations the flexibility to articulate the voices of social actors or the nuances of social processes. Otherwise what we achieve in predictability and explanation does not give us enough confidence about the individual's or group's sense of agency. Without this knowledge, we do not have a confident explanation because there is the possibility that time, space, and historical and cultural contexts can create variations that our statistical regularities and generalizations cannot capture. (9)
One can indeed raise some critical concern about the use of parsimonious explanation models, which is vehemently stressed in empiricist methods of enquiry. For example, critical reflection on the methodology of Theda Skocpol's States and Social Revolutions highlights some of the deficiencies associated with the preoccupation by empiricists to develop a parsimonious model of explanation. (10) She uses empiricist methodology (induction) in an interpretive-qualitative format to make a comparative historical study of national revolutions. To come up with a parsimonious model explaining the French, Russian, and Chinese revolutions across a time period of over a century (1789-1940s), she had to impose some assumptions on history and social reality and hold them constant. Michael Burowoy notes the following assumptions that Skocpol had to make in order to make her study epistemologically plausible:
a) All three revolutions belong to the same specie or class of elements. This perhaps is because she defines revolutions from their appearance (i.e., formal characteristics) and not essence. Yet several social phenomena can appear the same but actually have different essences.
b) The same "causal factors operate in all three revolutions." She believes there is indeed one theory of social revolutions. (11)
In making these assumptions, Skocpol ossified complex and diverse historical processes that stretched over a century by assuming that history and time had the same meaning and were therefore constant across the whole period covering her study. Like many empiricists, she allowed faithfulness to formal methodological requirements to override the need for thorough theorizing of time and the temporality of how historical events unfold. Furthermore, in treating the three revolutions as independent cases, she ignored the effect of earlier revolutions on how the people in the later revolution subjectively perceived the need for revolutions. (12)
Skocpol wanted to generalize and therefore developed a universalizing and encompassing theory of social revolutions (13). Yet to be able to generalize, she had to transform the historical processes and situations she studied into variables, and by doing so, she unwittingly disaggregated the cases she studied from their complex historical and cultural contexts. (14) She also ignored the meaning and trajectory that time confers on events, social processes, their outcomes, and consequences. (15)
From the foregoing, we can assert that the variable-oriented approach of comparative studies ignores the micro-foundations of knowledge claims made at the general and macro level of analysis, thereby producing a social science that is detached from concrete social realities at the grassroots level. In effect, empiricists preoccupy themselves with lateral relations and generalizations because of the assumptions they make about the social world. I therefore proceed to highlight some of their specific assumptions about the social world that are in many respects problematic and often aren't explicitly acknowledged.
ASSUMPTIONS ABOUT THE SOCIAL WORLD MADE BY EMPIRICIST SCHOLARS IN SOCIAL SCIENCE RESEARCH
Empiricists assume that the social world is constituted by permanent entities which have varying attributes. This assumption, which makes statistical analysis possible, in practice, fails to conceptually and empirically theorize how the supposedly permanent entities produce the attributes they have. Isn't there something specific about the nature of the permanent entities that enable them to exude certain attributes, which makes it possible for empiricist social scientists to theoretically assume that the appearance of such entities is inevitably associated with certain effects or results? Empiricists do not address this issue adequately, yet their statistical analysis is unconsciously built on such a hidden assumption. (16)
Empiricists make the assumption that the effect of the operation of a variable on a case or situation does not have any constraining or modifying effect on the time or place when another variable operates. Andrew Abbott calls this assumption "case-wise independence assumption." (17) But we know that this is an untenable assumption in empirical reality. As Barrington Moore eloquently argues, the French Revolution has had temporal and spatial effects on subsequent revolutions in Europe and Asia. (18) Any model of explanation that makes the case-wise independence assumption in its analysis is, in a significant way, making a problematic assumption about social reality and history.
The empiricists also assume that variables used in a statistical model of explanation in a particular study have a single causal effect or relevance. In other words, its meaning within the context of a …