Paths to Democracy

Revolution and Totalitarianism

Rosemary H. T. O'Kane

• Publisher: Routledge
• Year: 2004
• ISBN:
  9781134378807
• Edition: 1
• Topic: Politics & International Relations
• Subtopic: Politics
• Index: Politics & International Relations

How and why countries become democracies remain intriguing questions. This innovative volume provides a theoretically informed comparative investigation of the links between revolutions, totalitarianism and democracy. It will appeal to those interested in the relationship between history and democracy and the implications for the understanding of democracy today.

Part I
Democracy

Method and theses

1 On method

Comparative historical analysis and politics

To move beyond description, the narrative of the path taken by a country to democracy, and to engage with general questions about democracy it is necessary to make comparisons. As C. Wright Mills (1970: 163) has it, ‘Comparisons are required in order to understand what may be the essential conditions of whatever we are trying to understand.’ As Sartori (1994a: 15) explains, ‘comparisons control – they control (verify or falsify) whether generalizations hold across the cases to which they apply’. Comparing for similarity and difference against controls is the comparative method and its purpose is to judge significance. For a subject such as democracy, how precise that judgement can be is limited, in practice, by a number of problems. The most widely recognized are those neatly put by Lijphart (1971: 685): ‘many variables, small number of cases’. The chances of a study of democracy escaping these problems are slim. It is, clearly, reasonable to expect paths to democracy to be complex and though in recent times democracies have grown greatly in number, established democracies remain few.

In addition to the problems of few cases and large numbers of variables the study of democracy is also constrained by conceptual logic. The extent to which general explanations can be claimed is limited both by the number of things to which the concept can apply (for example, the list of democracies) and by the capacity of the concept to drop properties and be defined not only in terms of what it is but also in terms of what it is not.¹ As Collier and Levitsky’s (1997) consideration of ‘democracy with adjectives’ has shown, in practice the properties of democracy, rather than being pared down, tend to be heaped up to incorporate more identifying characteristics. The consequent ‘conceptual stretching’ (Sartori 1970) thereby allows in cases that are not democracy. Employed in statistical analysis in search of general explanations stretched concepts run the very real risk of producing messy and even misleading research.² It follows, therefore, that for the study of democracy it is necessary to move down Sartori’s ‘ladder of abstraction’, away from global explanations towards more ‘country-by-country’ studies seeking descriptions of similarities and differences and comparing cases within an area (Sartori 1970: 1042).

In country-by-country studies the comparative method is still employed, similarities and differences are examined against controls in search of regularities and investigation of hypotheses. Country-by-country studies have the benefit that the concept of democracy employed in them is not required to have the scientific precision of an empirical universal concept. The loss is that lessons produced cannot make claims to be general explanations, theories. They remain lessons about situations favourable or unfavourable to democracy, concerned with necessary but not sufficient conditions. So long, however, as the claim for general theory is resisted, there is much to be said for studying small numbers of cases and particularly so where a comparative historical approach is adopted.

Comparative historical analysis

Skocpol (1984) has delineated three comparative historical approaches. The first is the application of a general model to one or more historical cases; the second is the use of concepts to interpret history, the ‘interpretive’ approach; the third is the analysis of causal regularities in history, the search for a ‘pattern’ of history.³ The second approach, the ‘interpretive’ approach, constructs a concept to aid understanding, in the sense of how things work, independently or together, to produce effects in history. Though concerned to understand occurrences in history this approach is not directed at highlighting generalities but at bringing understanding to particularities through comparison. Both the first and the third approaches seek causal explanations. The first approach is deductive, hypotheses to be tested are deduced from a theory or model and applied to a historical case or cases. Within this category comes Lakatos’s (1978) methodology of scientific research programme. The third approach is inductive; the explanation is generated from examination of the cases and alternative hypotheses are tested arising from consideration of those cases. In contrast to the interpretive approach, both the deductive and the inductive methods are concerned with sequences, that is they seek out conditions, events, situations which, logically, precede the event that they are intended to explain.

As a method of comparative historical analysis, Skocpol argues for the superiority of the third approach, the analysis of causal regularities. This inductive method, employed by Skocpol (1979) in States and Social Revolutions, is also the method employed by Moore (1969) in Social Origins of Dictatorship and Democracy. In arguing for the inductive method, Skocpol (1984: 365–8) stresses the problems of the other two methods: the interpretive method frustratingly stops short of investigating causal regularities; the deductive approach requires the availability of a model that has not been generated from the cases to which it is to be applied. If available, the application of such a pre-existing model may lead the researcher to overlook important contrary evidence and to squash evidence into shape to make it fit. With concentration on a few cases, cases may also be selected for their fit while other less supportive cases may be ignored. Skocpol’s broad concern is that restriction to one model or theory acts like a straitjacket around the research, preventing the study of the cases opening up new ideas.

In elaboration of her preferred method, the search for causal regularities in history, Skocpol (1984) argues as follows:

Neither the logic of a single overarching model nor the meaningful exploration of the complex particularities of each singular time and place takes priority. Instead, the investigator assumes that causal regularities – at least regularities of limited scope – may be found in history. He or she moves back and forth between aspects of historical cases and alternative hypotheses that may help to account for those regularities.

Ideas about causal regularities may come from two or more preexisting theories that are brought into confrontation with the historical evidence. Or they may be generated more inductively from the discovery of what Arthur Stinchcombe⁴ calls ‘causally significant analogies between instances’ during the course of a historical investigation.

(Skocpol 1984: 374–5)

As Skocpol (1984: 378–9) makes clear, in this approach to comparative historical analysis J. S. Mill’s methods of agreement and disagreement are applied, as outlined in his A System of Logic. In the method of agreement, comparisons are made between cases in the search for similarities and causal significance is ascribed to the factors that agree across the cases. In the method of disagreement, comparisons between cases are made on the basis of contrast. Differences are sought to explain why something happened in one case but not the other and it is the presence of the disagreed factor that is given causal significance.

Mill’s methods, however, are open to criticism and they are particularly vulnerable to criticism when applied to a small number of cases (Lieberson 1992). The major problem is that, limited to a small number of cases, the methods are unable to examine probabilistic relationships and must assume that relationships between variables are deterministic. Yet, in the real world, most hypotheses are probabilistic, which relates to a second problem. The methods are also unsuited to revealing the interaction between variables and the assumption that each variable is independent of each other goes against the nature of hypotheses involving many variables reacting on each other. Furthermore, such complex hypotheses are the type most usual in the study of subjects such as democracy or revolution. Single cause hypotheses, to which the methods of agreement and disagreement are suited, are rare in such research areas.⁵ Because of the small number of cases, the methods of agreement and disagreement also run the risk of associations being found and given causal significance when, in fact, they are due entirely to chance (Lieberson 1992: 113).

In A System of Logic, first published in 1724, Mill (1967: 253) identifies not just two but ‘four methods of experimental inquiry’. In addition to the method of agreement and the method of disagreement, he also identifies the ‘method of residues’, which is a joint method of agreement and disagreement, and the ‘method of concomitant variations’, which is concerned with cases where a phenomenon is inseparably present for all cases but where variations occur. Mill (ibid.: 285–91) is aware of the limitations of his methods and is particularly concerned with cases where causes are plural, where causes are not independent of each other and where the number of cases is small. It is because of his concerns that he generally recommends his joint method, the method of residues, which combines use of both the methods of agreement and disagreement. Importantly, offering all four as purely inductive methods, he accepts that deciding on what to compare may be shaped by hypotheses deduced from existing theories (laws) (ibid.: 287).

Though Mill is aware of some, at least, of the limitations of his methods, even where there are sufficient cases and hypotheses are simple the crucial problem remains that the choice of cases for the methods of agreement and disagreement will differ and the combination of the two approaches to seeking regularities may therefore produce misleading findings. As illustration, in States and Social Revolutions, Skocpol offers France, Russia and China as cases of ‘social revolution’ and also of ‘bourgeois revolution’ (France) and of ‘communist revolution’ (Russia and China). The hypothesis under consideration determines whether a case is to be used for contrast (bourgeois or communist) or compared for similarity (social revolution). Clearly, if France and China are compared as communist revolutions then it is doubtful that much will be learnt of value about communist revolutions. Less obviously, comparison of France and China as social revolutions may also mislead understanding. For example, the Communist Party and the guerrilla army played crucial roles in one of the cases (China) but not in the other (France) and as a consequence their relevance will be downplayed, even ignored, in the explanation for social revolutions.⁶

Selectivity bias, the selection of cases because they are most likely to confirm the hypothesis or support the theory being induced also presents problems (King, Keohane and Verba 1994: 128–9). This criticism has been levelled at Moore’s Social Origins for excluding small countries: the argument is discussed below in more detail. Selection bias is especially problematic when cases chosen are only those of examples of the phenomenon to be studied (selection on the dependent variable) because, lacking variation, the controls essential for the comparative method are absent. This is a criticism levelled at Skocpol (1979) for her choice of three social revolutions, France, Russia and China. In contrast, Moore (1969) has cases not only of revolution leading to democracy (England, America and France) and of revolutions leading to communist regimes (China with Russia as a lesser control case) but also countries that avoided revolutions and had fascist regimes (Japan with Germany and Italy as lesser control cases). Skocpol modifies her dependent variable selection bias through consideration of the lesser control cases of England, Germany and Japan.

The combined methods of agreement and disagreement do not, then, escape the problem that Skocpol attributes to the deductive method of, potentially, misleading understanding through downplaying or ignoring crucial factors, while wrongly highlighting the importance of others. While it may be true that the deductive approach, in viewing everything through a preconceived theory or model, may portray the evidence in a biased light, in the inductive method it is the overarching explanation that may be concealed. In the inductive approach, the underlying conditions that cause the variables under analysis to behave as they do may never be investigated. It is this that is of greatest concern to Burawoy (1989).

In opposition to Skocpol, Burawoy (1989) has argued strongly for the advantages of the deductive method over the application of Mill’s inductive methods employed in the search for causal patterns in history. More specifically, Burawoy makes a detailed case for the superiority of the methodology of scientific research programme, as proposed by Lakatos (1978). The methodology involves not simply the application of a theory but incorporates the dynamic of developing new adaptations to the theory in ‘a progressive defense of the hard core (theory)’ that ‘takes the form of an expanding belt of theories that increase the corroborated empirical content and solve successive puzzles’ (Burawoy 1989: 761). New hypotheses may be developed, but they remain adaptations of the core theory; cases are not used for refutation of the theory but, through appreciation of its shortcomings, for its strengthening and refinement.

Burawoy demonstrates and defends his position through comparison of Skocpol’s States and Social Revolutions, as an example of the inductive approach, with Trotsky’s application of Marx’s theory in the analysis of the Russian Revolution, as an example of the deductive research programme approach. Burawoy (1989: 762) argues, first, that the methodology of research programme has the greater chance of expanding our knowledge because, in moving beyond simply accounting for what happened to producing new lessons, it has the capacity for generating predictions that may, in some instances, be later corroborated. Second, Burawoy (ibid.: 769–70) argues that while the application of Mill’s logic rules out the possibility that one case (of revolution) affects another, a research programme can build in, as he puts it, ‘historical emulation, borrowings or breakthroughs’ (ibid.: 770). These are exactly the kinds of complexities for which the methods of agreement and disagreement are poorly designed.

In sum, Burawoy’s argument is that, on balance, it is the application of a general theory to a historical case or cases that is the more scientific approach than is the search for regularities through the application of Mill’s system of logic. A research programme seeks out problem cases and engages in problem-solving and stresses falsification and prediction over verification. Furthermore, Burawoy (1989: 777–8) contends that, in practice, Skocpol succeeds in producing a stimulating new idea on the causes of revolutions not because she slavishly follows Mill’s inductive approach but because she drops the application of his methods at crucial points and introduces hypotheses derived from existing theories (Moore’s and Marx’s) along with insights brought to the analysis through her own experiences of the present. As explained above, however, the inductive method as proposed by Skocpol and explicit in her statement that ‘ideas about causal regularities may come from two or more preexisting theories that are brought into confrontation with the historical evidence’ (Skocpol 1984: 374–5), does not rule out pre-existing theoretical hypotheses. As also explained, in allowing that ideas about causal regularities may come not only from the cases but also from pre-existing theories, Mill’s ‘laws’, Skocpol is not deviating from Mill.

Generating hypotheses

The importance of hypotheses deduced from pre-existing theories is common to both Burawoy’s and Skocpol’s approaches. The difference between their methods is not simply that the former is structured through the application of one core theory while the latter involves hypotheses deduced from several models or theories together with hypotheses induced from the cases. Burawoy’s defence of Trotsky’s approach lies also in its value for practice. Trotsky wanted revolution of a particular kind, one that produced a communist society, and his exploration of the Russian case through Marx’s theory, focusing on anomalies, produced hypotheses about future practice. In this lies an important lesson about the kinds of hypotheses that need to be investigated in a comparative historical study of democracy. Just as Trotsky wanted Marx’s theory to come true and so applied Marx’s theory analytically to Russia not simply to gain deeper theoretical understanding of revolution but also to draw practical lessons for making revolution, so those involved in making democracy want it to happen. Their whole purpose is to devise schemes, structures, means, procedures and constitutions in order to make democracy a reality.

In seeking to understand how and why paths have been taken to democracy and how and why democracies have adopted their particular form, it is necessary to consider not only social scientific hypotheses concerned with the underlying conditions of democracy. It is also necessary to include consideration of the fact that the actors were trying to create democracy and that their ideas and decisions were likely to impact on the choices made about political institutions and procedures. It follows that ideas about democracy may, for example, be affected by the works of political philosophers being discussed at the time when the new political system is being set up⁷ or by ideas being debated about the attractions or otherwise of earlier or contemporary democracies elsewhere. It follows also that the constitutional design chosen and the ideas behind its choice need to be considered in analysis.

The idea that the ideas and practicalities of democracy debated at the time at which the institutions and processes of democracy are chosen may have significant and independent effects on political outcomes raises questions not only about the importance of actors and ideas but also about the nature of history as explanation. History is not a narrative from which social scientists glean evidence to be employed as data for the testing of hypotheses. There is not one history but many approaches to a particular phenomenon or era. There is social history, economic history, political history, the history of ideas, international history and so on. Added to this, t...

Table of contents

• Cover
• Title
• Copyright
• Contents
• Preface
• Acknowledgements
• PART I. Democracy: Method and theses
• PART II. Revolution: Chance for democracy
• PART III. Totalitarianism: Antithesis of democracy
• PART IV. Democracy as synthesis
• PART V. Conclusion: Lessons for democracy
• Notes
• Bibliography
• Index