リアリズムが国際政治学をダメにした: ハンス・J・モーゲンソウの場合 3-3

Pocket

第7章: 評価 ─ リアリスト・パラダイムの妥当性

前章までの分析で、1950年代以降、リアリズムが国際政治学のパラダイムであることが明らかになった。本章以降、リアリスト・パラダイムが国際事象を正確に説明・予測しているか調べてみる。

パラダイムの評価法
①正確性の基準(the criterion of accuracy): まだ反証されていない仮説の数

Once the production of knowledge has been selected as the major criterion, the next problem is to specify a set of criteria that can be used to determine whether or not a paradigm has produced any knowledge. It was seen in chapter 1 that whether a paradigm produced knowledge could be determined by examining the empirical content of its theories, that is, the number of hypotheses that have failed to be falsified. This criterion will be called the criterion of accuracy, since it reflects the ability of the paradigm to predict behavior accurately. (p.122)

②中心性の基準(the criterion of centrality): パラダイムの中でとくに中心的な仮説群の中でまだ反証されていない仮説の数

It was also seen in chapter 1 that the ability of a theory to produce hypotheses that fail to be falsified is only a minimum requirement. More important, a theory must fail to falsify hypotheses that are intended to test its central propositions, where centrality is defined as the level of generality, the scope, and the uniqueness of the proposition (see Stinchcombe 1968: 17-22). The reason for this rule is that the central propositions form the heart of the theory, and if they are falsified, then any incidental propositions that fail to be falsified can be easily incorporated into a rival theory (if they are not already part of that rival theory). Applying this same logic to paradigms, it can be said that a paradigm’s central propositions must fail to be falsified when tested. The latter principle will be called the criterion of centrality. (p.122)

③科学的重要性の基準(the criterion of scientific importance): パラダイムが提供する知識が取るに足りないことでなく価値あるものであること

The criteria of accuracy and centrality provide two rules for determining whether a paradigm produces knowledge. Production of knowledge, however, is only a necessary condition for paradigm adequacy. The knowledge must also be of some value. A number of secondary criteria could be provided to assess the value of the produced knowledge, but there is not much consensus in the field over what those criteria might be. One major criterion that scholars agree on is that the knowledge should not be trivial. Recognizing that other secondary criteria can be employed, this analysis will only employ one, that the knowledge should be non-obvious to a large segment of scholars in the field. This criterion will be called the criterion of scientific importance. (pp.122-3)

パラダイム評価の枠組み

上記の3つの基準をテストする。

In order to determine the extent to which the realist paradigm has satisfied the three criteria of paradigm adequacy, the following propositions will be tested:
4. The realist paradigm should tend to produce hypotheses that fail to be falsified.
5. The central propositions of the realist paradigm should tend to produce hypotheses that fail to be falsified.
6. Realist hypotheses that fail to be falsified should be of scientific importance.
If the above propositions fail to be falsified, then it can be concluded that the realist paradigm has been an adequate guide to scientific international relations inquiry. If the above propositions are falsified, then the claim that the realist paradigm was not very accurate in explaining behavior will be given credence. (p.123)

評価基準は、リアリスト・パラダイムが、その他大勢の非リアリストの議論よりもより国際事象をうまく説明しているかどうかである。かなり大雑把なテスト方法にも思えますが、Vasquezはこれがリアリストには有利な(つまり、リアリストよりも非リアリストの主張の方が正しい場合、リアリストへの打撃は大きい)crucial testなのでこのようなテストでもかまわないことを示唆しています。

The decision-rule that first comes to mind would be to employ a statistical significance level (such as 0.05). To insist that a paradigm in international relations produce a statistically significant number of “accurate” hypotheses, however, would be quite unfair, given the youthfulness of the discipline and the exploratory nature of much research. A fairer requirement might be one suggested by Lakatos (1970). He states that a theory’s adequacy can be evaluated by comparing the empirical content of one theory with the empirical content of a rival theory (Lakatos 1970: 116). Applying the same logic to paradigms, a decision-rule that would permit the three propositions to be tested would be to insist that the realist paradigm produce proportionally more knowledge than its rival paradigms. The problem with applying this rule is that research in rival paradigms such as Marxism, transnational relations, or issue politics has not been conducted in the field. The only alternative is to compare the performance of the realist paradigm with the non-realist hypotheses that have been tested in the field (about 7 percent of all the hypotheses [see figure 6.3]). These non-realist hypotheses share the common characteristic of “not being realist,” but they do not share a well-defined rival paradigm. To expect such a “non-paradigm,” which has so few tests, to produce proportionally more accurate findings than the realist paradigm is giving the latter more than the benefit of the doubt. Nevertheless, if the realist paradigm failed to pass this test, it would demonstrate that the realist paradigm was not adequate and suggest that even a simple rejection of one or more of the realist assumptions might provide a better guide to research. For these reasons, the decision-rule that proportionally more realist than non-realist hypotheses should fail to be falsified or be of scientific importance will be employed. (p.124)

データ
以下参照

The sample consists of all correlational/explanatory articles listed in Jones and Singer, Beyond Conjecture in International Politics, that employ inductive statistics or measures of association to test hypotheses. A content analysis of the original articles produced a sample of 7,827 hypotheses. The following information was collected on these hypotheses: number of hypotheses tested in article; number of independent variables; actor, topic of inquiry, and paradigm of independent and dependent variables; name of independent and dependent variables; paradigm of hypothesis; statistics employed; significance level; strength of association. Reliability of the coded part of the data was established as 0.90. Since questions of data validity were discussed in chapter 6, there is no need to repeat the arguments here. (p.125)

正確性の基準のテスト
リアリスト仮説の93.1%が反証される。それに対して反証された非リアリスト仮説は83.1%。反証された割合は非リアリスト仮説の方が小さいのでよ、正確性の基準のテストはパスしませんでした。

The accuracy of a paradigm, therefore, can be operationally stipulated as the extent to which a paradigm produces hypotheses which, when tested by the use of inductive statistics and measures of association, are found to be statistically significant and have strong measures of association. (p.126)

The findings of the test of hypothesis 4 are reported in figure 7.1. It can be seen from this figure that 93.1 percent of the realist hypotheses were falsified, compared to 83.1 percent of the non-realist hypotheses. This means that 93.1 percent of the realist hypotheses and 83.1 percent of the non-realist hypotheses fell into the weak category of PPI (B); that is, they were statistically insignificant or had a measure of association of |0.50| or less. Turning to the stronger categories in PPI (B), it can be seen that only 2.2 percent of the realist hypotheses have a measure of association higher than |0.71| (categories 75 and 100) compared to 4.6 percent of the non-realist hypotheses. (p.131)

The test results of hypothesis 4 indicate that the realist paradigm has not been very successful in passing empirical texts. Although early success would not be expected, one would not expect about 90 percent of over 7,000 realist hypotheses to be falsified. Also, the fact that non-realist hypotheses, which consist of simply rejecting the fundamental assumptions of realism, can more successfully pass empirical tests than the realist paradigm, which has been the object of much work, raises serious questions about the accuracy of the realist paradigm. In light of these findings, it can be said that proposition 4 has not been supported, and it can be tentatively concluded that the realist paradigm has not, up to this time, satisfied the criterion of accuracy. (131)

中心性の基準のテスト
このテストでは、リアリストパラダイムにおいて特に重要な仮説(generality, scope, uniquenessの高い仮説)が実証テストで反証されないか確かめます。

The criterion of centrality is based on the recognition that certain propositions in a paradigm are more important than others. They are more important either because the adherents of the paradigm claim that these propositions have greater theoretical explanatory power or because they are what distinguishes the paradigm from rival paradigms. Because these propositions form, in a sense, the heart of the paradigm, it is important that hypotheses testing these propositions should fail to be falsified. Unlike the criterion of accuracy, the criterion of centrality introduces a qualitative element in assessing paradigm adequacy. It does not treat every hypothesis as equal in importance, but establishes a category of hypotheses that are given more weight. (p.132)

Early in this chapter, centrality was defined as the level of generality, scope, and uniqueness of a proposition. The more universal the proposition, the greater its generality. For example, a proposition intended to hold for all nations during the last two hundred years is more general than a proposition intended to hold only for Latin American nations in the last twenty years. Scope refers to the variety of phenomena or behavior the proposition intends to explain. The greater the variety of phenomena a proposition intends to explain, the greater its scope. For example, a proposition that attempts to explain all kinds of inter-nation conflict-cooperation is obviously greater in scope than a proposition that attempts to explain only economic conflict-cooperation among nations. Uniqueness refers to whether rival theories contain the same proposition. Uniqueness is included because it is the criterion by which one theory or paradigm is distinguished from another. Unique propositions, no matter what their generality or scope, provide the reasons for selecting one theory or paradigm over a rival. These three definitions can be employed to operationally define the criterion of centrality as the failure to falsify hypotheses that: hold over long periods of time and a great deal of space; explain a variety of phenomena or behavior; and offer predictions that are not made by a rival paradigm(s). (p.133)

この基準に沿う仮説は対立/協力変数に関係する国力変数と同盟変数ということで、これらに関する仮説を中心的仮説、他の仮説を非中心的仮説とします。

One way of determining the importance or centrality of a hypothesis for a realist paradigm is to examine the relationships postulated among the most frequently used concepts in Politics Among Nations balance of power, national power, and war. Applying the operational criteria, albeit in a somewhat judgmental manner, it was found that propositions relating these concepts tended to be highly general, great in scope, and unique. They were general to the extent that they applied to all nations in the modern state system (i.e., since the Peace of Westphalia) (see Morgenthau 1960, 1973: 8-10). They were great in scope in that the balance of power and national power were intended to explain not only war but all types of conflict-cooperation in the system (see Morgenthau 1960, 1973: 27-28, and chs. 4 and 11). They were unique in that competing paradigms such as idealism, Marxism, transnational relations, and the issue politics paradigms did not offer them as explanations. In addition, these propositions are central to the paradigm in that, as was shown with the textual analysis presented in chapter 3, the relationship between national power and conflict-cooperation forms the key theoretical focus of Politics Among Nations. The notion of a balance of power can also be regarded as central because it sharpens the national power focus by describing the power relationship between two or more nations. (pp.133-4)

On the basis of this examination, it was decided that it would be valid to declare that the national power variables and alliance variables that were related to conflict-cooperation variables were indicators of central hypotheses in the realist paradigm. All other variable relationships were coded as non-central. This nominal classification was used in the analysis as the first measure of the nation of centrality. (p.134)

命題5. リアリスト・パラダイムが中心性の基準を満たすならば、中心性の命題は反証されない仮説を生み出しているはずだ(p.134)。
仮説5a. 国力と同盟を国家間の対立と協調に関連させるリアリストの仮説は、他の非中心的なリアリストもしくは非リアリストの仮説よりも反証されていないはずだ(p.135)。
仮説5b. 独立変数として国力と国家間同盟を扱っている仮説は違う独立変数を扱っている仮説よりも反証されていないはずだ(p.135)。
仮説5c. 従属変数として国家間の対立と協力を扱っている仮説は違う従属変数を扱っている仮説よりも反証されていないはずだ(p.135)。

いずれの仮説も反証されました。リアリストが特に重要と見なしている主張はとくに実証的に正しいというわけではないという事です。

The three tests of proposition 5 are hardly encouraging for the realist paradigm. It has been found that the central realist hypotheses that relate national power or inter-nation alliances with inter-nation conflict-cooperation employ national power or inter-nation alliances as predictors, or try to predict inter-nation conflict-cooperation have been consistently falsified. These findings indicate that the area of the realist paradigm that promised to be the most theoretically powerful, the central power politics framework, have been among the poorest performers in actually predicting behavior. It has been found that non-central realist hypotheses and non-realist hypotheses provide more adequate predictions of behavior, even though these hypotheses have not been as extensively elaborated and tested as the central realist hypotheses. In light of these findings, it can be said that proposition 5 has not been supported. Therefore, it can be tentatively concluded that the realist paradigm has not satisfied the criterion of centrality. (p.143)

科学的重要性の基準のテスト

命題6. もしリアリスト・パラダイムが科学的重要性の基準を満たしているならば、反証されなかったリアリスト仮説は重要であるに違いない(p.145)。
仮説6. 反証されていないリアリスト仮説は、反証されていない非リアリスト仮説よりも科学的に重要であるに違いない(p.145)。

分析の結果は、69.5%のリアリスト仮説および54.2%の非リアリスト仮説はtrivialな仮説にすぎない。非リアリストの仮説の方が成績が良いのでこのテストもパスしませんでした。

The performance of realist and non-realist hypotheses, as evaluated by the criterion of scientific importance, is reported in figure 7.2. It can be seen from this figure that about two-thirds (69.5 percent) of the realist hypotheses were declared trivial, compared to about half (54.2 percent) of the non-realist hypotheses. On the basis of this distribution it appears that the non-realist hypotheses have performed slightly better. This finding suggests that accepting rather than rejecting realist assumptions does not result in comparatively more scientifically important findings. However, before reaching this conclusion a more detailed review of the nontrivial findings is warranted. (p.146)

このあたりは、実証テストも雑すぎてあまり真剣に読めませんでした。

結論は、「リアリストだめじゃん!」

The tests of the three propositions lead one to conclude that the realist paradigm has not been a very adequate guide to knowledge. The use of quantitative analysis to test aspects of the realist paradigm, which began in 1956 and was well underway by the mid-1960s, did not produce much knowledge by the 1970s, although it commanded a great deal of effort. The field did not, as had been expected, move “beyond conjecture.” The question that must be answered is why this is the case. This book suggests that the reason for this dismal performance is that the view of the world provided by the realist paradigm is incorrect. (p.150)

It can therefore be concluded that while the present data analysis has not demonstrated beyond doubt that the realist paradigm is inadequate, it has raised the following questions about its adequacy. If the view of the world presented by the realist paradigm is correct or useful as a guide to understanding, why have so many hypotheses guided by this view been consistently falsified? If the view of the realist paradigm is correct, why have hypotheses that have rejected the view been falsified proportionally less often? If the view of the realist paradigm is correct, why have the central realist propositions, which have been extensively elaborated and tested, been consistently falsified? If the view of the realist paradigm is correct, why has the realist paradigm produced only 48 scientifically important findings out of 7,158 realist hypotheses that were tested from 1956 to 1970? (p.153)
These questions must be answered. In Kuhn’s terms, these questions pose an anomaly for the field. How the field deals with the anomaly depends on what individual scholars believe has caused the anomaly. The present chapter has gone about as far as possible in terms of delineating a “cause.” Until there is more evidence, a definitive assessment of the adequacy of the realist paradigm cannot be made. The present analysis has served to raise as a serious question the possibility that the most fundamental assumptions of the field are incorrect. (p.153)

第8章 1970年代における理論と研究 ─ 新たな変則事例

第7章までの実証分析で使われたデータは1960年代以前のものだったので、本章で1970年代に国際政治研究について概観しています。

第1の仮定について

国家が国際政治におけるもっとも重要なアクターであるという仮定と、それにふづぃして国家が合理的であるという仮定について心理学的な側面からと、group thinkやbureaucratic politics modelといった組織論的な側面から政策決定者は合理的に行動しないと批判しています。ただし、合理的アクターというのは仮定にすぎないのでそれが「事実」ではないと批判してもあまり意味はないかとか思います。

Even though the rational-actor model employed by the power politics theorists is a more sophisticated and less stringent version of the rational decision-making model discussed in public administration, it has been undercut by recent research in psychology. The first argument against the model is that individuals and groups generally do not make decisions in a rational manner, because they process information not on the basis of logical rules but on the basis of a set of psychological principles which do not necessarily correspond with logical reasoning (see Jervis 1976; Janis and Mann 1977).  (pp.156-7)

These are strong tendencies, but this model does not mean that good decisions cannot be made or that the tendencies cannot be controlled. Jervis (1976: 3-10, 165-172) relies on the example of scientific reasoning to show that such tendencies do not necessarily result in disastrous information processing. Nevertheless, it is clear that models of action, particularly foreign policy action, based on the assumption of selfish interest and/or calculation of costs and benefits, are too simplistic as either descriptions or predictions of behavior, and certainly as explanations. (p.157)

The second argument against the power politics version of the rational-actor model is that, since certain types of individuals and specific kinds of groups behave differently, it is incorrect to assume that they would all behave rationally. This means that a state’s foreign policy cannot be deduced on the basis of a rational national-interest calculus, because personal and/or idiosyncratic factors affect individual behavior, and internal structural characteristics affect group decision making. There is some scientific evidence within the field relevant to this problem, consisting mostly of some experiments conducted with the Inter-Nation Simulation. (p.157)

次に、リアリストの①国家が単一アクター(unitary actor)である、②非国家的アクターは重要でない、という考えが批判されています。

After the criticism of the rational-actor model, the two other major areas of the first assumption of the realist paradigm that have been criticized are the notion that nation-states are unitary actors and the idea that non-state actors are relatively unimportant. The idea of the nation-state as a unitary actor is sometimes referred to as the billiard ball model, or black-boxing internal politics (Burton et al. 1974: 6). Clearly, the research of G. H. Snyder and Diesing (1977) and of Jervis (1976), as well as the theoretical work of Allison (1971) and Halperin (1974), raises serious questions about treating decision making in a state as if the state were a single collectivity. Nation-states may not have a single interest or a single coherent policy developed by a cohesive group; instead, the foreign policy of a state may very well reflect internal political outcomes. In light of the research mentioned here, it cannot be automatically assumed that the state is a unitary actor; rather, this must be investigated empirically to determine which states can be treated in a billiard ball fashion and whether they can be treated that way in all issue areas. (p.165)

The review of the first assumption suggests that it is an inadequate guide to inquiry because the behavior of nation-states cannot be explained solely by the power realities of world politics. This is because individual decision makers will differ, and the ability of governments to act in a unitary fashion will vary. Recent research has delineated the pitfalls of trying to deduce the foreign policy of a state by using a rational-actor model. In addition, the bureaucratic politics perspective has caused questions to be raised about the potency of external factors on the foreign policy of a state in non-crisis situations. Finally, the presence of non-state actors and the role they play in penetration has raised questions about the fundamental conception provided by the paradigm’s state-centric emphasis. (p.167)

第3の仮定について

国際政治は権力と平和の追求であるという第3の仮定について、安全保障の分野ではそうかもしれないが、安全保障以外の分野(たとえば経済、食料、環境など)では当てはまらないという主張(そりゃそうだ。でもSo what?)

The analysis of issue areas suggests to some that power politics behavior is confined to territorial and military issues and does not reflect behavior in other issue areas (particularly economic questions, but also other transnational areas that need regulation – e.g., food, the sea, the environment, air travel, etc.). Handelman et al. (1973), Coplin (1974: ch. 13), Kihl (1971), Hopkins and Puchala (1978), and Vasquez (1974b) have given credence to this view, and a data-based study of issue areas by O’Leary (1976) shows that behavior does vary by issue area. This suggests that a period that appears power politics prone is probably dominated by certain kinds of issues or some other issue characteristic (see Dean and Vasquez 1976: 18-28). An issue politics paradigm could provide a very attractive alternative to the realist paradigm in that it provides a broader perspective that explains both power politics and non-power politics behavior and the relationship between the two. Before such a paradigm can be taken seriously, however, it requires a real theory (not just a framework) and some supporting evidence. By the end of the 1970s, unfortunately, neither was forthcoming. Nevertheless, these criticisms of the realist paradigm raise problems that the paradigm’s adherents must address and that may with further work provide important anomalies. (p.169)

勢力均衡政策は平和をもたらさないとリアリストを批判しています。

The studies on polarity and on the balance of power pose anomalies for the realist paradigm, or at least that aspect of it that places emphasis on alliances as a way toward peace. In the mid-1960s, many scholars debated whether a bipolar or multipolar system would produce peace. If the realist paradigm were an adequate guide to inquiry, at least one side would have been expected to be correct. Instead, both were wrong. The only major difference is whether one will pay the Grim Reaper all at once with a few severe wars, or on the installment plan with many wars. (p.172)

It is now clear that alliances do not produce peace but lead to war. Alliance making is an indicator that there is a danger of war in the near future (less than four years). This means that the attempt to balance power is itself part of the very behavior that leads to war. This conclusion supports the earlier claim that power politics is an image of the world that encourages behavior that helps bring about war. Since it is now known that alliances, no matter what their form, do not being about peace, the theoretically interesting question is what causes actors to seek alliances. This question begins to push beyond the parameters of the third assumption. (p.172)

新しいパラダイムに向けて

What is significant about the preceding findings is not any single definitive finding (each individual piece of research could be challenged or explained away), but the consistent pattern that appears to be emerging from the research. As the field truly begins to progress, propositions based on realist assumptions do not do as well as those that reject realist assumptions. This conclusion holds for both the first assumption and the third, and for studies dealing with both foreign policy and the causes of war. Specific research findings have been produced that would not have been expected if power politics explanations were accurate, and realist assumptions seem to ignore certain phenomena or ways of perceiving these phenomena that have later led to important theoretical explanations and accurate predictions. While there is no need at this point to decide whether the realist paradigm should be rejected, it can be concluded that the research of the 1970s has called that paradigm into question, that the paradigm has still failed to satisfy the criteria of accuracy and centrality, and that it has satisfied the criterion of scientific importance less well than status, social psychological, or cognitive psychological explanations of global behavior. (p.175)

If the analysis presented in this chapter and chapter 7 is correct, the most pressing task for the field is to develop an alternative paradigm. When the findings on status explanations are coupled with social psychological models, the elements of a potentially powerful non-realist paradigm begin to take shape. While a detailed exposition of those elements is beyond the scope of this book, the general outlines can be suggested, and the major problems that a new paradigm must deal with can be delineated. (p.176)

国際関係の科学的研究の未来

リアリスト・パラダイムをつぶさないと国際政治学の将来はないよという結論。

This book attributes the absence of many findings in the field to the dominance of an inadequate paradigm – that is, the realist paradigm. It assumes that although there may be some measurement error in research, the primary problem lies not in the research methodology of the field but in the incorrectness of the hypotheses that are being tested. Until a paradigm is found that shows promise of adequately explaining behavior, there will be no major progress in research. This implies that the realist paradigm must be rejected as the dominant paradigm in the field. Since the realist paradigm is not likely to be rejected in the absence of a better paradigm, the strategy that this explanation suggests is to have more paradigm diversity in the field. (p.179)

John A. VasquezのThe Power of Politics: From Classical Realism to NeotraditionalimのPart Iを読み直してみましたが、1983年に出版されたThe Power of Power Politics: A Critique(Part Iは1983年版の原稿をそのまま転載しています)が、国際政治学者にほとんど無視されたであろうことは容易に想像がつきます。それは通常科学と化したリアリストパラダイムが国際政治学コミュニティーの中で強固すぎたというよりも、国際政治学といったfragileな学問分野にパラダイムなるものは元々存在していなかったからです。Vasquezが言う現実主義パラダイムの3つの基本的仮定は、①国際関係におけるもっとも重要なアクターは国家である。②国際政治は国内政治と根本的に異なる。③国際関係は権力闘争である、とのことでしたが、①は便宜的な単純化にすぎず、③は国際政治におけるハイ・ポリティックスの分野ではそうだということにすぎないかと思います。Vasquezが扱ったデータは1950年代から1960年代にかけてのものですが、その当時、軍事問題への関心が高かったのは、リアリスト・パラダイムが指示したからというよりも、当時は米ソ冷戦が切迫した状況だったらでしょう。国際政治では多数の国家の上位に位置する上位アクターが存在しない点で、政府が存在する国内政治とは根本的に異なるという第2仮定は確かに重要なポイントですが、リベラリストもその前提を受けていれた上で(超国家的組織の設立を訴えるリベラリストを除いて)国家間協力の途を探っているわけだし、トマス・クーンの科学論を社会科学の中でもとくにinterdisciplinaryな何でもありの国際政治学という分野に適用できるかについてははなはだ疑問です。

では次回からPart IIを検討。ケネス・ウォルツやジョン・ミアシャイマーがぼろくそ叩かれています。

 

12月 19, 2014 · Pukuro · No Comments
Posted in: ☆社会科学

Leave a Reply