Week 4 Summaries

Jeffry A. Frieden, "Actors and Preferences in International Affairs," in Powell and Lake, eds., Strategic Choice and International Relations.

Two principal points: (1) preferences must be kept separate from the strategic setting in order to distinguish between the causal role of actors' interests and the causal role of the environment; and (2) whether preferences are variables of interest or control variables, scholars must be explicit about how preferences are determined.   Recall that the 'strategic setting' or 'environment' is defined by the actions available to the actors--the cells in a PD game--and an information structure that defines what the actors can know for sure and what they have to infer from the behavior of others (AND therefore may include institutions as well.)

Preferences are part of all explanations yet are not directly observable, so care and precision is needed.  An actor *prefers* some *outcomes* to others and pursues a *strategy* to achieve its most preferred possible outcome.  Preferences, strategies, and outcomes are all distinct but related concepts.  Within a given interaction, preferences are typically held constant in order to determine how preferences affect choices.  Yet, without information about the strategic setting and/or the process of preference formation, it is impossible to know how behavior maps back to preference.

There are three common sets of errors made with regard to preferences: (1) sins of confusion, confusing preferences with strategies; (2) sins of omission, focusing solely on preferences and ignoring the context; and (3) sins of commission, ignoring preferences altogether. 
Sins of confusion mix preferences and strategic setting: for example, to read that states maximize power and infer that this strategy is actually the state's preference, i.e., that power or survival is in the state's utility function.  Sins of omission assert that variation in outcomes has nothing to do with variations in preferences, and sins of commission assert that variation in outcomes is solely owing to variation in preferences.  Sins of omission arise when when one does not control for actors' preferences.  Sins of commission arise when analysts observe an outcome and draw a direct line back to the preferences of actors (ignoring the possibility that strategic interaction might have fundamentally transformed the process and its end point.)

Avoiding these sins requires the identification of preferences, which may be done by assumption, by observation, and/or by deduction.  Assumption might be simply assuming states prefer to maximize national wealth or size.  Economics assumes individuals and firms are wealth or profit-maximizing, but IR has different actors and different issues on many dimensions, so assuming state preferences can be hazardous.  Observation (or induction) is when the scholar attempts to determine the national preference by investigating the country's behavior.  Preferences are *revealed*.  National preferences are often traced to the ideological perspectives of national elites; inducing not 'the nation's' preferences, but those of powerful actors (individuals, or private or public groups) who determine national goals. The danger of this approach is confounding preferences with their effects.  This is especially difficult when trying to read preferences back from public postures countries take in bargaining.  **It is especially egregious to "induce" preferences from observed behavior and then use those preferences to explain this very behavior.**  Frieden's preferred method of defining preferences is to deduce or derive them on the basis of preexisting theory.  Of course, the prior preference (from which the preference to be used in analysis is derived) is, in fact, assumed or observed in precisely the ways described earlier.  Typically, the exercise is structured so that the features that determine the preferences to be derived are relatively easy to observe, so they are more easily validated.

In most cases, national preferences do not emerge seamlessly from existing theories.  Typically the application of theories of preferences to social collectivities requires a complementary theory of the aggregation of preferences, from individuals and firms up to groups, sectors, classes, and nations.

Ronald Rogowski, "Institutions as Constraints on Strategic Choice," in Lake and Powell, eds. Strategic Choice and International Relations, pp. 115-136.

Rogowski surveys the literature on the effect of institutions on foreign economic and security policy. Rogowski's survey of empirical and theoretical work challenges the realist model in which domestic institutions are entirely endogenous -- they are determined through selection in the anarchy of interstate-war. He suggests that domestic institutions determine foreign policy, and supports the observation that competition in the international arena may be less fierce than in the domestic economy.

The institutional features (independent variables) include the franchise (who makes the decision, including strength of different voting sectors (women, wealthy, bureaucracy, aristocrats, etc.), number of cross-cutting cleavages (class, religion, etc.), representational mechanisms (e.g., tenure in office, size of districts, number of representative bodies), and decision rules (number of veto points, amount of delegation). Dependent variables include substantive policy bias such as free trade or status quo, credibility of commitments, coherence (stability) of policy decisions, ability to mobilize resources and troops, and types of strategies.
The hypotheses presented are best viewed in table format, on page 135. These hypotheses are illustrative, and do not include all the hypotheses reviewed in the article. Hypotheses on the effect of the franchise include: female suffrage causes pacific policy, strong bureaucracy causes credible commitment, cross-cutting cleavages cause unstable policy, democracy causes greater extractive capacity, and democracy causes the strategy of demagoguery. Hypotheses on the effect of representational mechanisms include: larger constituency causes free trade, longer, more secure term of office causes greater credibility, single-member districts cause chance of accidental majorities, short terms and easy recall of representatives cause low slack between constituency wishes and representation, and more extractive capacity, and multiple bodies cause more bargaining, bluff, and signaling. Hypotheses on the effect of decision-rules include: more veto points cause status quo, credibility, instability, incoherence, and less extractive capacity; delegation restricts constituency access to specialists.

In conclusion, Rogowski calls for increased research on the under-studied question of the effect of domestic institutions on foreign policy, especially from a formal-theoretic and principal-agent perspective.

Grahm Allison, "Conceptual Models and the Cuban Missile Crisis," American Political Science Review 63 (September 1969), pp. 689-718.

Allison proposes three models for the analysis of foreign and military policy.
Model I, the Rational Policy Model, is used by most analysts. Happenings are understood as the more or less purposive acts of unified national governments.
According to model II, the Organizational Process Model, acts and choices are outputs of large organizations, functioning according to certain regular patterns of behavior.
Model III, the Bureaucratic Politics Model, analyzes the perceptions, motivations, positions, power and maneuvers of principal players from which the outcome emerged.

Model I

Many analysts explain and predict foreign policy using this model, in which the actor is a national government. The basic unit of analysis is policy as national choice. Happenings in foreign affairs are conceived as actions chosen by the nation or national government. Governments select the action that will maximize strategic goals and objectives.

The nation or government, conceived as a rational, unitary decision-maker, is thus the agent. This actor has one set of specified goals (a consistent utility function), one set of perceived options and a single estimate of the consequences that follow from each alternative. The rational agent selects the alternative whose consequences rank highest in terms of his goals and objectives. Analysts who use this model thus view an action as a calculated response to a strategic problem. Their explanations attempt to show what goal the government was pursuing in committing the act and how this action was a reasonable choice, given the nation's objectives.

The Rational Policy Model has several variants. Each variant places the action within a value-maximizing framework, given certain constraints. One group of variants includes national propensities reflected in an "operational code", concern with special objectives or special principles of action. A second cluster of variants focuses on the individual leader or leadership group as the actor whose preference function is maximized. 
Model II

Governments consist of large organizations among which primary responsibility for particular areas is divided. The actor is not a monolithic nation or government but rather a constellation of loosely allied organizations on top of which government leaders sit. Government behavior relevant to any important problem reflects the independent output of several organizations, partially coordinated by government leaders.

Each organization has a fixed set of standard operating procedures. The behavior of these organizations - and consequently the government - in a particular instance is thus determined primarily by routines established prior to that instance. The leaders can exercise some choice in combining outputs, but the mass of behavior is determined by previously established procedures. It does not constitute far-sighted, flexible adaptation to the specific issue.
The organizations have parochial priorities, perceptions and interests due to several factors: 1. primary responsibility to a narrow set of problems. 2. availability of selective information. 3. tenure of individuals on the organization. 4. small group pressures within the organization. Organizations also try to avoid uncertainty and do not attempt to estimate the probability distribution of future occurrences.

Organizations may change and learn over time. Learning and change follow in large part from existing procedures but sometime dramatic changes may occur, especially after major performance failures. Confronted with an undeniable failure of procedures, authorities outside the organization demand change and the existing personnel are less resistant to change.

Model III

The bureaucratic politics model, in contrast with model I, sees no unitary actor but rather many individual players. Each player occupies a position in an administration (e.g. president, minister of defense, permanent government officials, members of the press). The players participate in a competitive bargaining game, whose outcome is government behavior.

The players focus not on a single strategic issue but on many diverse intra-national problems as well. They have no consistent set of strategic objectives but rather various conceptions of national, organizational and personal goals.

The position of each player determines his priorities and perceptions as well as his advantages and handicaps in the game. Government decisions are not made by rational choice, but by "pulling and hauling that is politics." They result from compromise, coalition, competition and confusion among officials. 
Allison uses each model to analyze the US behavior in the Cuban missile Crisis. In model I the blockade of Cuba results from a consideration of the advantages and disadvantages of each alternative. The analysis based on model II focuses on the role of the CIA, the air force and the navy. The model III analysis views the decision as resulting from the debate between the top decision makers. Kennedy initially supported an air strike but Mcnamara and Robert Kennedy led him to change his mind.

Allison's conclusion is that alternative models and frames of references lead to different explanations, and analysts should be self-conscious about the net they employ.    

George W. Downs and David M. Rocke, Tacit Bargaining, Arms Races, and Arms Control, pgs 92-100.  

This article explores when/why the UNITARY ACTOR ASSUMPTION may be OK..

Realists accept this assumption; bureaucratic-politics/process and social choice scholars do not.  The social choice counterargument rests upon Arrow's Impossibility Theorem and the Condorcet cycling paradox, which demonstrate that under many circumstances policy priorities cannot be dependably aggregated across society.

Key question: What is the empirical validity of the unitary actors assumption? We don't know, because there are not enough cases to make a good test.  
But we can talk about more or less confidence in the appropriateness of this assumption.

New question: Under what circumstances can the unitary actor assumption work?

1. Bueno de Mesquita (War Trap): Under the common condition when there is a strong executive leadership in foreign policy, because when one member's vote is weighted much more heavily than the others, that individual is a necessary part of most (all) winning coalitions.  Leaders may not have the power to thrust a nation into war unilaterally, but have the power to veto other attempts to do so.  Standard operating procedures that define the nature of organizational heuristics are often suspended for rare and important events, when the preferences of specific leaders (Richelieu, Roosevelt, Hitler, etc.) rule the day.  Britain/US response to powerful enemies has been to abandon democratic procedure.
Lesson: Executives' agenda-setting power, ability to appoint key officials, access to information, and bureaucracies' hierarchical nature mean Arrow's theorem is not as relevant, and therefore it is not so naive to accept the unitary actor assumption, especially w/in the foreign policy realm.

2. When one state adopts a more unitary (authoritarian/dictatorial) approach  in its own politics, other states are likely to follow suit because of "attribution bias", which is when people overestimate the role of internal disposition and underestimate the role of external situation.  In other words, if you're in charge of your country, you are more likely to interpret the behavior of a rival country as the revealed preferences of its leadership.
But, even w/out attribution bias, authors say that rival states cannot afford to 'forgive' activities (skirmishes, harassment, treaty violations) claimed to be accidental, unintentional, or outside the leadership's direct control...BECAUSE then they would happen more frequently, and WHERE WOULD IT END?!  So, states have an incentive to act like a unitary actor and treat their rivals as if they were unitary actors.

3. Recent formal-theoretic work has shown that when supermajorities of 64% are necessary to make policy, then the mean voter will rule--this avoids Condorcet's cycling problem.  We may think that foreign policy frequently has to meet a higher standard of government and public acceptance; if so, this perspective may be a good reason to think the unitary actor assumption is ok.

4. Another formal-theoretic approach shows that when the final decision is the executive’s (who is subject to pressure, which is costly for the influencers to exercise), then the final decision is equivalent to that of a hypothetical actor whose preferences are a weighted average of all actors (including the executive.)

Micahel W. Doyle, "Kant, Libreal Legacies, and Foreign Affairs," Philosophy and Public Affairs 12:3 (Summer, 1983), pp. 205-235.

In this article, Doyle attempts to provide an explanation for the liberal peace phenomenon, which states that "even though liberal states have become involved in numerous wars with nonliberal states, constitutionally secure liberal states have yet to engage in war with one another." (213)  His answer lies in the constitutional structures of liberal states.

He first challenges the Realist explanations for the liberal peace—prudent diplomacy, similarity social structures, specific regional attributes and historical alliances.  The liberal peace cannot be attributed merely to the above explanations, since prudent diplomacy often requires preemptive strike, similar social structures of feudal and communist societies did not result in peace, and the liberal peace existed during the interwar period when there was no hegemon.  The argument that liberal states lack conflict of interests merely begs the question.

Doyle’s explanation for the liberal peace is that “a natural evolution will produce a harmony from the very disharmony of men against their will.” (227)  This evolution is argued to derive from three sources of liberalism: constitutional law, international law, and cosmopolitan law.  Constitutional law—republican representation and separation of power—are “the means by which to prepare for and meet foreign threats and to tame the ambitions of selfish and aggressive individuals.” (228)  Once this form of government is established, wars would appear as the disaster to the people’s welfare, and the people would be cautious in commencing wars due to high costs associated with them.

International law—a guarantee of respect—complements the constitutional guarantee of caution.  An understanding of the legitimate rights of all citizens and all republics comes into play, and sets up the moral foundations for the liberal peace.  “The domestically just republics, which rest on consent, presume foreign republics to be also consensual and therefore deserving of accommodation.  The experience of cooperation helps engender further cooperative behavior.” (230)

Lastly, cosmopolitan law adds material incentives to moral commitments.  The cosmopolitan law of universal hospitality permits the “spirit of commerce” to take hold of every nation, thus impelling states to promote peace.  In other words, economic interdependence leads states to enhance each other’s security by means of alliance, create crosscutting transnational ties that serve as lobbies for mutual accommodation.  Together, the three laws, that underpin liberalism, tames the effects of international anarchy.  Together, they made the political bond of liberal rights and interests a firm foundation for mutual nonaggression.

Helen V. Milner, Interests, Institutions, and Information: Domestic Politics and Information.  Princeton: Princeton University Press, 1997.  Chapters 3 and 4. (pp. 67-134)

Chapter 3:  “A Model of the Two-Level Game (Co-authored with B. Peter Rosendorff)”

Milner and Rosendorff model a two-level game with four actors.  The foreign country (F) is a unitary actor; the domestic country is disaggregated into three sub-state actors: the domestic executive (P), the domestic legislature (C), and domestic interest groups (E).  Each player wants the policy closest to its ideal point:  f for F, p for P, c for C, and e for E.  Utility decreases linearly and symmetrically as the implemented policy deviates from each player’s respective ideal.  Outcomes result from interaction at both levels of the game.  

Milner and Rosendorff use a spatial model to represent the international bargaining game and base this level of the two-level interaction on realist assumptions.  (p. 69-70)  The international level of the game is modeled using a Nash Bargaining Solution (which maximizes the sum of the products of the distance between the solution selected along the pareto frontier for each player).  The NBS does not have a “well-defined institutional structure; politics on that level are assumed to be anarchic, and international negotiations are generally conducted without a constitutionally mandated sequence of moves.”  (p. 70, also see p. 72)  

Milner and Rosendorff view the domestic interaction as central to achieving a NBS.  To model the domestic interaction, they employ a version of the agenda-setter, take-it-or leave it bargaining game, modified to include two agenda setters (F and P).  C votes the proposal up or down, resulting in the agreement point a from the status quo q.  If there is no agreement, then the equilibrium reverts to q.  

Milner and Rosendorff analyzed several situations and found that under:
International negotiation without domestic politics, no cooperation is a common outcome in the complete information setting.  When q is not between p and f, agreement is possible, but as the distance between p and f grows, the area of no agreement grows.  
Domestic politics with complete, symmetric information, the preferences of C affect the range of a if c < q < (2c – p).  (This is given by the NBS.)  If c is far from q, C’s influence over a weakens.  Thus, a is less likely to be achieved with c.  
Divided government with complete, symmetric information, the farther c is from p, the less likely a will be achieved.  If agreement is achieved, it will be closer to c than to p.  
Domestic politics with asymmetric information, cooperation is less likely because P and F are fully informed, but C is not sure about F’s preferences.  
Domestic politics with asymmetric information and an interest group, makes a more likely than without E.  E is not P, but has complete information about the agreement such that C listens for E’s endorsement, obtains information from this signal, and then votes on a.  
Divided government with asymmetric information and an interest group, a moves with respect to c and p.  
Multiple interest groups and asymmetric information, the interest group nearest the status quo exerts the greatest influence over the median legislator.  Additional endorsements provide C with more information and help it obtain an agreement closer to c.  

Chapter 4: “Political Institutions and International Cooperation”

Milner argues that domestic political institutions affect the distribution of power between national actors, determining whose interests will dominate policy making.  “Variations or changes in this institutional relationship influence the probability and terms of international cooperation.”  (p. 99)  The five key elements of the legislative process are:  agenda setting, amendment, ratification or veto, use of referendums, and side payments.  While the model in chapter 3 only considered agenda setting and ratification, chapter 4 considers how changes in the distribution of the five legislative processes affect international cooperation.   

Milner finds that when control of legislative processes “is concentrated in the executive, the executive’s preferences prevail.  When control is dispersed to the legislature, its preferences will come to the fore.”  (p. 127)  In terms of the specific processes, Milner asserts that in foreign affairs, the executive is usually the agenda setter, but that if that power falls to the legislative branch, then the legislature can set the terms of the debate.  In terms of amendment power, Milner observes, “In most parliamentary systems, the executive tends to be the agenda setter and the parliament usually has full power of amendment.”  (p. 104)  Amendment power decreases the probability of the legislature rejecting the bill.  While veto power constrains the agenda setter, amendment power is more significant because it can shift the agreement point.  The power to call referendums either allows the executive to bypass or override the legislature (if the executive has the authority to call referendums) or (if constitutionally mandated) merely reverts to the bargaining model in Chapter 3, with the median voter (instead of the median legislator) as the pivot.  Finally, side-payments may come in many forms.  Milner concludes that the probability of successful cooperative agreement is highest when decision-making power is concentrated in the most dovish domestic political actor.  

John Owen, “How Liberalism Produces Democratic Peace,” International Security 19:2 (Fall 1994), pp. 87-125.

As a background, Owen highlights three common weaknesses of the “democratic peace” proposition (i.e. that democracies seldom, if ever, go to war against one another): (1) There is a problem with proposing common definitions for war and democracy; (2) Wars are so rare that random chance could account for the democratic peace observations; (3) Democratic peace lacks a convincing theoretical foundation and accordingly a sufficient causal mechanism that might explain it is not available to us. The article addresses the 3rd weakness. Owen argues that liberal ideas cause liberal democracies to try and avoid fighting each other and that these ideas prod liberal states into war with illiberal states. Owen describes a mechanism whereby liberal ideas lead to a liberal ideology and to liberal institutions and both, in tandem, bring about democratic peace.

From an ideological perspective, liberals believe that all individuals are fundamentally equal and are best off pursuing self-preservation and material well-being. Since freedom is required for these pursuits and since peace is required for freedom; liberals conclude that coercion and violence are counter-productive. Liberals also believe that democracies seek their citizens’ true interests and that thus by definition they are trustworthy. Non-democracies may be dangerous because they seek other ends, such as conquest or plunder.   

From an institutional perspective, democratic structures ensure free debate and by that help constrain the government and help translate liberal preferences into foreign policy (even in cases when illiberal statesmen are in power). That is because illiberal leaders are unable to rally the public to fight other fellow democracies. However the above explanation is likely to take place, according to Owen, only when liberals consider the other state democratic. In other words, if its peer states do not believe it is a liberal democracy, they will not treat it as one. Hence, Owen’s explanation takes into account the importance of perceptions.     

Owen reviews two previous attempts to explain democratic peace: (1) Structural accounts that attribute the democratic peace to institutional constraints; (2) Normative accounts that attribute the democratic peace to ideas and norms held by democracies. For example, democracies believe it unjust and imprudent to fight one another.

Following an examination of the typical statistical data sets that were used to back up each of these two theories, Owen concludes that there is no clear ‘winner’. In fact, by carrying out a “process tracing” analysis on a number of historical case studies, he concludes that normative theory failed to take perceptions into account. At the same time, Owen found that democratic structures were just as likely to drive states to war as to restrain them from it. Therefore as an alternative, Owen proposes to synthesize both explanations. According to his consolidated explanation, the liberal ideology motivates some citizens against war with a fellow democracy and, at the same time, democratic institutions allow this ideology to affect the actual foreign policy formulation process.

Owen stresses that claiming that liberalism pushes democracies to peace with each other does not imply that power politics have no force in determining the foreign policies of liberal democracies. Instead, liberalism is an additional force that contributes to foreign policy shaping.

To empirically test his hypothesis regarding the causes of democratic peace (namely that a liberal democracy will only avoid war with a state that it believes to be liberal), Owen derives a set of expectations (or hypotheses) that should occur. Accordingly, if these hypotheses are falsified, Owen’s deems his theory inadequate. The expectations/hypotheses are: (1) Liberal trust states they consider liberal and mistrust states they consider illiberal. (2) When liberals observe a foreign state becoming liberal by their own standards, they will expect pacific relations with it. (3) Liberal will claim that fellow liberal democracies share their ends, and that illiberal states do not. (4) Liberals will not change their assessments of foreign states during crises with those states unless those states change their institutions (otherwise power politics or another force will determine what label liberals attached to foreign states). (5) Liberal elites will agitate for their policies during war threatening crises. (6) During crises, statesmen will be constrained to follow liberal policy.

To illustrate (not test) the validity of his argument, Owen presents four of twelve war-threatening historical case studies that he used in order to derive his argument. All twelve cases: (1) involve the US; (2) occur before 1945 (in order to neutralize the effects of bipolarity and nuclear weapons); (3) allow the perceptions and governmental systems of the other state to vary. The four cases presented in the article are: Franco-American relations in 1796-8, and Anglo-American relations during 1803-12, 1861-63, and 1895-96. While the cases demonstrate the validity of Owen’s argument, they do not test it (since the argument was derived from them). In three of the cases liberalism helped to prevent a war. In one (Anglo-American relations from 1803-12), liberalism helped bring on a war. The cases also emphasized the importance of perceptions (e.g. Americans did not perceive England as liberal in 1803-12 but later changed their perception).

Moreover, while Owen argues that realists are wrong in denying the existence of the democratic peace, he does not deny the role the power politics play in liberal democratic foreign policy. Liberals can care about the balance of power and still view it as part of a larger picture of international politics. Thus in the context of the democratic peace, a synthesis of realism and liberalism seems possible. However democratic peace does not necessarily lead to perpetual peace since threats to liberalism may lead to threats to peace.

From an IR perspective, Owen maintains that the democratic peace provides strong evidence that ideas matter in international relations, both as shapers of national interest and as builders of democratic institutions.                        

Philip G. Roeder, “Soviet Polices and Kremlin Politics,” International Studies Quarterly 28 (1984), pp. 171-193. 

Roeder argues that domestic politics matters in the construction of foreign policy.  He points out that Soviet foreign policy was not the product of a unified rational actor, but a result of the distribution of power and functions within the Kremlin.  He divides Soviet politics from 1953 to 1977 into five regime types, and uses them to show that the consistency, coherency, responsiveness, and risk-behavior of the Soviet Union varied over time because the domestic regime type varied over time. 

“The causal nexus between the two dimensions of power (consolidation and competition) may be summarized” as (179-180)

I.1:   The consistency of policy varies inversely with the competition for power.
II.1:  The coherence of policy varies inversely with the competition for power. 
II.2:  The coherence of policy varies directly with the consolidation of decisionmaking authority. 
III.1: The responsiveness of policy varies inversely with the competition for power. 
III.2: The responsiveness of policy varies curvilinearly with the consolidation of decisionmaking authority, reaching a maximum at intermediate levels of consolidation.
IV.1: Risk-taking increases directly with the competition for power. 
IV.2: Risk-taking increases directly with the consolidation of decisionmaking authority.

Regime Types

A pluralistic regime (from Stalin’s death in March, 1953 to Khrushchev’s victory over the Anti-Party Group in June, 1957) is “associated with the succession crisis” (174) and characterized by the relatively equal distribution of power between the Party, State, Military, and Police.  Roeder expects little consistency, coherence, responsiveness, and low levels of risk-taking during this period. 

The directive period (from June, 1957 to the U-2 incident in May, 1960) was characterized by the “consolidation of decisionmaking authority in Khrushchev’s hands and very limited competition for power.”  (175)  Roeder expects high consistency and coherence, but low levels of responsiveness and risk-taking. 

The primatial period (from May, 1960 to Khrushchev’s fall in October, 1964) was characterized by Khrushchev’s struggle for survival (primus inter pares behavior), while retaining disproportionate control over foreign policy.  Roeder expects little consistency, coherence, or responsiveness, coupled with risky behavior. 

During the oligarchic period (from October, 1964 to the early 1970s), there was “limited competition of power and dispersed authority over policy.”  (176)  “[C]ommitment to ‘proceduralism’ limited the conflict within the elite.”  (177)  Roeder expects high consistency, but little coherence or responsiveness, coupled with low levels of risk-taking behavior. 

Roeder distinguishes the cartelistic period (from about April, 1973 to April, 1977) from the oligarchic period.  During the cartelistic period, competition for power within the central elite was further reduced, and elite participation in decisionmaking was narrowed on functional grounds.  Roeder predicts high consistency, coherence, and responsiveness, combined with low levels of risk-taking. 


Roeder uses the monthly data from the Conflict and Peace Data Bank (COPDAB), which contains responses and effects from “reportable” or “newsworthy” events over the period in question.  (182-3)  “Each event is classified by five attributes relevant tot eh present study—date, actor-state, target-state, scale value measuring the intensity of conflict or cooperation, and issue area.  The monthly aggregate data use here measure the magnitude (that is, the frequency multiplied by the intensity) of conflict and cooperation in each issue area by summing the number of events weighted by the intensity of each event.” (183)  He excludes the 14 data points associated with the Berlin crisis, the Cuban missle crisis, and the Czechoslovakian crisis; without these outliers, he has 276 data points.  He excludes the crises on the grounds that they “upset the prevailing decisionmaking patterns.  In addition to this theoretical reason … there is also a strong methodological reason—the assumption of normality in the t-test.”  (184) 


Of Roeder’s 20 predictions (5 regime types, four measures), he finds 17 of them statistically significant at the 5% level.  The three exceptions are:  consistency and coherence during the cartelistic period, and coherence during the directive period.  The correlation between predicted and actual behavior leads Roeder to conclude that American analysts should keep in mind that Soviet actions in foreign policy may result from domestic events, and that in periods of greater dispersal of authority, it becomes difficult to discern Soviet intent because multiple sub-state actors may be driving policy.  (189)