ArticlePDF Available

Only Friends Can Betray You: International Rivalry and Domestic Politics

Authors:

Abstract and Figures

This article argues that dramatic political change in State A poses a threat to the interests that other states B share with it. The more salient those interests are to State B, the greater the threat posed by domestic political change within State A. Thus major changes in one state place the leaders of formerly friendly states into a domain of losses, motivating risk-seeking behavior in hopes of reversing these losses. Conversely, the new leaders of the state undergoing domestic political change initiate similarly risk-seeking behavior to defend their new endowments. The conflicts that result sow the seeds of long-lasting enmity (rivalry) between former partners. I test this argument in a dataset of rivalry onset during the period 1950–2005 and find evidence that in the wake of dramatic political change in State A, rivalries are most likely to form between A and its close partners.
Content may be subject to copyright.
Only Friends Can Betray You: International Rivalry and
Domestic Politics
Richard Saunders
Department of Political Science
Florida State University
rjs15d@fsu.edu
July 21, 2020
Abstract This article argues that dramatic political change in State A poses a threat
to the interests that other states B share with it. The more salient those interests are to
State B, the greater the threat posed by domestic political change within State A. Thus
major changes in one state place the leaders of formerly friendly states into a domain of
losses, motivating risk-seeking behavior in hopes of reversing these losses. Conversely, the
new leaders of the state undergoing domestic political change initiate similarly risk-seeking
behavior to defend their new endowments. The conflicts that result sow the seeds of long-
lasting enmity (rivalry) between former partners. I test this argument in a dataset of rivalry
onset during the period 1950-2005 and find evidence that in the wake of dramatic political
change in State A, rivalries are most likely to form between A and its close partners.
Word Count: 10981
Introduction
What causes rivalry? This should be a question importance to scholars of interstate
conflict. The persistence of three conflicts –in Iraq, Syria and Yemen – is frequently at-
tributed to activities of rivals Iran and Saudi Arabia. Similarly, much worry is expressed
by policymakers and pundits about the possible resurrection of the rivalry between Russia
and the West due to growing Russian influence in the Balkans and the Middle East. Finally,
growth in Chinese capabilities and renewed insistence from Beijing regarding Chinese control
of the Western Pacific threatens to intensify a number of rivalries in the Pacific. Given this
apparent resurgence in antagonism between old and new rivals alike, and the threat rivals
pose to global stability, understanding the way in which rivalries form should be one of the
primary goals of interstate conflict research. Rivalries make up only 1 to 5% of dyads, but
account for 50 to 80% of all interstate wars (Stinnet & Diehl 2001, Colaresi Rasler & Thomp-
son 2007). Unfortunately, research into rivalry formation has identified only few potential
causes –large systemic shocks, such as world war or major realignment of capabilities (Diehl
& Goertz 2001, Goertz & Diehl 1995); the existence of unresolved issues of contention, such
as territorial disputes (Vasquez & Leskiw 2001, Stinnet & Diehl 2001); or the prevalence of
power-politics behavior in international disputes (Valeriano 2013).
Importantly, research on the causes of rivalry has focused on explanations at the inter-
national level. These arguments fail to explain the emergence of rivalries like those between
the United States and Cuba or the U.S. and Iran, which were not accompanied by interna-
tional shocks or preexisting disputes. An explanation rooted in changing domestic politics
is necessary to explain these relationships. To date, relatively little work has been done to
understand the linkages between domestic politics and rivalry formation.1Rather, research
has turned away from the causes of rivalry to focus on the effects of rivalry (see Conrad
1An exception to this is the literature linking state independence and rivalry formation (Diehl & Goertz
2001, Stinnet & Diehl 2001, Goertz & Diehl 1995).
1
2011, Findley et al. 2012, Boutton 2014, Haynes 2015). This turn away from exploring the
causes of rivalry is surprising, especially given the lack of understanding of the domestic
influences on rivalry formation. The causes of any given rivalrous relationship should be
expected to influence the behavior of the rivals toward one another. It seems unreasonable
to think that we can adequately understand the effects of rivalry –the behaviors of rivals –
without a thorough understanding of why states become antagonistic toward one another in
the first place. Additionally, rivalrous behavior must, in large part, be driven by the inter-
play of domestic political interests within the two states engaged in rivalry. If researchers
are to develop a thorough understanding of either the causes or effects of rivalry, the effect
of domestic politics on rivalrous behavior can no longer be ignored. As such, this analysis
seeks to make a valuable contribution to our understanding of rivalry formation by more
fully exploring a largely unprobed link between political change within a state and the onset
of international rivalry.
In exploring this relationship between domestic politics and rivalry formation, I theorize
the existence of a conditional relationship between domestic political change in one state and
the interests that state shares with others. When two states share highly salient common
interests dramatic change in the domestic politics of one state increases the likelihood of
rivalry formation in that dyad. Put bluntly, major political changes can transform friendly
states into enemies. I argue that this is the case because changing domestic politics within
one state will threaten the interests shared between that state and others with which it
interacts, providing incentive for those other states to attempt to reassert the previous status
quo. This leads both states to view the other as ’enemy’, increasing suspicion of the other
state’s behavior in all areas and beginning a process of issue accumulation that culminates
in long-lasting enmity (Dreyer 2010a; 2010b).
In the remainder of this paper, I test this argument in a dataset of rivalry onset for the
period 1950-2005. I find that both revolution in a state and significant change in a leader’s
2
support coalition (SOLS change) are associated with an increase in the likelihood that state
enters into a rivalry within the next five years. Further analysis reveals that rivalry onset
is most likely to occur between a state undergoing changes to its domestic politics and its
friends in the international system –specifically those states B with which it shares a military
alliance, significant foreign policy preferences or significant trade. This finding illuminates
a previously unknown pathway to rivalry and demonstrates the importance of studying the
influence of domestic politics on rivalry formation.
Systemic and Domestic Shocks
Research on the causes of rivalry has focused on interstate level explanations, examining
the effect of large international shocks –world wars, the end of colonialism or the re-arranging
of the global balance of power –on rivalry formation. Goertz & Diehl (1995, 2001) argue
that these dramatic upheavals significantly alter the interests of states. Due to these altered
national interests, new sets of states come into conflict, while in other cases, previously
contentious issues become less salient. Thus large systemic shocks are expected to lead to
the end of some rivalries and the beginning of others (Diehl & Goertz 2001, Goertz & Diehl
1995). Further research has added that a stalemated dispute between two states, representing
a disagreement that is not sufficiently resolved by an initial round of conflict, is a primary
motivator of rivalry when both states in question continue to work toward victory (Stinnet &
Diehl 2001). Valeriano (2013) employs an international-level explanation of rivalry that does
not rely on system shocks, but in which rivalries grow from an initial dispute due to states’
application of power-politics strategies –alliance building, arms buildups, and escalation of
demands –as their primary tool in resolving these disputes.
While a literature exists studying the links between domestic politics and war (Davies
2002, Mansfield & Snyder 1995) relatively little research on the domestic political determi-
nants of rivalry exists. One exception can be found in the argument that domestic political
shocks, such as state birth, motivate the formation of many rivalries (Diehl & Goertz 2001,
3
Goertz & Diehl 1995, Stinnet & Diehl 2001). More recent research has gone on to provide
further support for this argument. Colaresi, Rasler and Thompson (2007 p.83) note that
roughly half of rivalries that include at least one state that gained independence after 1816
begin within a short period of that independence event. Conrad and Souva (2011) find evi-
dence to suggest that rivalry is unlikely to form between democratic states. Goertz & Diehl
(1995), however find no link between regime change and rivalry onset.
By focusing on changes to the international system to explain rivalry, most analysts fail to
account for the pre-existing relationships between states and how domestic political change
threatens these relationships. Prior to some form of domestic shock in one state, that state
and others are already engaged in various and diverse relationships. These heterogeneous
preexisting relationships set equally heterogeneous incentives that other states face when
responding to political shocks within the initial state. Thus, it seems necessary that the
effect of a large political change in one state upon its relationships with other states must
be conditional upon the character of those preexisting relationships. States that are enemies
may see their disputes resolved by a change in the domestic politics of one, as was found by
Bennett (1997; 1998). States that share deep interests with one another, however, should be
particularly effected by dramatic changes in the domestic politics of a close partner. I argue
below, that these deeply interconnected states are likely to move from a state of friendship
to one of rivalry due to major changes in the domestic political makeup of one.
The literature on revolution and civil war provides some additional insight on the role
of domestic political change in international conflict. Scholars in the realist tradition (Walt,
1996; Maoz 1989) argue that revolutions alter the balance of power and perceptions of ag-
gressive intent, creating a two-way pressure toward conflict. Revolutions often weaken the
revolutionary state and encouraging outside actors to prey upon it during its moment of
weakness (Walt, 1996, p. 331-333). The existence of these predatory incentives also con-
tribute to spirals of suspicion that motivate revolutionary leaders to wage preemptive wars in
4
hopes of holding off perceived external enemies while consolidating power (p. 334). A second
strain of research focuses on the characteristics of revolutionary leaders. Jeff Colgan (2010,
2012, 2013) argues that participation in a revolutionary movement selects for aggressive and
ambitious personalities, meaning that the leaders who are installed by revolutions will be
especially likely to initiate conflict against neighbors. Colgan and Weeks (2015) add to this
that revolution is especially likely to lead to paternalistic forms of government that cannot
constrain leaders aggressive ambitions. Gleditsch, Salehyan and Schultz (2008) point out
that civil wars can lead to international conflict when outside states intervene to influence
the outcome of the conflict.2While each of these arguments posits a link between revolution
and international conflict, they imply that revolutionary regimes should be equally likely to
come into conflict with any of their neighbors. They provide little understanding of why
some relationships with post-revolutionary regimes turn hostile while others do not.
Domestic Political Change & Policy Change
In the following section, I lay out my argument regarding the effect of changing domestic
politics in one state –State A –upon the relationship between that state and others –States
B. Domestic political change refers to change in the policy preferences of the support coali-
tion upon which a state’s leader relies to maintain office. This change can occur in two
general ways. It is possible that the composition of the support coalition remains unchanged
but that the policy preferences of its members change in response to the natural evolution
of problems and opportunities facing a society. This sort of domestic political change is
relatively commonplace, occurring within most regimes as time progresses. A second form
of domestic political change is that which brings new elements of society and their interests
into the support coalition while excluding other groups that previously held power. I argue
that changes to the composition of the support coalition represent a more significant level
2As discussed in the empirical section below, I explicitly rule out that intervention in civil war drives the
findings of this analysis by dropping all cases in which one state in a dyad intervenes in the other state’s
civil war.
5
of change in aggregate preferences than does incremental creep.
I use the term radical domestic political change to refer to domestically driven events
that lead to a near complete replacement of one support coalition by another. This can
include coups, revolutions and other forms of sweeping government replacement. Changes
to the structure of domestic politics are likely not, of themselves, sufficient to cause rivalry
with a neighbor. If a new regime were to come to power and follow precisely the policies
of the previous regime, there is no ex-ante reason to expect this regime change to cause
conflict3. However, radical domestic political changes are likely to result in actual change
in foreign and domestic policies pursued by the state. These radical changes wipe away the
former political structure and bring an entirely new support coalition to power that holds
a different set of preferences as compared to the previous coalition, thus providing a new
leader with significant incentives to reform the policies followed by the state.
Policy Change and the Ex-Ante Relationship
I argue that domestic political change drives change in policies pursued by states, and
because of this, can motivate states to engage in rival behavior when they did not previously.
Whether a given state’s change of domestic politics serves to motivate rivalry is dependent
upon the ex-ante relationships between that State A and those States B with which it
interacts. When two states are "friendly," –when states hold many trade, security or other
interests in one another –sufficiently large changes in the domestic politics of one state of the
pair establish an expectation that major policy changes will follow, which may be a threat
to those interests that the opposite state possesses in it.
In such an environment, domestic political actors in State B who hold economic interests
in State A will feel these interests to be under threat. Given these fears, if even minor
detrimental policy change follows the change in State A’s domestic politics, these elites will
3Thyne and co-authors provide an example. They attribute the stability of the Egypt-Israeli relationship
following Egypt’s 2013 coup to the immediate pledge by coup-leaders to maintain previous defense policies
(Thyne et al. 2018).
6
pressure their own government to intervene to protect their interests at this first sign of
danger. Similarly, major political change in a state that is important to the opposite state’s
national security is likely to cause publics in State B to experience an increase in perceptions
of threat and insecurity. This will motivate calls for the government to intervene to protect
the security of its people. Thus, the leadership of State B will fear a loss of domestic support
if they fail to take action to restore the status quo.
For their part, the leadership of State A prioritizes maintaining a grip on their recently-
won power. This requires that the leadership maintain its domestic support coalition. To
the extent that the support coalition backing the new leadership is different from that of
the old leadership, we would expect that the new leaders will be required to respond to a
different set of preferences. Thus, a leader’s desire to change policy will be determined by the
difference between their supporting coalition and the previous leader’s coalition. Further,
having recently supplanted a leader or regime that failed to cater to the demands of society
should serve to make the new leadership wary of suffering the same fate. That one leader or
regime has been recently replaced suggests that important elements of society desire major
change in government policies. If the new leadership fails to provide this policy change by
acquiescing to foreign demands, this allows room for other members of the reform movement
or counter-revolutionary elements to replace the new leadership before it can consolidate its
hold on power.
Domestic Political Change and Rivalry
Prospect theory, formulated by Kahneman and Tversky (1979), provides an understand-
ing as to how domestic political change ultimately leads states into rivalry. According to
prospect theory, actors make choices in a process consisting of two phases.4In the first edit-
ing phase, actors organize and simplify the available options for later evaluation. Framing
effects, are argued to significantly affect the succeeding evaluation process. While Kahneman
4See Levy, 1992; 1996 for a thorough discussion of prospect theory and its implications for IR research.
7
and Tversky discuss a number of specific framing effects that occur during the editing pro-
cess, for purposes of this discussion, the most relevant aspect of the editing phase is coding
of the location of an actor’s reference point. Decision-makers define the value of outcomes
in terms of gains or losses from a reference point –generally the status quo –rather than in
terms of the absolute value of the outcome.
Once the available options have been edited, an evaluation phase begins in which the
decision-maker selects the option with the highest value as determined by the product of
avalue function and a probability weighting function. These functions combine to provide
several expectations about decision-making under conditions of risk. Most important of
these for purposes of this analysis is an observed tendency for actors to be risk-seeking in a
domain of losses and risk averse in a domain of gains (Tversky & Kahneman 1992). When
contemplating certain losses, actors are likely to over-estimate the value of a risky gamble
that offers only a moderate chance of reversing those certain losses, even if the gamble risks
significant additional losses.
The behavioral expectations provided by prospect theory serve to explain why domestic
political change in a "friendly" dyad will lead to conflict5. Following radical political change
in a friendly state, State B is likely to suffer a loss in terms of economic interests, alliance
benefits6, cooperation in foreign policy, or all of the above. This policy change harms the
leadership of State B in two ways. First, the loss of an allied government or economic partner
damages the security interests and prestige of State B in the international environment,
placing the leadership of State B in a domain of losses in foreign affairs. Similarly, loss of an
ally or trade partner harms the interests of domestic political and economic elites in State
5While the usefulness of prospect theory for explaining political phenomena has been questioned (Levy
1997), it has been usefully applied to choices made by leaders in a number of crisis situations such as those
of Kennedy and Khrushchev during the Cuban missile crisis (Haas 2001), Carter during the Iranian hostage
crisis (McDermott 1992) and Bush during the run-up to the 1991 Gulf War (McDermott & Kugler 2001).
The arguments put forward in this analysis are similarly centered around crisis decisions made by state
leaders, thus prospect theory is applicable and provides for interesting insights.
6see Morrow (1991) for a discussion of alliance benefits.
8
B, which drive demands for action by the leader of State B and threatens to undermine the
leader’s tenure should demands not be met. That the loss of an ally or economic partner
can place the leader of a state in a domain of losses across a number of dimensions is vital to
a prospect-theory centered explanation of conflict. The losses suffered by a leader must be
sufficiently grave as to make him or her unwilling to accept the new status quo as the reference
point from which deviations will be evaluated (McDermott 1992). Given these losses, we
should expect the leadership of State B to prioritize re-establishing the previous status quo,
even if this means undertaking risky behaviors that threaten significant additional losses, so
long as these behaviors also provide a reasonable chance to avoid paying the certain cost of
accepting a revised status quo. Practically, this loss-aversion manifests itself in actions such
as supporting counter-revolutionaries (as in the Bay of Pigs invasion), aiding third parties in
trying to overthrow the new regime (as with the significant Western support for Iraq during
the Iraq-Iran war), or direct military action. These attempts at loss aversion establish enmity
between the states and cement enemy images in the minds of policymakers and the populace
of both states. Once this perception of hostility on the part of the other has taken root,
issue disputes tend to accumulate between the two states, cementing rivalry (Dreyer 2010).
An example of the process described above can be seen in the relationship between the
United States and Cuba around the Cuban revolution. Before the revolution, the U.S. and
Cuba shared a defensive alliance through the Organization of American States. They also
participated in a strong trading relationship in which American nationals possessed a great
deal of property within Cuba. Estimates suggest that as much as 65% of the Cuban sugar
industry was under U.S. ownership and some 95% of Cuban exports were bound for the
U.S. market (Fitzgibbon & Thomas 1972 p. 335-337). In the immediate wake of the Cuban
Revolution, these economic ties remained intact. However, the policy goals of the Castro
regime differed significantly from those of the Batista government, ensuring that a period of
economic and social upheaval would follow the success of the Cuban Revolution.
9
American property interests came under threat in May of 1959 when Fidel Castro began
to implement a policy of land reform that would result in the seizure of much American-owned
property (Snyder 1999 p. 273). Further, throughout the first few years of the reform process,
some 600 American firms were nationalized (Snyder 1999 p. 276). The Cuban Revolution
inflicted humiliation and economic losses on the United States through the removal of the
Batista regime with which the U.S. government had close security ties and also throughout
the following several years as the reform process increasingly put U.S. economic interests in
Cuba at risk.
I contend that it was the threat posed to U.S. interests by Castro’s reforms that sowed
the seeds of rivalry between the U.S. and Cuba. The Cuban land reforms of 1959 motivated
President Eisenhower, in March of 1960, to authorize agents of the Central Intelligence
Agency to begin arming and training Cuban exiles to build a paramilitary invasion force.
This represents a very risky endeavor that, if failed, would cause foreign and domestic polit-
ical losses for the president, but if successful would restore the pre-existing status quo. This
risky decision would ultimately culminate in the failed Bay of Pigs invasion in April 1961
and would go on to motivate further hostility between the two states. Not until December 2,
1961 – after the beginning of Cuba/U.S. hostilities –did Castro align with the Soviet Union
and declare Cuba a Communist country (Bain, 2011 p.112). Thus, while Cuba’s alliance
with the Soviets certainly sustained the rivalry between Cuba and America by making it a
part of the Cold War competition, the order of events suggests that the American threat to
Cuba caused Cuba to seek out Soviet protection, rather than the other way around. Once
this initial hostility took root in the minds of the people and policy-makers of both countries,
every action of the other state was viewed through a lens of suspicion, and issues of con-
tention began to accumulate quickly over ideological and positional concerns (Dreyer 2010a).
Long-lasting enmity was the result.
I argue that the initiation of rivalry between Cuba and the United States is in no way
10
a unique occurrence. Rather, I argue that the U.S./Cuba relationship is illustrative of the
process that a number of other dyads have undergone on their way to a rivalrous relationship.
Tables 3 and 4 display two lists of rivalries that were initiated within 5 rears of a major
domestic political change as well as a summary of the major ex-ante interest shared between
those states.7Between 23% and 31% of all rivalries initiated during the 1950-2005 time
period seem to follow the pattern of the U.S./Cuba example.8
Given the expectations discussed above, I derive the following hypotheses regarding the
conditional effect of shared interests between states on radical policy change, which I then
test empirically.
H1: The effect of radical policy change upon the probability of rivalry formation in the
dyad should be zero in the absence of shared interests between states.
H2: The effect of radical policy change upon the probability of rivalry formation should
be positive in the presence of significant shared interests between two states.
Theories rooted in rational choice have difficulty explaining this conditional relationship.
From a bargaining perspective, rivalry onset is akin to conflict onset, thus there must be
a mechanism that triggers bargaining failure. A change in domestic politics or policy pref-
erences does not lead to a large and rapid shift in power or other credible commitment
problem. To the extent that these changes could exacerbate information problems, there is
little reason to think this would be more likely between former allies than between formerly
neutral states. Similarly, while theories rooted in expected utility might suggest that these
changes bring new actors to power who may then become a security threat to other states,
7In Table 3, this radical domestic political change is a revolution. Table 4 displays rivalries following
within 5 years of a change in source of leader support as drawn from the CHISOLS dataset.
8I note that, at least in the time period under analysis, it appears that the results presented below
are driven by a relative handful of particularly belligerent states. Egypt, Libya, Iran, Uganda, the United
States, Sudan and Saudi Arabia all become involved in multiple rivalries during this period, with Iran being
especially prevalent. Given this, I include Table 18 in the appendix in which I replicate the main findings of
this paper in a sample where I have dropped all dyads containing Iran. The results of the primary analysis
hold in this sample.
11
this perspective cannot explain why the new threat operates in the conditional manner I
have hypothesized. The rational response to revolution in a state would seem to be one of
guarded neutrality meant to avoid the additional costs of conflict. Only an argument based
on loss aversion is able to explain how large changes of domestic politics push two formerly
friendly nations past neutrality into enmity.
Radical Change and Outside Intervention
If future change in policy within State A threatens the interests of the opposing state,
why does State B allow the domestic political change in the first place? Why does B not act
to preserve the initial government of A, thus preserving the ex-ante status quo? In many
cases, this is exactly what does occur. Witness the long history of outside interventions by
one state to prop up the government of another (Butler 2003, Saideman 2001, Pearson 1974).
There are, however, at least five reasons for which a state may not intervene in another state
to preserve its own interests.
First, states may possess insufficient capabilities to ensure victory to an allied government
in their civil war. Second, when states have the capability to support an allied government,
third party support of the rebels may still lead to the downfall of the allied government.
In either of these cases, State B’s inability to prevent a change of government in A does
not imply that B prefers that this change occur. While B may not be able to remove the
new government of A, hostility between the two is likely.
Third, potential intervening states may lack the political will to support a friendly gov-
ernment even when they possess the capabilities. A leader may face negative domestic conse-
quences for intervening in the allied state that are sufficient to make intervention politically
unacceptable.
Fourth, it is also possible that militarily potent third parties miscalculate the cost of
regime change in the ally. A rebel movement fighting against an allied state can be expected
to hold an ideology that is unfavorable to the third party. However, international relations
12
often make for strange bedfellows. There is no immutable law that the third party could not
work to achieve good relations with a victorious rebel group. Thus the third party may not
see the rebel group as a sufficient threat to their interests, only to realize their error after the
group assumes power and begins implementing its policy reforms. Rebel and revolutionary
groups are aware that allies of the regime must make this calculation when contemplating
an intervention to support the regime, and sometimes work to conceal their true ideological
stance or the extent of their policy goals to induce uncertainty over the cost that a third
party will pay for failing to intervene (Werner 2000). However, when the victorious rebel
group implements its preferred set of policy reforms, this uncertainty is resolved. To the
extent that these new policies damage the interests of the third party, conflict is likely to
result.
Fifth, and finally, it is entirely possible that State B does not appreciate that major
changes are occurring in State A. Not all radical changes are the result of protracted civil
war. The South Korean April Revolution of 1960 spanned only 14 days from the outbreak
of protests on April 11 to the resignation of Syngman Rhee on April 26 (Kim & Kim, 1964).
Most states would be hard pressed to recognize that the protest movement represented a
genuine threat to the regime and to then mount a major military operation on a timeline of
14 days.
In any of these cases we should see that State B does not act to preserve the ex-ante
status quo even though its interests are at risk. However, once radical change has come to
pass, State B will suffer damage to the interests that it shared with State A and will enter
a domain of losses as discussed in the previous section.
Research Design
Unit of Analysis and Scope
I construct a dataset consisting of dyad-year observations during the period 1950-2005
13
for all politically relevant dyads in which both states have population 500,000 or more.9The
argument presented above centers around rivalry that stems from dramatic political change
within an existing state rather than rivalry stemming from state birth. Thus, I drop all
dyad-years from the sample in which one or both members of the dyad are new entrants
to the system –along with the attendant rivalry-onsets. This procedure should ensure that
findings presented below result only from dramatic political change within existing states
rather than from state birth. This provides a sample of 43,144 observations.10 I fit a series
of logistic regression models on this sample, which are presented in the next section.11
Operationalization
Rivalry Onset: I adopt a conceptualization of rivalry put forward by William Thompson
(Thompson 1995; Thompson & Dreyer 2011). Thompson’s conceptualization emphasizes
that rivals must mutually believe themselves and their interests to be under threat due to
the existence and activities of the other state (Thompson & Dreyer 2011, p.2). Thus, while
the existence of a hostile relationship between states is necessary to the formation of rivalry
in this conceptualization, there exists no set threshold of militarized action that defines the
start-point of a rivalry. Rather, Thompson examines historical records and primary sources
for evidence that policymakers in both states in a dyad view the other as threatening, and
that both states believe a military confrontation to be likely, and thus take steps to prepare
for conflict (Thompson & Dreyer 2011 p.3).
The dependent variable is the initiation of rivalry within a dyad. The variable Rivalry
Onset is coded using the inventory of rivalries developed by Thompson and Dreyer (2011),
and takes the value "1" in the dyad-year in which a rivalry was coded to have begun, and
9Lags of trade data begin in 1950 and lags of the main revolution variable run out in 2005.
106,271 observations are lost due to missing data on Joint U.N. voting. 19 observations are lost due to
missing V-Dem data.
11In the appendix, I include a replication of this main analysis using the Rare Events Logit (King & Zeng,
2001). Coefficients for the three interaction terms remain positive in all cases. The interaction between
trade dependence and revolution, however, fails to achieve statistical significance at conventional levels in
the RE-Logit.
14
"0" otherwise.12 There are 55 cases of rivalry initiation in the sample under investigation.
Observations in which rivalry is ongoing are dropped from the dataset because no new
rivalry-onset is possible.
In 10 cases, a rivalry is coded to have begun in the same year a change in domestic
politics is coded to have occurred. In these instances I consult the narratives constructed by
Thompson and Dreyer to determine which came first: political change or rivalry. To ensure
temporal precedence, I examine the narratives for these 10 rivalries to see if Thompson and
Dreyer make a note that the state not undergoing revolution intervened in any way during
the revolutionary event or civil war leading to regime change. If Thompson and Dreyer note
that one state intervened in another prior to both the political change and the formation of
the rivalry, the dependent variable is coded "0" to limit the possibility that cases of reverse
causality influence the findings of this analysis. In three of the 10 cases so examined, the
second state in the dyad intervened in the other prior to the political change event. Thus,
for these 3 cases, Rivalry Initiation is coded "0".
Measuring Domestic Political Change: I employ two measures of domestic political
change in this analysis. In the main analysis I use a measure derived from Jeff Colgan’s
(2012) revolutionary leader dataset as an indicator of radical political change within one
state. To determine whether a revolution has occurred, Colgan begins with ARCHIGOS
leader transitions (Goemans et al., 2009), then consults historical records for evidence of
change within seven different policy domains. These seven policy areas include measures of
changes to executive selection and power; property ownership laws; the relationship between
state and religion; the official ideology of the state; protections for ethnicity or gender; the
official state name; or the establishment of a revolutionary council or committee (Colgan
2012). The variable Revolution is coded a "1" for a dyad in any year in which a new leader
12For purposes of robustness testing, I also fit a series of models on a dataset including only rivalries that
appear in both Thompson & Dreyer’s 2011 dataset and Klein, Goertz & Diehl’s 2006 inventory. Results
follow those of the main analysis and can be found in Table 16 in the online appendix.
15
comes to power in at least one of the two states who institutes major policy changes in at
least three of these seven categories.13 This measure is useful for the current analysis be-
cause it accounts for not only change in regime or sources of leader support, but also directly
measures the extent of subsequent change in domestic policies.
Note that, while this variable is coded using Colgan’s 2012 data, the variable constructed
for this analysis is not identical to Colgan’s variable. Colgan’s measure covers the entire
tenure of a leader –thus taking the value "1" for any year in which that leader remains in
office (Colgan 2010 p.675). I am interested only in the initiation of major political change
in a state, thus I focus on the year in which a new leader comes to power. Further, I
am concerned primarily with the initial revolutionary event rather than the behavior of
revolutionary leaders. Thus, I also do not include the start of subsequent terms of office of
a revolutionary leader, nor do I include normal transitions from one leader of the revolution
to another. I construct a variable that takes the value "1" only for the first year in a leader’s
tenure. Thus, Colgan’s radical ideology variable takes the value "1" for Chile for years 1973-
1989 (Pinochet’s tenure), the Revolution variable employed in this analysis takes the value
"1" for Chile only in 1973. Further, Colgan’s measure includes Ferrer’s second term in Costa
Rica (starting in 1953) when I do not. Thus Costa Rica is coded, in my data, as undergoing
a revolution only at the start of Ferrer’s first term in 1948.
A second measure of political change is used in robustness tests. This variable is coded
using the Change in Source of Leader Support (CHISOLS) dataset (Mattes, Leeds & Mat-
sumura 2016). SOLS Change takes the value of "1" for any dyad-year in which one state is
coded in CHISOLS to have undergone a change in the source of leader support that did not
13While the selection of a threshold of three areas of policy change is arbitrary, Tables 12-14 in the appendix
display the robustness of the findings discussed below to the use of alternate thresholds.
16
result from a peaceful democratic election.14
Between 1950 and 2005, there are 80 Revolutions and 330 instances of SOLS change in
my sample.
Shared Interests: The argument presented above centers on the behavior of states that
share a close relationship following a major domestic change in one. I use three measures of
shared interest. First, I use the existence of a defense pact between two states in a dyad.
The signing of a defense pact between two states is an observable signal on behalf of each
state that it possesses an interest in preserving the status quo in the other. Usefully, this
interest in preserving the allied state may result from a number of factors such as defense
concerns, economic interdependence, or shared ideology. We should expect that the existence
of a publicly declared defense pact is an observable indicator of a range of shared interests
that may be put at risk due to changes in the domestic policy of the allied state. Further,
it seems reasonable to expect that states will only act upon interests that their leaders or
powerful domestic groups recognize as salient, thus, using the declaration of a defense pact
–a clear signal of a recognition of important shared interests –provides for an appropriate
test of the hypothesis. The variable Al liance is coded using the Alliance Treaty Obligations
and Provisions (ATOP) dataset (Leeds, Ritter & Mitchell 2002). This variable takes the
value "1" for any year in which a defensive alliance exists between the states in a dyad and
"0" otherwise.
I also employ a measure of shared interests based upon trade ties. To measure Trade
Dependence, I use data taken from the Gledditsch (2002) "Expanded Trade and GDP Data."
Trade Dependence is a continuous variable constructed using the weak-link procedure. For
14SOLS changes resulting from peaceful democratic election indicate only a change in the political party
holding the executive office. The theory offered above claims that major political change is necessary to
trigger rivalry initiation. It seems unlikely that a change as commonplace as the peaceful hand-off of power
from one party to another would represent a radical change in domestic politics. Thus, to ensure that this
measure matches the concept discussed above, peaceful electoral transitions are coded a "0" in this measure
of policy change.
17
each State A in a dyad I sum the total value of imports to and exports from State B, and
then divide this sum by State A’s GDP, thus creating a ratio of State A’s trade with State
B to State A’s total domestic output.15 I repeat this procedure reversing states A and B,
then keep the lowest value of the two –the weak link. This raw measure ranges between 0
and 0.21. Thus, I normalize the variable between 0 and 1 to make coefficient estimates more
easily comparable across variables.
Third, I use a measure of shared foreign policy interests that is derived from roll call
votes in the United Nations General Assembly (Bailey, Strezhnev and Voeten 2017). The
source dataset provides a spatial measure of voting in which each value represents a state’s
ideal point. Thus, as two state’s values draw closer together this represent more congruent
foreign policy preferences. To construct my measure of foreign policy similarity, I first take
the absolute difference between dyad members’ ideal points. I then normalize these values
between 0 and 1 to make estimates more easily comparable across variables. Finally, I
subtract the resulting value from 1 to invert it. This procedure results in a continuous
variable such that a value of "1" represents the maximum and "0" the minimum congruence
between two states’ U.N. General Assembly voting record.
Finally, I also use these measures of shared interest to construct a latent variable Shared
Interests to more thoroughly account for the multiple aspects of states’ shared interests in
a single model and also to better account for unobservable aspects of shared interests. This
latent variable is built using linear structural equation modeling and ranges between0.46
and 0.72 with mean 0.000 and standard deviation 0.14. This latent variable has high face
validity. Among the 50 relationships with the fewest shared interests, all are observations
that include the Unites States and a Warsaw Pact member state, and all occur during the
1950’s, at the height of the Cold War. Among the 50 most aligned observations, 41 are
observations of the relationship between Belgium and the Netherlands –arguably the closest
15This is expressed as I mportBtoA +E xportAtoB
GDPA.
18
interstate relationship in the world. The remaining 9 most-aligned observations are split
between Belgium-Luxembourg and France-Germany observations during the 1980’s. This
measure also tracks changes in shared interests well. The U.S.-Iran relationship in 1978
prior to the Iranian revolution, for example, scores above the 90th percentile in terms of
shared interests. In 1980, one year after the revolution, U.S.-Iran shared interests have fallen
to the bottom 5 percent.
Interaction Terms: Finally, I code interactions between the aforementioned measures of
shared interests and the measures of policy change discussed above. For purposes of this
analysis, I consider a five-year period of time following political change within one state in
a dyad. I expect that a new leader who comes to power with the hope of implementing a
radical policy program will have managed to consolidate power and implement a sufficiently
large subset of the planned reforms for other states to gauge their ultimate extent. Thus, it
is during this five-year period that other states would judge the threats to their interests, and
I would expect to see any conflict motivated by the revolution to occur within this period
of time. As such, the primary independent variables of interest –the interactions between
measures of shared interests and measures of policy change –take a non-zero value during
the five years t1t5following a policy change event, representing the five-year transition
period after a change in policy t0, and "0" for all other observations.16
I also employ a number of control variables in hopes of mitigating the possibility of
spuriousness or omitted variable bias influencing the findings. All covariates are lagged 6
years to ensure that no covariate value for any year during a five-year observation window
is taken from a post-revolutionary time period.
Contiguity: Contiguity of borders is known to contribute to the likelihood of conflict,
including rivalry (Stinnet & Diehl 2001, Lemke & Reed 2001) and should be expected to
16In robustness testing I examine the effect of using observation windows of different lengths. Tables 7-9
in the appendix display the results of analyses using observation windows of 2-8 years.
19
influence alliance formation and trade as well. Further, controlling for contiguity has been
shown to be necessary for avoiding omitted variable bias in studies positing a positive as-
sociation between alliances and conflict behavior (Ray 1995). This variable is coded "1" if
the states in a dyad share a land or river border according to the Correlates of War Direct
Contiguity V3.2 dataset (Stinnett et al. 2002), and "0" otherwise.
Military Capability Disparity: Thompson and Dreyer argue explicitly that some level of
competitiveness in military capabilities is necessary to rivalry formation (2011 p.3). Addi-
tionally, evidence suggests that the balance of capabilities affects the likelihood that defense
pacts are signed between states (Walt 1985). Thus, as with shared borders, we must control
for the distribution of military capabilities in a dyad to reduce the likelihood that any effects
found in our analysis are spurious.
To account for military capabilities, I construct a variable using the Composite Indicator
of National Capabilities (CINC) (Singer, Bremer & Stuckey 1972). This variable takes a
value equal to the greater of the two capability scores in the dyad divided by the sum of
both capability scores.17 This measure provides a ratio of the military capabilities possessed
by the more powerful state in the dyad as compared to the total military capability present
in the dyad (Gleditsch, et al. 2008).
Joint Democracy As with the other controls discussed above, the level of jointly demo-
cratic institutions in a dyad have been shown to influence the interests shared between
states (Leeds, Ritter & Long 2002), as well as the likelihood of rivalry formation in the dyad
(Conrad & Souva 2011), and thus must be controlled for.
Democracy is constructed using the V-Dem Electoral Democracy Index (Coppedge et al.
2018) and is a continuous variable ranging between 0 and 1. I use a weak-link procedure,
taking the lesser of the two Electoral Democracy Index scores in the dyad as a measure of
the overall level of democracy in the dyad.
17This is expressed as MaxC apability
MaxC apability+M inCapability .
20
Systemic Shock Shocks to the international order, such as world wars, decolonization and
the end of the Cold War have been shown to re-balance power between states and contribute
to rivalry onset (Rider & Owsiak, 2015; Diehl, 1995). I control for systemic shocks using
the procedure introduced by Diehl 1995. This variable takes the value "1" for a 10-year span
following the end of the Cold War or any significant changes in the territory of major powers.
I draw the specific measure used in this analysis from Rider and Owsiak’s (2015) replication
data due to it’s longer temporal span.
Ongoing Territorial Dispute The existence of ongoing territorial disputes has commonly
been found to contribute to rivalry (Rider & Owsiak, 2015; Vasquez, 2009; Hensel, 2001). I
control for the existence of an ongoing territorial dispute using data drawn from the Issue
Correlates of War (ICOW) (Frederick et al., 2017).
Civil War The existence of a civil war prior to revolution allows for the possibility that
an external state provides support to the government forces and thus effectively initiates
conflict with a rebel group before it gains power. Thus, to prevent the possibility that
external support for the regime explains findings regarding domestic political change and
rivalry onset, I control for the occurrence of a civil war antecedent to revolution or SOLS
change. I take data on civil war from the Correlates of War Intrastate Wars dataset V4.0
(Dixon & Sarkees, 2015).
I also include a control for the Cold War which takes the value "1" for years 1950-1989.
The Cold War was a time period in which both rivalry across blocs, and alliances and
alignment of policy and trade within blocks was the rule.
Finally, I control for the temporally dependent nature of conflict by using a cubic poly-
nomial of the years since the end of the last rivalry in a dyad as suggested by Carter and
Signorino (2010).
Results
I begin testing the argument above by first assessing the relationship between revolution
21
and rivalry onset. Table 1 displays the simple bi-variate relationship between a dyad’s
experience of a revolution within the last five years and subsequent onset of rivalry. The
table displays observed frequencies without parentheses as well as the frequency that we
would expect to observe if there were no systematic relationship (inside parentheses). In this
bi-variate analysis we note a strongly significant relationship between revolution and rivalry.
Just over three times as many rivalries begin in the immediate wake of a revolution in one of
the states than would be expected by chance.18 Further, nearly one quarter of rivalries (24%)
that begin during the 1950-2005 time period do so within 5 years of a revolution within one
state in the rival pair.19 20 This relationship is worth noting on its own. It conforms with
previous research on rivalry in that it shows that major domestic political shocks seem to
be associated with subsequent rivalry onset(Goertz & Diehl 1995). However, where previous
studies have only found support that state birth or independence (Goertz & Diehl 1995,
Stinnet & Diehl 2001, Colaresi et al. 2007) but not regime change (Goertz & Diehl 1995)
have an effect on rivalry, the current study demonstrates that major changes to the structure
of domestic politics within an existing state can also provide an important spark to ignite
rivalry.
————
Table 1 about here
————
Table 2 displays the results of a series of logistic regression models that test the hypothesis
discussed above. Robust standard errors are clustered at the dyad level. Figures 1 through 6
display average marginal effects for the three interactions of interest. Model one demonstrates
the correlation between revolution and rivalry onset. Models 2-4 display the interactive effect
18Table 5 in the appendix displays a similarly significant relationship between CHISOLS SOLS changes
and revolution. 83% more rivalries form in the wake of a SOLS change than would be expected by chance.
19A list of rivalries that begin within 5 years of a revolution can be found in Table 3.
20For SOLS change, the figure stands at 31% of rivalries forming within 5 years.
22
of revolution conditional upon the three measures of shared interests between states that are
discussed above. .
————
Table 2 about here
————
As we would expect given the bi-variate correlations seen in Table 1, Model 1 indicates
that there remains a positive relationship between revolution within the last five years and the
beginning of a subsequent rivalry even while controlling for other commonly theorized causes.
However, as predicted in the hypothesis presented above, we see that this relationship is
reduced in magnitude and statistical significance and may disappear entirely when accounting
for the effect of shared interests in a dyad. In Model 2, Revolution takes on a negative sign
and approaches significance (P>|Z| = 0.13). The models including individual measures of
shared interest provide mixed support for the hypothesis that revolutions should not lead
to rivalry. However, Model 5 presents the findings when modeling the effect of unobservable
latent interests on rivalry formation. In Model 5 we see the predicted relationships between
revolutionary change, shared interests and rivalry onset. When accounting for this more
thorough latent measure of shared interests, revolution has no discernible independent effect
on rivalry formation. The interaction of latent shared interests and revolution, however,
takes a positive and strongly significant coefficient. This effect is equivalent to a 10-fold
increase in the probability of rivalry onset over the baseline. This seems to indicate that
while revolution is positively associated with rivalry onset, these post-revolutionary rivalries
do not form at random. Instead, they are determined by the ex-ante relationship between
the revolutionary state and the friends and allies of the old regime. Given the findings
presented here, it may only be when two states share close ties ex-ante that the occurrence
23
of a revolution will drive them into conflict ex-post.21 This finding provides support for both
hypotheses discussed in section 2. Revolutionary states appear to find themselves drawn into
conflict because revolution threatens to destroy existing relationships between states, thus
inflicts losses upon those states that share valuable ties with the revolutionary state. This
begins an intractable antagonistic relationship between the new regime and states aligned
with the old regime giving birth to rivalry.
It should be noted that, while marginal effect sizes are small in absolute value, they
represent a substantive increase over the base probability of rivalry onset in the sample
(0.001). Figure 1 displays that a revolution in a dyad that is maximally aligned in terms of
foreign policy interests leads to a six-fold increase in the probability of onset as compared
to a dyad that is minimally aligned (an increase of 0.0066).22 Similarly, revolution in one
state of an allied pair leads to an increase of double the base probability of rivalry onset
after revolution as compared to a non-allied dyad (an increase of 0.0023). This effect is even
more pronounced when observing the marginal effects associated with the latent Shared
Interest variable –displayed in Figure 2. In this case, movement from minimum to the 99th
percentile of shared interest23 is associate with an increase of an order of magnitude (0.011) in
the probability of rivalry onset across those values distinguishable from 0.24 Further, these
marginal effects are larger than those associated with the occurrence of a systemic shock
(0.00025) and are of a similar magnitude to the effect associated with an ongoing territorial
21The models presented here are insufficient to assess whether high levels of shared interest are a necessary
condition for revolution to result in rivalry. I include in the appendix to this paper a test of necessary
conditions devised by Braumoeller and Goertz (2000) as well as a discussion of the findings. It seems likely
that high levels of shared interests are indeed necessary for domestic political shocks like revolution or regime
change to lead to rivalry onset.
22Moving from the first to third quartile in foreign policy similarity reduces the effect to a doubling of the
base probability or 0.0023, similar to the effect size found for an alliance.
23As discussed above this measure ranges between 0.46 and 0.72 with mean 0.000 and standard deviation
0.144, however, 99% of values fall below .4. Thus, making predictions for values of Shared Interest above the
99th percentile would be irresponsible due to lack of data to support prediction over a relatively large range
of values.
24When moving from first to third quartile, an increase of 140% the base rate (0.0014) is seen.
24
dispute (0.0069)25 –two of the most widely noted causes of interstate rivalry (Rider & Owsiak
2015, Colaresi et al., 2007; Hensel, 2001; Thompson, 2001; Vasquez & Leskiw, 2001; Stinnet
& Diehl, 2001; Lemke & Reed, 2001; Goertz & Diehl, 1995).
—————–
Figures 1-2 about here
—————–
While rivalry onsets are rare, rivalries account for a large majority of interstate wars.
Additionally, the extreme competition between rivals such as Saudi Arabia and Iran over
regional influence can destabilize entire regions. This competitive behavior between rivals can
ignite and exacerbate interstate conflict between the rivals themselves and can contribute to
both interstate conflict in the region, as well as intrastate conflict in nearby states. Thus, the
extreme consequences associated with rivalry ensure that it cannot be ignored as a subject
of study, despite its rarity.
Discussion
The findings presented above have several interesting implications for research on rivalry
as well as research on the conflict-proneness of revolutionary states. First, revolution is one
clear pathway to rivalry that has not been adequately explored by previous analyses that
focus primarily on systemic shocks and state birth. Major shifts in the domestic political
landscape of a state have important implications for rivalry formation. Future research
should seek to determine the extent to which rivalries born from revolution are comparable
to other rivalries in terms of behavior and other characteristics, and also how significant
the extent of political change within a state must be before it motivates outside powers to
intervene.
25Both of these are marginal effect sizes derived from Model 1
25
Second, previous research has offered a number of explanations for the empirical regularity
that post-revolutionary states seem to engage in conflict at a disproportionate level. The
finding presented here casts doubt on some of these arguments and complicates others. Post-
revolutionary conflict does not seem to be driven by opportunistic predatory behavior on the
part of either the revolutionary state or enemy states hoping to capitalize on the moment
of weakness that occurs due to revolution (Maoz 1989). Rather, it seems to be driven by
disruptions of pre-revolutionary interdependent relationships. Additionally, if the process
of fighting a revolution selects for aggressive risk-seeking personalities in post-revolutionary
leaders (Colgan 2010), it also seems that it provides those leaders with incentives to engage in
aggression specifically toward friends and allies of the old regime rather than opportunistic
aggression toward weaker targets of opportunity. Further explanation of the motivations
behind a given revolution may provide more clarity as to why this is the case.
Conclusion
In recent years, attempts to understand rivalry appear to have turned away from exam-
ining the causes of these particularly dangerous relationships. Instead, most research into
interstate rivalry now appears to focus on the ways in which rivalries influence other aspects
of the international system. I argue that this is problematic. Attempting to explore the effect
of rivalry without first understanding its causes will contribute to significant theoretical and
empirical problems. Most research designs require the scholar to have a clear understand-
ing of the underlying process that generates observed data sufficiently well as to be able to
identify and account for spurious relationships between the independent variable –existence
of a rivalry–and our dependent variable –the behavior of rivals. Without developing a better
understanding of the causes of rivalry, it is difficult to believe that we can achieve this.
In this paper I have attempted to address our collective lack of understanding of the
causes of rivalry, at least in part, by deriving and testing a new argument regarding the root
causes of rivalry formation. As discussed above, this argument centers around the general
26
logic that, when one state undergoes extreme changes to its domestic politics, this endangers
its shared relations with other states. Those states are motivated to intervene in the affairs
of their straying partner in order to restore the previous status quo and avoid the certain
cost of allowing reform in the partner state. These interventions lead to rivalry.
This is a novel and valuable insight. However, this investigation represents only a first,
if promising, attempt to better understand a very complex relationship between states, and
illuminates only one process by which states can fall into this relationship. As such, much
additional research is required. Further, the logic of the argument presented here goes beyond
the narrow test presented. We should expect that the types of interests discussed here are
not the only ones that can motivate rivalry in the wake of radical political change. Additional
research should expand upon the insight developed in this investigation by examining these
various other forms of shared ties to determine which may be salient enough to motivate
conflict when threatened by the activities of the states they tie together, and how significant
an event of domestic upheaval must be before it motivates conflict.
27
Citations
Bailey, M.A., Strezhnev, A. and Voeten, E., 2017. Estimating dynamic state preferences
from United Nations voting data. Journal of Conflict Resolution, 61(2), pp.430-456.
Bain, M. J. (2011). Russia and Cuba:“doomed” comrades?. Communist and Post-Communist
Studies, 44(2), 111-118.
Bennett, D. Scott. 1997. "Democracy, regime change, and rivalry termination." International
Interactions 22(4): 369-97.
Bennett, D. Scott. 1998. "Integrating and Testing Models of Rivalry Duration." American
Journal of Political Science 42(4): 1200.
Boutton, Andrew. 2014. "US foreign aid, interstate rivalry, and incentives for counterterror-
ism cooperation." Journal of Peace Research 51(6): 741-54.
Butler, Michael J. 2003. "U.S. Military Intervention in Crisis, 1945-1994." An Empirical
Inquiry Of Just War Theory 47(2): 226-248
Bueno de Mesquita, Bruce. "The war trap." (1981).
Carter, D. B., & Signorino, C. S. (2010). Back to the future: Modeling time dependence in
binary data. Political Analysis, 18(3), 271-292.
Chiba, Daina, Jesse C. Johnson, and Brett Ashley Leeds. 2015. "Careful Commitments:
Democratic States and Alliance Design." The Journal of Politics 77(4): 968-82.
Colaresi, Michael P., Karen Rasler, and William R. Thompson. (2007) "Defining and iden-
tifying strategic rivalries in world politics." Strategic Rivalries in World Politics: 21-72.
Colgan, J. D. (2013). Domestic revolutionary leaders and international conflict. World
Politics, 65(4), 656-690.
Colgan, Jeff D. 2010. "Oil and Revolutionary Governments: Fuel for International Conflict."
International Organization 64(04): 661-694.
Colgan, Jeff D. 2012. "Measuring revolution." Conflict Management and Peace Science,
29(4), 444-467.
Conrad, Justin. 2011. "Interstate Rivalry and Terrorism." Journal of Conflict Resolution
55(4): 529-55.
Colgan, J. D., & Weeks, J. L. (2015). Revolution, personalist dictatorships, and international
conflict. International Organization, 69(1), 163-194.
Conrad, Justin, and Mark Souva. 2011. "Regime Similarity and Rivalry." International
Interactions 37(1): 1-28.
Davies, Graeme A. M. 2002. Domestic strife and the initiation of international conflicts: A
directed dyad analysis, 1950-1982. Journal of Conflict Resolution 46 (5): 672-92.
Coppedge, Michael, John Gerring, Carl Henrik Knutsen, Staffan I. Lindberg, Svend-Erik
Skaaning, Jan Teorell, David Altman, Michael Bernhard, Agnes Cornell, M. Steven Fish,
28
Haakon Gjerløw, Adam Glynn, Allen Hicken, Joshua Krusell, Anna Lührmann, Kyle L.
Marquardt, Kelly McMann, Valeriya Mechkova, Moa Olin, Pamela Paxton, Daniel Pemstein,
Brigitte Seim, Rachel Sigman, Jeffrey Staton, Aksel Sundtröm, Eitan Tzelgov, Luca Uberti,
Yi-ting Wang, Tore Wig, and Daniel Ziblatt. 2018. "V-Dem Codebook v8" Varieties of
Democracy (V-Dem) Project.
Diehl, P. F., & Goertz, G. (2001). War and peace in international rivalry. University of
Michigan Press.
Dixon, J. S., & Sarkees, M. R. (2015). A Guide to Intra-state Wars: An Examination of
Civil, Regional, and Intercommunal Wars, 1816-2014. CQ Press.
Dreyer, D. R. (2010). Issue conflict accumulation and the dynamics of strategic rivalry.
International Studies Quarterly, 54(3), 779-795.
Dreyer, D. R. (2010). One Issue Leads to Another: Issue Spirals and the Sino-Vietnamese
War. Foreign Policy Analysis, 6(4), 297-315.
Findley, Michael, James Piazza, and Joseph K. Young. 2010. "Games Rivals Play: Terrorism
in International Rivalries." SSRN Electronic Journal.
Fitzgibbon, Russell H., and Hugh Thomas. 1972. "Cuba: The Pursuit of Freedom." The
American Historical Review 77(3): 847.
Frederick, B. A., Hensel, P. R., & Macaulay, C. (2017). The issue correlates of war territorial
claims data, 1816–2001. Journal of Peace Research, 54(1), 99-108.
Geddes, B., Wright, J., & Frantz, E. (2014). Autocratic breakdown and regime transitions:
A new data set. Perspectives on Politics, 12(2), 313-331.
Gleditsch, Kristian Skrede. 2002. "Expanded Trade and GDP Data." Journal of Conflict
Resolution 46(5): 712-24.
Gleditsch, K. S., Salehyan, I., & Schultz, K. (2008). Fighting at home, fighting abroad: How
civil wars lead to international disputes. Journal of Conflict Resolution, 52(4), 479-506.
Goertz, Gary, and Paul F. Diehl. 1992. "The empirical importance of enduring rivalries."
International Interactions 18(2): 151-63.
Goertz, Gary, and Paul F. Diehl. 1995. "The Initiation and Termination of Enduring
Rivalries: The Impact of Political Shocks." American Journal of Political Science 39(1): 30.
Haas, Mark L. 2001. "Prospect Theory and the Cuban Missile Crisis" International Studies
Quarterly 45: 241-270.
Haynes, Kyle. 2015. "Diversionary conflict: Demonizing enemies or demonstrating compe-
tence?" Conflict Management and Peace Science 34(4): 337-58.
Hensel, Paul R. 2001. "Contentious Issues and World Politics: The Management of Territo-
rial Claims in the Americas, 1816-1992." International Studies Quarterly Vol. 45, No. 1, pp.
81-109
29
Hensel, P. R. (2013). General Coding Manual: Issue Correlates of War (ICOW) Project.
University of North Texas.
Jervis, R. (1992). Political implications of loss aversion. Political psychology, 187-204.
Kahneman, Daniel, Jack L. Knetsch and Richard H. Thaler. 1990. "Experimental Tests
of the Endowment Effect and the Coase Theorem." Journal of Political Economy. 98(6).
1325-1348
Kahneman, Daniel and Amos Tversky. 1979. "Prospect Theory: An Analysis of Decision
under Risk." Econometrica. Vol. 47, No. 2, pp. 263-292
Kim, C. E., & Kim, K. S. (1964). The April 1960 Korean Student Movement. Western
Political Quarterly, 17(1), 83-92.
King, G., & Zeng, L. (2001). Logistic regression in rare events data. Political analysis, 9(2),
137-163.
Klein, James P., Gary Goertz, and Paul F. Diehl. 2006. "The New Rivalry Dataset: Proce-
dures and Patterns." Journal of Peace Research 43(3): 331-48.
Leeds, Brett Ashley. 2003. "Do Alliances Deter Aggression? The Influence of Military
Alliances on the Initiation of Militarized Interstate Disputes." American Journal of Political
Science 47(3): 427.
Leeds, B., Ritter, J., Mitchell, S., & Long, A. (2002). Alliance treaty obligations and provi-
sions, 1815-1944. International Interactions, 28(3), 237-260.
Lemke, Douglas, and William Reed. 2001. "War and Rivalry among Great Powers." Ameri-
can Journal of Political Science 45(2): 457.
Levy. 1992. "An Introduction to Prospect Theory." Political Psychology 171-168.
Levy. 1996. "Loss Aversion, Framing, and Bargaining: The Implications of Prospect Theory
for International Conflict." International Political Science Review Vol 17, No. 2, 179-195.
Maoz, Z., 1989. Joining the club of nations: Political development and international conflict,
1816–1976. International Studies Quarterly, 33(2), pp.199-231.
Mansfield, Edward, and Jack Snyder. 1995. Democratization and the danger of war. Inter-
national Security 20 (1): 5-38.
Mattes, Michaela, Brett Ashley Leeds, and Naoko Matsumura. 2016. "Measuring change in
source of leader support." Journal of Peace Research 53(2): 259-67.
McDermott, R. (1992). Prospect Theory in International Relations: The Iranian Hostage
Rescue Mission. Political Psychology, 13(2), 237-263.
McDermott, Rose. 2001. "The Psychological Ideas of Amos Tversky and Their Relevance
for Political Science." Journal of Theoretical Politics. 13(1): 5-33
McDermott, Rose and Jack Kugler. 2001. "Comparing Rational Choice and Prospect Theory
Analyses: The US Decision to Launch Operation ’Desert Storm’, January 1991."The Journal
of Strategic Studies. 24(3): 49-85
30
Morrow, James D. 1991. "Alliances and Asymmetry: An Alternative to the Capability
Aggregation Model of Alliances." American Journal of Political Science 35(4): 904.
Ostovar, Afshon. P. (2009). Guardians of the Islamic Revolution Ideology, Politics, and
the Development of Military Power in Iran (1979-2009) (Unpublished doctoral dissertation).
The University of Michigan, Ann Arbor, MI
Pearson, Fredrick S. 1974. "Geographic Proximity and Foreign Military Intervention" Journal
of Conflict Resolution, Vol. 18 No. 3
Ray, J. L. (1995). Democracy and international conflict (p. 854). Columbia: University of
South Carolina Press.
Rider, T. J., & Owsiak, A. P. (2015). Border settlement, commitment problems, and the
causes of contiguous rivalry. Journal of Peace Research, 52(4), 508-521.
Saideman, Stephen M. 2001. The Ties That Divide: Ethnic Politics, Foreign Policy, and
International Conflict. New York: Columbia University Press.
Signorino, Curtis S., and Jeffrey M. Ritter. 1999. "Tau-b or Not Tau-b: Measuring the
Similarity of Foreign Policy Positions." International Studies Quarterly 43(1): 115-44.
Singer, J. David; Stuart Bremer & John Stuckey, (1972). ’Capability Distribution, Uncer-
tainty, and Major Power War, 1820-1965’, in Bruce Russett, ed., Peace, War, and Numbers.
Beverly Hills, CA: Sage (19-48).
Snyder, Robert S. 1999. "The U.S. and third World Revolutionary States: Understanding
the Breakdown in Relations." International Studies Quarterly 43(2): 265-90.
Stinnet, Douglas M., Jaroslav Tir, Philip Schafer, Paul F. Diehl, and Charles Gochman.
2002. "The Correlates of War Project Direct Contiguity Data, Version 3." Conflict Manage-
ment and Peace Science 19(2):58-66.
Stinnett, Douglas M., and Paul F. Diehl. 2001. "The Path(s) to Rivalry: Behavioral and
Structural Explanations of Rivalry Development." The Journal of Politics 63(3): 717-40.
Thaler, Richard. 1980. "Toward a Positive Theory of Consumer Choice." Journal of Eco-
nomic Behavior and Organization 1: 39-60
Thompson, William R. 1995. "Principal Rivalries." Journal of Conflict Resolution 39(2):
195-223.
Thompson, William R., and David R. Dreyer. Handbook of International Rivalries: 1494-
2010. Washington D.C.: Congressional Quarterly Press, 2011
Thyne, C., Powell, J., Parrott, S., & VanMeter, E. (2018). Even Generals Need Friends:
How Domestic and International Reactions to Coups Influence Regime Survival. Journal of
Conflict Resolution, 62(7), 1406–1432.
Tversky, Amos and Daniel Kahneman. 1992. "Advances in Prospect Theory: Cumulative
Representation of Uncertainty." Journal of Risk and Uncertainty. 5: 297-323
31
Valeriano, B. (2013). Becoming rivals: The process of interstate rivalry development. Rout-
ledge.
Vasquez, John, and Christopher S. Leskiw. 2001 "The Origins and War Proneness of Inter-
state Rivalries" Annu. Rev. Polit. Sci. 2001. 4:295-316
Vasquez, J. A. (2009). The war puzzle revisited. Cambridge University Press.
Voeten, Erik. 2000. "Clashes in the Assembly." International Organization 54(2): 185-215.
Walt, Stephen M. 1985. "Alliance Formation and the Balance of World Power." International
Security 9(4): 3.
Walt, Stephen. 1996. Revolution and war. Ithaca, NY: Cornell University Press
Werner, S. (2000). Deterring intervention: The stakes of war and third-party involvement.
American Journal of Political Science, 720-732.
32
Main Tables and Figures:
Table 1: Bivariate Relationship - Revolution and Rivalry Onset (All dyads 1950-2005)
No Revolution-Last 5 Years Revolution-Last 5 Years Total
No Rivalry Onset 48,807
(48,797.3)
4,118
(4,127.7) 52,925
Rivalry Onset 41
(50.7)
14
(4.3)) 55
Total 48,848 4,132 52,980
Pearson chi2 = 23.8663 Pr = 0.000
Note: Cells display the observed frequency over the expected frequency (in parentheses.)
33
Table 2: Logistic Regression of Rivalry Initiation 1950-2005
(1) (2) (3) (4) (5)
Revolution Foreign Policy Alliance Trade Latent Interests
b/se b/se b/se b/se b/se
Revolution 1.977*** -4.497 0.868 1.843*** 0.578
(0.523) (2.901) (0.703) (0.539) (0.912)
Revolution ×Policy 7.266*
(3.143)
Revolution ×Alliance 3.612*
(1.442)
Revolution ×Trade 18.464+
(11.127)
Revolution ×Shared Interest 10.388**
(3.629)
Shared Interest -2.122
(2.351)
Alliance -0.349 -0.376 -2.565* -0.352
(0.543) (0.546) (1.264) (0.539)
Policy 3.068+ 1.018 3.114+ 3.099+
(1.811) (1.746) (1.790) (1.796)
Trade 1.643 2.384 4.637 -2.086
(4.172) (4.185) (4.613) (7.428)
Civil War 0.639 0.587 1.029 0.657 0.794
(0.770) (0.783) (0.750) (0.780) (0.707)
Capability Disparity -4.588* -4.457* -4.451* -4.477* -4.304+
(2.246) (2.173) (2.236) (2.266) (2.214)
Democracy -2.899* -2.934* -2.615* -2.797* -2.609*
(1.249) (1.266) (1.256) (1.318) (1.128)
Contiguity 0.346 0.469 0.381 0.346 0.881
(0.876) (0.911) (0.923) (0.885) (1.024)
Cold War 0.374 0.320 0.215 0.363 -0.008
(0.879) (0.920) (0.862) (0.883) (0.836)
Systemic Shock 1.172 1.180 1.159 1.167 1.005
(0.936) (0.959) (0.938) (0.934) (0.902)
Ongoing Territorial Dispute 2.686** 2.838** 3.371** 2.703** 2.872**
(0.923) (0.952) (1.088) (0.939) (1.060)
Constant -6.625+ -5.022 -6.478+ -6.697+ -4.084
(3.988) (3.711) (3.782) (4.011) (2.740)
Observations 36854 36854 36854 36854 35829
Robust standard errors clustered on the dyad –in parentheses
Cubic polynomial of years of peace included in model but not reported
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001 in 2-tail test
34
0
.002
.004
.006
.008
.01
Marginal Effect of Revolution
0 .2 .4 .6 .8 1
Change in Policy Similarity
95% CI
Figure 1: Effect of Revolution conditional on Policy Similarity
0
.01
.02
.03
.04
Marginal Effect of Revolution
-.4 -.2 0 .2 .4
Change in Latent Interests
95% CI
Figure 2: Effect of Revolution conditional on Shared Interests
35
Appendix A:
36
Table 3: Rivalries Following Within 5 Years of Revolution - 1950-2005
Country A Country B Revolution Date Rivalry Onset Alliance
Top 25% Annually
Trade Dependence
Top 25% Annually
Policy Similarity Major Interest
USA Cuba 1959 1960 Yes Yes Yes Yes
USA China 1949 1950 Yes Yes
Honduras El Salvador 1979 1980 Yes Yes Yes
Senegal Guinea 1984 1989 Yes Yes
Uganda Kenya 1986 1987 Yes Yes
Uganda Tanzania 1971 1972 Yes Yes Yes
Uganda Sudan 1989 1994 Yes Yes
Libya Sudan 1969 1973 Yes Yes
Libya Egypt 1969 1973 Yes Yes Yes
Sudan Egypt 1989 1991 Yes Yes Yes Yes
Iran Afghanistan 1996 1997 Yes Yes
Iran Saudi Arabia 1979 1980 NO
Iran Egypt 1979 1980 NO
Iran Egypt 1953 1955 Yes Yes
Iran Israel 1979 1980 Yes Yes
Iraq Saudi Arabia 1968 1969 Yes Yes Yes
Cambodia Vietnam 1975 1976 Yes Yes
Nicaragua Columbia 1979 1980 Yes Yes Yes
Egypt Saudi Arabia 1956 1957 Yes Yes Yes Yes
Totals 19 Revolutions 8 7 6 17
74 total rivalries are initiated 1950-2005.
Revolutions in dyads that share important interests account for 23%.
Table 4: Rivalries Following Within 5 Years of SOLS Change - 1950-2005
State A State B SOLS Change SOLS Rivalry Alliance
Top 25% Annually
Policy Similarity
Top 25% Annually
Trade Dependence Major Interest
United States Cuba 1957 1960 Yes Yes Yes Yes
United States China 1949 1950 Yes Yes
Honduras Nicaragua 1979 1980 Yes Yes Yes
Senegal Mauritania 1984 1989 Yes Yes Yes
Senegal Guinea 1984 1989 Yes Yes
Democratic Republic of Congo Rwanda 1994 1996 Yes Yes
Uganda Kenya 1986 1987 Yes Yes
Uganda Tanzania 1971 1972 Yes Yes Yes
Uganda Rwanda 1994 1999 Yes Yes
Uganda Sudan 1989 1994 Yes Yes
Kenya Sudan 1985 1989 Yes Yes Yes
Zimbabwe South Africa 1979 1980 Yes Yes
Libya Sudan 1969 1973 Yes Yes
Libya Egypt 1969 1973 Yes Yes Yes
Sudan Egypt 1989 1991 Yes Yes Yes Yes
Iran Afghanistan 1997 1997 Yes Yes
Iran Saudi Arabia 1979 1980 NO
Iran Egypt 1953 1955 Yes Yes
Iran Egypt 1979 1980 NO
Iran Israel 1979 1980 Yes Yes
Iran Syria 1953 1958 Yes Yes Yes
Iraq Saudi Arabia 1968 1969 Yes Yes Yes Yes
Cambodia Vietnam 1975 1976 Yes Yes
Nicaragua Columbia 1979 1980 Yes Yes Yes
Egypt Saudi Arabia 1956 1957 Yes Yes Yes Yes
25 SOLS Changes 9 7 13 23
74 total rivalries are initiated 1950-2005.
SOLS changes in dyads that share important interests account for 31%.
37
Table 5: Rare Events Logistic Regression of Rivalry Initiation 1950-2005
(Revolution Only) (Shared Foreign Policy) (Alliance) (Trade Dependence) (Latent Interests)
ˆ
β/se ˆ
β/se ˆ
β/se ˆ
β/se
[.3em] Revolution 1.978*** -3.301 0.963 1.901*** 0.723
(0.523) (2.900) (0.703) (0.538) (0.912)
X Foreign Policy 6.017+
(3.142)
X Alliance 2.985*
(1.441)
X Trade Dependence 9.366
(11.122)
X Shared Interests 9.462**
(3.627)
Shared Interests -2.079
(2.350)
Shared Foreign Policy 2.689 0.649 2.749 2.760
(1.811) (1.745) (1.789) (1.796)
Alliance -0.331 -0.357 -2.028 -0.348
(0.543) (0.545) (1.264) (0.539)
Trade Dependence 10.149* 11.013** 13.474** 14.320+
(4.170) (4.183) (4.611) (7.425)
Civil War 0.765 0.696 1.118 0.780 0.903
(0.770) (0.783) (0.750) (0.780) (0.707)
Capability Disparity -4.300+ -4.176+ -4.122+ -4.183+ -4.134+
(2.245) (2.172) (2.235) (2.265) (2.213)
Democracy -2.335+ -2.344+ -2.100+ -2.209+ -1.903+
(1.248) (1.266) (1.255) (1.318) (1.127)
Contiguity 0.320 0.429 0.357 0.315 0.839
(0.876) (0.910) (0.922) (0.885) (1.024)
Cold War 0.192 0.160 0.038 0.184 -0.159
(0.879) (0.920) (0.861) (0.883) (0.836)
Systemic Shock 1.000 1.029 0.971 0.992 0.819
(0.936) (0.959) (0.938) (0.933) (0.901)
Ongoing Territorial Dispute 2.809** 2.952** 3.431** 2.820** 2.985**
(0.923) (0.952) (1.088) (0.938) (1.060)
Constant -6.221 -4.632 -6.094 -6.344 -3.932
(3.986) (3.710) (3.780) (4.009) (2.739)
Observations 36854 36854 36854 36854 35829
Robust standard errors clustered on the dyad –in parentheses
Cubic polynomial of years of peace included in model but not reported
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001 in 2-tail test
38
Table 6: Bivariate Relationship - SOLS Change and Rivalry Onset (1950-2005)
No Revolution-Last 5 Years Revolution-Last 5 Years Total
No Rivalry Onset 43,503
(43,494.8)
9,422
(9,430.2) 52,925
Rivalry Onset 37
(45.2)
18
(9.8) 55
Total 43,540 9,440 52,980
Pearson chi2(1) = 8.3577 Pr = 0.004
Note: Cells display the observed frequency over the expected frequency (in parentheses.)
Table 7: Robustness to change in observation window - Foreign Policy
(1) (2) (3) (4) (5) (6) (7)
2-Year 3-Year 4-Year 5-Year 6-Year 7-Year 8-Year
b/se b/se b/se b/se b/se b/se b/se
Revolution ×Shared Foreign Policy 8.377+ 12.412* 3.489 7.266* 7.420* 7.004* 6.222*
(4.452) (4.965) (2.768) (3.143) (2.984) (3.156) (3.164)
Shared Foreign Policy 2.346 2.531 2.529 1.018 0.706 0.941 0.961
(2.128) (1.925) (1.940) (1.746) (1.794) (1.875) (1.568)
Revolution -6.046 -10.265* -1.509 -4.497 -4.633+ -4.669 -4.258
(4.135) (4.757) (2.541) (2.901) (2.781) (2.995) (3.096)
Alliance -0.612 -0.600 -0.717 -0.376 -0.292 -0.111 -0.570
(0.455) (0.476) (0.510) (0.546) (0.577) (0.600) (0.659)
Trade Dependence 1.475 -1.060 -0.363 2.384 5.750 7.170* 4.217
(6.687) (7.034) (5.496) (4.185) (3.645) (3.649) (3.761)
Civil War 1.077 1.074 0.273 0.587 0.618 1.247+ 1.287+
(0.710) (0.774) (0.852) (0.783) (0.785) (0.671) (0.774)
Capability Disparity -3.889* -3.719* -3.934* -4.457* -3.030 -2.682 -4.115+
(1.614) (1.662) (1.850) (2.173) (2.080) (2.286) (2.470)
Democracy -5.629*** -3.756** -2.524* -2.934* -3.599* -4.903** -2.827+
(1.530) (1.399) (1.057) (1.266) (1.431) (1.642) (1.578)
Contiguity 1.225 0.907 0.493 0.469 0.668 0.507 0.223
(0.836) (0.800) (0.813) (0.911) (0.922) (0.981) (1.053)
Cold War -0.429 0.404 0.847 0.320 0.343 0.319 0.356
(0.947) (0.863) (1.087) (0.920) (0.921) (0.910) (0.875)
Systemic Shock 0.585 1.271 1.928+ 1.180 1.261 1.299 1.410
(0.949) (0.849) (1.092) (0.959) (0.980) (0.971) (0.959)
Ongoing Territorial Dispute 2.659** 2.397** 2.479** 2.838** 3.045** 3.084** 2.876**
(0.989) (0.928) (0.946) (0.952) (0.969) (0.994) (1.027)
Constant -5.082 -6.202+ -6.960+ -5.022 -6.130+ -6.378+ -5.007
(3.386) (3.286) (3.683) (3.711) (3.629) (3.848) (3.764)
Observations 38118 38049 37959 36854 35783 34689 33602
Robust standard errors clustered on the dyad
Cubic polynomial of years of peace included in model but not reported
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001 in 2-tail test
39
Table 8: Robustness to change in observation window - Alliance
(1) (2) (3) (4) (5) (6) (7)
2-Year 3-Year 4-Year 5-Year 6-Year 7-Year 8-Year
b/se b/se b/se b/se b/se b/se b/se
Alliance ×Revolution 0.966 0.939 1.735 3.612* 3.703* 4.341** 3.636*
(0.971) (1.006) (1.092) (1.442) (1.500) (1.470) (1.566)
Alliance -0.891 -0.871 -1.411+ -2.565* -2.599+ -2.694* -2.687*
(0.589) (0.613) (0.793) (1.264) (1.338) (1.338) (1.363)
Revolution 1.258* 0.864 1.007+ 0.868 0.780 -0.006 0.159
(0.604) (0.634) (0.599) (0.703) (0.721) (0.728) (0.744)
Shared Foreign Policy 3.445 3.872+ 3.363+ 3.114+ 3.013+ 3.058+ 2.740+
(2.215) (2.118) (1.790) (1.790) (1.783) (1.777) (1.503)
Trade Dependence 2.042 -0.653 0.504 4.637 8.315* 10.326** 6.139
(6.527) (6.909) (5.563) (4.613) (4.105) (3.698) (4.045)
Civil War 1.074 1.124 0.551 1.029 1.079 1.824** 1.798*
(0.678) (0.718) (0.818) (0.750) (0.752) (0.607) (0.741)
Capability Disparity -3.850* -3.758* -3.888* -4.451* -2.937 -2.450 -4.089
(1.631) (1.676) (1.871) (2.236) (2.161) (2.328) (2.504)
Democracy -5.421*** -3.582** -2.431* -2.615* -3.446* -4.923** -2.713+
(1.469) (1.320) (1.055) (1.256) (1.443) (1.597) (1.483)
Contiguity 1.160 0.807 0.492 0.381 0.625 0.508 0.124
(0.835) (0.784) (0.827) (0.923) (0.900) (0.964) (1.071)
Cold War -0.422 0.420 0.745 0.215 0.225 0.196 0.295
(0.938) (0.855) (1.041) (0.862) (0.848) (0.836) (0.821)
Systemic Shock 0.561 1.264 1.848+ 1.159 1.214 1.320 1.483
(0.946) (0.844) (1.054) (0.938) (0.947) (0.950) (0.949)
Ongoing Territorial Dispute 2.656** 2.403* 2.680* 3.371** 3.625*** 3.942*** 3.591**
(1.023) (0.979) (1.048) (1.088) (1.098) (1.197) (1.238)
Constant -6.045+ -7.298* -7.529* -6.478+ -7.751* -7.994* -6.270+
(3.481) (3.487) (3.548) (3.782) (3.688) (3.802) (3.690)
Observations 38118 38049 37959 36854 35783 34689 33602
Robust standard errors clustered on the dyad
Cubic polynomial of years of peace included in model but not reported
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001 in 2-tail test
40
Table 9: Robustness to change in observation window - Trade
(1) (2) (3) (4) (5) (6) (7)
2-Year 3-Year 4-Year 5-Year 6-Year 7-Year 8-Year
b/se b/se b/se b/se b/se b/se b/se
Revolution ×Trade Dependence 24.005+ 29.957+ 25.574* 18.464+ 24.896** 17.789+ 19.413+
(13.939) (16.791) (12.758) (11.127) (9.336) (10.397) (10.974)
Trade Dependence -3.673 -9.156 -6.428 -2.086 0.893 4.928 0.992
(12.797) (15.926) (11.276) (7.428) (6.314) (4.525) (6.080)
Revolution 1.387* 0.952+ 1.393* 1.843*** 1.735** 1.404* 1.041
(0.539) (0.575) (0.546) (0.539) (0.565) (0.604) (0.639)
Shared Foreign Policy 3.403 3.827+ 3.400+ 3.099+ 3.016+ 3.023+ 2.736+
(2.207) (2.109) (1.796) (1.796) (1.776) (1.762) (1.443)
Alliance -0.589 -0.566 -0.702 -0.352 -0.281 -0.133 -0.603
(0.457) (0.476) (0.511) (0.539) (0.569) (0.597) (0.661)
Civil War 1.063 1.115 0.382 0.657 0.685 1.325+ 1.430+
(0.720) (0.757) (0.878) (0.780) (0.784) (0.695) (0.780)
Capability Disparity -3.837* -3.739* -3.863* -4.477* -2.906 -2.668 -4.106
(1.644) (1.684) (1.885) (2.266) (2.168) (2.376) (2.574)
Democracy -5.520*** -3.543* -2.377* -2.797* -3.299* -4.641** -2.561
(1.607) (1.455) (1.145) (1.318) (1.401) (1.573) (1.576)
Contiguity 1.128 0.773 0.461 0.346 0.550 0.368 0.103
(0.852) (0.806) (0.819) (0.885) (0.877) (0.936) (1.023)
Cold War -0.373 0.509 0.889 0.363 0.392 0.390 0.405
(0.981) (0.905) (1.134) (0.883) (0.877) (0.869) (0.842)
Systemic Shock 0.596 1.339 1.927+ 1.167 1.281 1.359 1.463
(0.990) (0.889) (1.140) (0.934) (0.950) (0.943) (0.915)
Ongoing Territorial Dispute 2.620** 2.341* 2.448** 2.703** 2.897** 2.956** 2.752**
(0.987) (0.936) (0.948) (0.939) (0.949) (0.989) (1.026)
Constant -6.034+ -7.344* -7.754* -6.697+ -8.098* -8.078* -6.413
(3.547) (3.564) (3.723) (4.011) (3.910) (4.109) (3.955)
Observations 38118 38049 37959 36854 35783 34689 33602
Robust standard errors clustered on the dyad
Cubic polynomial of years of peace included in model but not reported
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001 in 2-tail test
41
Table 10: Robustness to change in observation window - Latent Interests
(1) (2) (3) (4) (5) (6) (7)
2-Year 3-Year 4-Year 5-Year 6-Year 7-Year 8-Year
b/se b/se b/se b/se b/se b/se b/se
Revolution ×Shared Interests 8.051 8.527 8.837* 14.104** 14.920** 15.003* 10.152*
(5.067) (5.374) (4.373) (5.013) (5.411) (6.038) (4.997)
Shared Interests -5.290 -4.401 -7.000+ -7.081+ -5.872 -5.661 -5.987
(4.028) (3.782) (3.752) (4.218) (4.351) (4.486) (4.457)
Revolution 0.466 -0.023 0.425 0.080 -0.078 -0.564 -0.029
(0.965) (1.061) (0.901) (1.114) (1.220) (1.363) (1.151)
Trade Dependence 4.561 1.620 4.439 6.645 9.973* 11.497* 8.378+
(6.503) (6.911) (5.509) (5.450) (4.630) (4.951) (5.044)
Shared Foreign Policy 5.272+ 5.240* 5.673** 4.850* 4.045+ 4.074+ 4.075*
(2.828) (2.606) (2.184) (2.010) (2.125) (2.323) (1.873)
Alliance -0.304 -0.386 -0.353 -0.420 -0.526 -0.333 -0.434
(0.595) (0.587) (0.628) (0.635) (0.675) (0.666) (0.758)
Civil War 1.029 1.051 0.337 0.647 0.720 1.400* 1.389+
(0.690) (0.735) (0.863) (0.757) (0.769) (0.659) (0.781)
Capability Disparity -3.887* -3.759* -3.931* -4.383+ -2.749 -2.270 -3.978
(1.611) (1.662) (1.869) (2.250) (2.155) (2.368) (2.555)
Democracy -5.489*** -3.654** -2.502* -2.830* -3.577* -5.102** -2.961+
(1.519) (1.354) (1.047) (1.275) (1.417) (1.681) (1.614)
Contiguity 1.414 1.039 0.871 0.763 0.912 0.807 0.522
(0.882) (0.818) (0.876) (1.008) (0.999) (1.053) (1.123)
Cold War -0.578 0.246 0.543 -0.070 -0.017 -0.019 0.038
(0.960) (0.892) (1.120) (0.894) (0.883) (0.875) (0.870)
Systemic Shock 0.476 1.175 1.715 0.903 0.995 1.064 1.185
(0.944) (0.862) (1.114) (0.913) (0.931) (0.931) (0.932)
Ongoing Territorial Dispute 2.387* 2.209* 2.211* 2.677* 3.057** 3.206** 2.768*
(1.083) (1.001) (1.120) (1.104) (1.081) (1.116) (1.198)
Constant -7.096+ -8.001* -8.955* -7.377* -8.238* -8.447* -6.939*
(3.658) (3.530) (3.644) (3.489) (3.382) (3.571) (3.406)
Observations 37600 37356 37096 35823 34586 33326 32081
42
Table 11: Logistic Regression 1950-2005 (CHISOLS Change Variable)
(Revolution Only) (Shared Foreign Policy) (Alliance) (Trade Dependence) (Latent Interests)
ˆ
β/se ˆ
β/se ˆ
β/se ˆ
β/se ˆ
β/se
SOLS Change 1.494** -3.904+ 0.747 1.343** -0.019
(0.460) (2.254) (0.564) (0.469) (0.837)
SOLS Change ×Shared Foreign Policy 6.134*
(2.484)
SOLS Change ×Alliance 17.798***
(4.173)
SOLS Change ×Trade Dependence 19.250+
(10.134)
SOLS Change ×Shared Interests 13.041**
(4.036)
Shared Interests -9.756*
(4.575)
Shared Foreign Policy 2.569 0.461 2.913+ 2.630 5.403*
(1.723) (1.740) (1.711) (1.736) (2.126)
Alliance -0.335 -0.348 -17.426*** -0.323 -0.318
(0.567) (0.578) (3.951) (0.565) (0.599)
Trade Dependence 3.126 4.287 12.451+ -0.292 12.012*
(3.515) (3.493) (7.010) (6.115) (4.728)
Civil War 1.324+ 1.325+ 1.545* 1.401+ 1.370+
(0.738) (0.744) (0.751) (0.774) (0.744)
Capability Disparity -4.580+ -4.313+ -4.220+ -4.542+ -4.244+
(2.370) (2.274) (2.363) (2.399) (2.403)
Democracy -2.948* -2.984* -3.047* -2.805* -3.150*
(1.295) (1.330) (1.370) (1.329) (1.333)
Contiguity 0.276 0.369 0.351 0.228 0.739
(0.861) (0.881) (0.892) (0.887) (1.026)
Cold War 0.391 0.350 0.401 0.380 0.030
(0.828) (0.865) (0.846) (0.834) (0.837)
Systemic Shock 1.084 1.029 1.089 1.109 0.828
(0.844) (0.877) (0.863) (0.841) (0.835)
Ongoing Territorial Dispute 2.159* 2.181* 2.049* 2.130* 2.190*
(0.907) (0.931) (0.946) (0.908) (0.954)
Constant -6.182 -4.627 -6.522+ -6.210 -7.893*
(3.984) (3.688) (3.889) (4.031) (3.620)
Observations 36854 36854 36854 36854 35823
Robust standard errors clustered on the dyad
Cubic polynomial of years of peace included in model but not reported
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001 in 2-tail test
43
Table 12: Logistic Regression 1950-2005 (Threshold Robustness - Alliances)
(1) (2) (3) (4) (5) (6)
Threshold =1 Threshold = 2 Threshold = 3 Threshold = 4 Threshold = 5 Threshold = 6
b/se b/se b/se b/se b/se b/se
Threshold = 1 ×Alliance 16.961***
(3.901)
Threshold = 1 0.810
(0.598)
Threshold = 2 ×Alliance 16.869***
(3.797)
Threshold = 2 1.140+
(0.602)
Threshold = 3 ×Alliance 3.621*
(1.449)
Threshold = 3 0.881
(0.702)
Threshold = 4 ×Alliance 4.075**
(1.472)
Threshold = 4 0.645
(0.799)
Threshold = 5 ×Alliance 17.540***
(1.072)
Threshold = 5 -13.559***
(0.730)
Threshold = 6 ×Alliance 18.813***
(1.288)
Threshold = 6 -14.294***
(0.776)
Alliance -16.521*** -16.220*** -2.555* -2.623* -1.521+ -0.663
(3.803) (3.715) (1.284) (1.262) (0.801) (0.581)
Trade Dependence 11.152 9.798 4.702 3.175 0.503 0.340
(6.851) (7.250) (4.447) (5.513) (6.373) (4.504)
Shared Foreign Policy 3.245+ 3.203+ 3.385+ 3.646+ 3.505+ 3.216+
(1.947) (1.929) (1.936) (2.008) (1.889) (1.838)
Civil War 1.403+ 1.138 1.057 1.155 2.118** 2.509***
(0.742) (0.735) (0.747) (0.751) (0.688) (0.752)
Capability Disparity -4.564* -4.545* -4.809* -4.797** -5.730*** -5.367**
(1.926) (1.907) (1.886) (1.811) (1.686) (1.866)
Democracy -2.905* -2.533+ -2.599* -2.313+ -2.842* -3.028**
(1.324) (1.308) (1.259) (1.231) (1.153) (1.107)
Cold War 0.356 0.235 0.358 0.402 0.415 0.509
(0.943) (0.939) (0.948) (0.946) (0.912) (0.911)
Systemic Shock 1.275 1.365 1.353 1.343 1.197 1.392
(0.984) (1.007) (1.033) (1.028) (0.962) (0.978)
Ongoing Territorial Dispute 2.773** 3.050*** 3.368** 3.315** 2.753** 2.447*
(0.872) (0.892) (1.092) (1.108) (1.025) (0.953)
Constant -6.190 -6.269+ -6.315+ -6.470+ -5.485 -5.771
(3.807) (3.720) (3.736) (3.732) (3.437) (3.628)
Observations 36854 36854 36854 36854 36854 36854
Robust standard errors clustered on the dyad
Cubic polynomial of years of peace included in model but not reported
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001 in 2-tail test
44
Table 13: Logistic Regression 1950-2005 (Threshold Robustness - Shared Foreign Policy)
(1) (2) (3) (4) (5) (6)
Threshold = 1 Threshold = 2 Threshold = 3 Threshold = 4 Threshold = 5 Threshold = 6
b/se b/se b/se b/se b/se b/se
Threshold = 1 ×Shared Foreign Policy 6.073*
(2.725)
Threshold = 1 -3.717
(2.431)
Threshold = 2 ×Shared Foreign Policy 7.013*
(2.807)
Threshold = 2 -4.103+
(2.491)
Threshold = 3 ×Shared Foreign Policy 7.100*
(3.174)
Threshold = 3 -4.331
(2.898)
Threshold = 4 ×Shared Foreign Policy 7.595*
(3.687)
Threshold = 4 -4.860
(3.363)
Threshold = 5 ×Shared Foreign Policy 1.125
(2.370)
Threshold = 5 0.325
(2.100)
Threshold = 6 ×Shared Foreign Policy -3.359
(2.510)
Threshold = 6 3.893*
(1.937)
Shared Foreign Policy 0.949 0.548 1.384 1.532 3.103 3.481+
(2.026) (2.009) (1.962) (1.941) (2.200) (1.990)
Alliance -0.420 -0.343 -0.347 -0.238 -0.147 -0.187
(0.555) (0.554) (0.543) (0.521) (0.588) (0.592)
Trade Dependence 4.309 4.078 2.479 1.461 -0.153 -0.560
(3.584) (3.630) (4.008) (4.242) (5.076) (4.898)
Civil War 1.156 0.839 0.614 0.861 1.366 1.652
(0.753) (0.759) (0.787) (0.807) (0.919) (1.141)
Capability Disparity -4.761* -4.834* -4.886* -4.824** -5.064** -5.200**
(1.920) (1.904) (1.899) (1.859) (1.816) (1.771)
Democracy -3.104* -3.017* -2.977* -2.762* -3.041** -3.079**
(1.306) (1.353) (1.246) (1.186) (1.081) (1.076)
Cold War 0.382 0.315 0.494 0.507 0.529 0.417
(0.928) (0.938) (0.973) (0.978) (0.955) (0.898)
Systemic Shock 1.210 1.315 1.413 1.443 1.366 1.199
(0.997) (1.005) (1.048) (1.038) (1.010) (0.970)
Ongoing Territorial Dispute 2.664** 2.815** 2.820** 2.668** 2.257* 2.157*
(0.869) (0.868) (0.976) (0.973) (0.948) (0.933)
Constant -4.398 -4.193 -4.841 -4.895 -5.693 -5.644
(3.648) (3.559) (3.703) (3.671) (3.883) (3.665)
Observations 36854 36854 36854 36854 36854 36854
Robust standard errors clustered on the dyad
Cubic polynomial of years of peace included in model but not reported
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001 in 2-tail test
45
Table 14: Logistic Regression 1950-2005 (Threshold Robustness - Trade Dependence)
(1) (2) (3) (4) (5) (6)
Threshold = 1 Threshold = 2 Threshold = 3 Threshold = 4 Threshold = 5 Threshold = 6
b/se b/se b/se b/se b/se b/se
Threshold = 1 ×Trade Dependence 17.764
(11.247)
Threshold = 1 1.514**
(0.469)
Threshold = 2 ×Trade Dependence 16.048
(10.976)
Threshold = 2 1.956***
(0.499)
Threshold = 3 ×Trade Dependence 18.398+
(10.939)
Threshold = 3 1.854***
(0.533)
Threshold = 4 ×Trade Dependence -2.162
(18.406)
Threshold = 4 1.937***
(0.568)
Threshold = 5 ×Trade Dependence 28.546
(27.044)
Threshold = 5 1.090
(0.789)
Threshold = 6 ×Trade Dependence -1766.735
(1610.176)
Threshold = 6 2.090
(1.305)
Trade Dep endence -0.191 -0.175 -1.884 1.193 -1.427 -0.652
(6.102) (5.925) (6.967) (4.101) (5.954) (4.788)
Shared Foreign Policy 3.125 3.122 3.337+ 3.300+ 3.381+ 3.460+
(1.954) (1.954) (1.967) (1.979) (1.885) (1.882)
Alliance -0.421 -0.336 -0.331 -0.253 -0.175 -0.199
(0.547) (0.545) (0.534) (0.524) (0.585) (0.586)
Civil War 1.151 0.862 0.682 0.935 1.473 1.758+
(0.745) (0.747) (0.779) (0.796) (0.903) (1.015)
Capability Disparity -4.736* -4.754* -4.815* -4.864** -5.046** -5.574**
(1.979) (1.953) (1.907) (1.862) (1.805) (1.860)
Democracy -2.852* -2.764* -2.845* -2.788* -3.012** -3.049**
(1.309) (1.334) (1.280) (1.167) (1.095) (1.082)
Cold War 0.379 0.316 0.486 0.515 0.504 0.398
(0.933) (0.935) (0.971) (0.968) (0.950) (0.901)
Systemic Shock 1.219 1.273 1.339 1.380 1.289 1.098
(1.005) (1.022) (1.061) (1.055) (1.032) (0.993)
Ongoing Territorial Dispute 2.509** 2.607** 2.688** 2.562** 2.275* 2.161*
(0.873) (0.867) (0.959) (0.950) (0.950) (0.933)
Constant -6.255 -6.414 -6.515+ -6.384 -5.876 -5.360
(3.929) (3.908) (3.920) (3.887) (3.643) (3.615)
Observations 36854 36854 36854 36854 36854 36854
multicolumn6lRobust standard errors clustered on the dyad
Cubic polynomial of years of peace included in model but not reported
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001 in 2-tail test
46
Table 15: Logistic Regression 1950-2005 (Threshold Robustness - Latent Shared Interests)
(1) (2) (3) (4) (5) (6)
Threshold = 1 Threshold = 2 Threshold = 3 Threshold = 4 Threshold = 5 Threshold = 6
b/se b/se b/se b/se b/se b/se
Threshold = 1 ×Shared Interests 13.159***
(3.972)
Threshold = 1 0.140
(0.876)
Threshold = 2 ×Shared Interests 14.469***
(4.083)
Threshold = 2 0.389
(0.906)
Threshold = 3 ×Shared Interests 13.720**
(4.609)
Threshold = 3 0.169
(1.027)
Threshold = 4 ×Shared Interests 13.383**
(4.652)
Threshold = 4 0.102
(1.065)
Threshold = 5 ×Shared Interests 19.638*
(9.704)
Threshold = 5 -1.694
(2.112)
Threshold = 6 ×Shared Interests 39.122**
(11.979)
Threshold = 6 -5.751*
(2.746)
Shared Interests -8.612+ -8.822* -5.931 -5.322 -4.255 -3.458
(4.494) (4.391) (4.244) (4.212) (4.054) (4.033)
Trade Dep endence 11.084* 10.249* 6.051 4.273 2.389 2.633
(5.078) (5.223) (5.683) (6.347) (7.516) (5.510)
Shared Foreign Policy 5.606** 5.464** 4.840* 4.801* 4.739* 4.587*
(2.175) (2.039) (1.985) (1.934) (1.989) (1.979)
Alliance -0.498 -0.444 -0.462 -0.388 -0.129 -0.042
(0.690) (0.678) (0.665) (0.660) (0.705) (0.678)
Civil War 1.177 0.834 0.680 0.936 1.537+ 2.127*
(0.751) (0.754) (0.767) (0.790) (0.886) (0.890)
Capability Disparity -4.786* -4.954* -5.013* -4.886* -5.567** -5.486**
(2.190) (2.104) (2.088) (2.053) (2.035) (1.996)
Democracy -3.119* -2.966* -2.862* -2.648* -3.136** -3.176**
(1.293) (1.335) (1.261) (1.235) (1.167) (1.095)
Cold War 0.147 0.004 0.190 0.256 0.259 0.376
(0.925) (0.937) (0.960) (0.960) (0.926) (0.914)
Systemic Shock 1.141 1.196 1.253 1.282 1.138 1.272
(1.010) (1.010) (1.040) (1.038) (1.008) (0.986)
Ongoing Territorial Dispute 2.526** 2.611** 2.732** 2.646* 2.244* 2.155*
(0.910) (0.913) (1.048) (1.042) (1.001) (0.954)
Constant -7.277* -7.058* -6.689* -6.768* -5.871+ -6.136+
(3.539) (3.348) (3.390) (3.318) (3.313) (3.286)
Observations 35823 35823 35823 35823 35823 35823
multicolumn6lRobust standard errors clustered on the dyad
Cubic polynomial of years of peace included in model but not reported
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001 in 2-tail test
47
Table 16: Logistic Regression 1950-2005 (Consensus Rivalries)
(Revolution Only) (Shared Foreign Policy) (Alliance) (Trade Dependence) (Latent Interests)
ˆ
β/se ˆ
β/se ˆ
β/se ˆ
β/se ˆ
β/se
Revolution 1.430** -3.498 0.373 1.293* -0.696
(0.527) (3.067) (0.770) (0.551) (0.963)
Revolution ×Shared Foreign Policy 5.561+
(3.342)
Revolution ×Alliance 3.294*
(1.420)
Revolution ×Trade Dependence 19.430
(11.981)
Revolution ×Shared Interest 15.603***
(4.088)
Shared Interest -5.082
(4.820)
Shared Foreign Policy 2.969 1.825 3.129+ 2.978 4.539*
(1.860) (1.926) (1.834) (1.862) (2.042)
Alliance -0.399 -0.410 -2.198+ -0.405 -0.636
(0.607) (0.607) (1.248) (0.605) (0.824)
Trade Dependence 0.872 1.379 3.157 -2.545 4.207
(4.219) (4.200) (4.814) (6.851) (6.583)
Civil War 1.099 1.053 1.425+ 1.104 1.090
(0.783) (0.791) (0.746) (0.798) (0.773)
Capability Disparity -5.092** -5.035** -4.942* -4.990* -5.191*
(1.948) (1.954) (1.938) (1.957) (2.161)
Democracy -2.306* -2.324* -2.040+ -2.198+ -2.277*
(1.106) (1.114) (1.143) (1.160) (1.130)
Cold War 0.547 0.540 0.506 0.538 0.273
(0.952) (0.970) (0.962) (0.964) (0.977)
Systemic Shock 1.024 1.067 1.114 1.035 0.981
(1.091) (1.098) (1.081) (1.098) (1.091)
Ongoing Territorial Dispute 1.851+ 1.940+ 2.393* 1.871+ 1.948+
(1.036) (1.046) (1.102) (1.044) (1.070)
Constant -5.727 -4.791 -5.813 -5.784 -6.114+
(3.815) (3.741) (3.729) (3.839) (3.397)
Observations 36854 36854 36854 36854 35823
multicolumn5lRobust standard errors clustered on the dyad
Cubic polynomial of years of peace included in model but not reported
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001 in 2-tail test
48
High Interest Only Samples
Table 17: Logistic Regression 1950-2005 (Shared Interests Sample Only)
(1) (2) (3) (4)
Alliance U.N. Voting 75th% Trade 75th% Combined Sample
Revolution 2.9704*** 2.5332*** 2.0929* 2.2253***
0.8648 0.8019 0.9298 0.4456
Contiguity -0.1307 -0.3063* -0.2865+ -0.2149
0.2633 0.1324 0.2189 0.1900
Capability Disparity -4.3152+ 1.4827 -6.8133* -3.1651+
2.8067 2.3666 3.5379 2.3590
Democracy -4.9004* -3.8427* -4.5898* -3.1154**
2.2139 1.9758 2.2262 1.0965
Systemic Shock 0.3191 2.1126* * 0.3927
1.5111 0.9085 0.6542
Ongoing Territorial Dispute 2.5733+ 2.1541+ 4.2655*** 2.2929**
1.6124 1.6227 1.0980 0.8583
Cold War 0.1052 1.2974 -0.1746 0.0773
1.2446 1.0208 0.8288 0.5201
Major Power in Dyad 0.9373 -1.4241 1.1435 -0.2688
1.3235 1.1243 1.5180 0.7895
Peace Years 0.0396 -0.0493 -0.0205 0.0126
0.0423 0.0947 0.0681 0.0290
Peace Years2-0.0009 0.0027 0.0000 -0.0004
0.0012 0.0038 0.0013 0.0006
Peace Years30.0000 -0.0000 -0.0000 0.0000
0.0000 0.0000 0.0000 0.0000
Constant -3.8499+ -7.6729** -0.0828 -3.4590*
2.9127 2.6932 2.3835 1.8725
Observations 8681 8967 14337 22528
Robust standard errors clustered on the dyad
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001 in 2-tail test
* Systemic Shock is collinear with outcome in this sample.
49
Necessary Condition Testing:
I assess the necessity of high levels of shared interest for domestic political changes to
cause rivalry using a methodology developed by Braumoeller and Goertz (2000). I limit the
sample under observation to only the 74 cases of rivalry onset that occurs between 1950
and 2005. I then assess whether rivalry onset occurs after revolution only in cases where
high levels of shared interest are present. To satisfy the criteria proposed by Braumoeller
and Goertz, high levels of shared interest must be present in every case in which domestic
political change leads to rivalry. As Braumoeller and Goertz point out, one counter-case
does not necessarily refute a claim regarding necessary conditions. Measurement error in the
variables of interest can introduce false counter-examples. Thus, it must be demonstrated
that any case in which rivalry occurs after a revolution but in the absence of high shared
interests is likely a case of measurement error in regard to shared interests. I consider high
levels of shared interest to exist in a dyad in any case where the states share an formal defense
pact, and any case in which at least one state’s measure of trade dependence on or foreign
policy similarity with the other falls into the highest quartile of that state’s relationships
with all other states for the year prior to revolution or SOLS change. The two tables below
present the relationship between shared interests and revolutions prior to rivalry formation.
Table 16: Rivalry Onsets 1950-2005
Low Shared Interest High Shared Interest
No Prior SOLS Change 38 11
Prior SOLS Change 2 23
Table 17: Rivalry Onsets 1950-2005
Low Shared Interest High Shared Interest
No Prior Revolution 38 17
Prior Revolution 2 17
A total of 2 counter-cases exist across the two measures of domestic political change:
Iran-Egypt 1980
Iran-Saudi Arabia 1980
The cases involving Iran seem likely cases of measurement error. Both cases narrowly
miss the cutoff to be included as high-interest on both the trade dependence and shared
foreign policy measures. The Iran-Egypt and Iran-Saudi dyads fall into the 70th and 63rd
percentile in terms of trade dependence, respectively. Additionally, an assumption that one’s
"friends" are those whom one agrees with more often than one agrees with the average state
(i.e. if both states are in the third or fourth quartile in terms of join U.N. voting) would
remove both cases as counter-examples. Further, none of the measures of shared interest
proposed here directly include the Middle Eastern religious divide which provided the Shah
and Saudi and Egyptian leaders with a shared interest in suppressing Shi’ite political power.
Historical evidence also suggests that the mid-to-late 1970’s were a period in which Iran
50
sought increasingly closed ties with Egypt and Saudi Arabia in hopes of building a security
community capable of resisting great power influence (Ramazani, 1976). Iran provided pilots
and planes to Saudi Arabia during the 1973 Arab-Israeli war and in 1974 Iran signed a "billion
dollar protocol" with Egypt to finance the rebuilding of Port Said and widening of the Suez
Canal after the war. Iran then provided full support to Anwar Sadat in his attempts to
resolve the Arab-Israeli conflict (p. 175-176). Similarly, after the British withdrawal from
the Persian Gulf, Iran sought out cooperation with Saudi Arabia, giving up its territorial
claims to Bahrain and resolving Iranian-Saudi disputes over the continental shelf (p.176).
Saudi-Iranian relations appear to have been at a high point in the lead up to the Iranian
revolution. Thus, it is difficult to conclude that these counter examples are clearly cases in
which the states in question do not share strong joint interests.
I argue that these findings provide qualified support for the belief that major domestic
political changes will only lead to rivalry in the presence of high ex-ante shared interests.
When states have little invested in one another, a revolution or other major political change
in one state has little effect on the other, thus should not be thought to trigger conflict.
References
Ramazani, R. K. (1976). Iran’s Search for Regional Cooperation. The Middle East
Journal, 173-186.
51
Table 18: Logistic Regression of Rivalry Initiation 1950-2005: Dropping dyads containing Iran
(1) (2) (3) (4) (5)
Revolution Foreign Policy Alliance Trade Latent Interests
b/se b/se b/se b/se b/se
Revolution 1.808*** -3.767 0.530 1.661** 0.454
(0.543) (2.709) (0.758) (0.557) (0.987)
Revolution ×Policy 6.277*
(2.908)
Revolution ×Alliance 3.858**
(1.465)
Revolution ×Trade 19.705+
(11.323)
Revolution ×Shared Interests 9.953**
(3.813)
Shared Interests -2.123
(2.309)
Alliance -0.353 -0.374 -2.614* -0.348
(0.532) (0.532) (1.274) (0.529)
Policy 2.869+ 1.172 2.896+ 2.891+
(1.672) (1.681) (1.674) (1.660)
Trade 1.241 1.886 4.256 -3.104
(4.244) (4.221) (4.722) (8.079)
Civil War 1.048 0.999 1.386+ 1.050 1.059
(0.771) (0.779) (0.716) (0.777) (0.725)
Capability Disparity -5.601* -5.461* -5.428* -5.470* -5.146*
(2.530) (2.433) (2.461) (2.549) (2.398)
Democracy -2.914* -2.928* -2.508* -2.794* -2.541*
(1.283) (1.282) (1.239) (1.358) (1.125)
Contiguity 0.166 0.287 0.202 0.165 0.750
(0.949) (0.951) (0.985) (0.963) (1.067)
Cold War 0.412 0.351 0.239 0.402 0.023
(0.855) (0.898) (0.851) (0.864) (0.832)
Systemic Shock 0.966 0.958 1.014 0.970 0.854
(0.910) (0.938) (0.932) (0.910) (0.898)
Ongoing Territorial Dispute 2.707** 2.839** 3.385** 2.725** 2.860**
(0.952) (0.982) (1.153) (0.972) (1.105)
Constant -5.563 -4.256 -5.434 -5.633 -3.341
(4.015) (3.851) (3.778) (4.056) (2.887)
Observations 36419 36419 36419 36419 35400
Robust standard errors clustered on the dyad –in parentheses
Cubic polynomial of years of peace included in model but not reported
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001 in 2-tail test
52
Table 19: Logistic Regression of Rivalry Initiation 1950-2005: Dropping Major Power/Minor Power Dyads
(1) (2) (3) (4) (5)
Revolution Foreign Policy Alliance Trade Latent Interests
b/se b/se b/se b/se b/se
Revolution 1.808*** -4.564 0.827 1.582** 1.720**
(0.497) (3.118) (0.735) (0.545) (0.588)
Revolution ×Policy 7.117*
(3.403)
Revolution ×Alliance 2.931*
(1.288)
Revolution ×Trade 28.838*
(14.105)
Revolution ×Shared Interests 6.751**
(2.491)
Shared Interests -1.927+
(0.990)
Policy 4.233* 1.790 3.970* 4.366*
(1.684) (1.616) (1.744) (1.702)
Alliance -0.338 -0.349 -2.065+ -0.299
(0.531) (0.533) (1.095) (0.533)
Trade 1.106 1.641 5.348 -7.592
(5.658) (5.536) (4.679) (9.900)
Civil War 0.717 0.646 1.026 0.824 0.890
(0.826) (0.836) (0.790) (0.876) (0.777)
Capability Disparity -2.513 -2.280 -2.107 -2.473 -2.152
(1.662) (1.681) (1.693) (1.671) (1.752)
Contiguity -3.710*** -3.021*** -3.451*** -3.822*** -0.741
(0.843) (0.809) (0.828) (0.879) (0.792)
Ongoing Territorial Dispute * * * * *
(.) (.) (.) (.) (.)
Democracy -3.739* -3.623* -3.858* -3.772* -1.974+
(1.716) (1.622) (1.570) (1.843) (1.196)
Cold War -0.159 -0.661 -0.322 -0.193 -0.535
(0.632) (0.642) (0.729) (0.652) (0.777)
Systemic Shock 0.747 0.299 0.677 0.721 0.617
(0.693) (0.652) (0.763) (0.719) (0.765)
Constant -4.456* -2.695 -4.298* -4.436* -3.779**
(1.966) (1.740) (1.966) (1.957) (1.318)
Observations 6309 6309 6309 6309 6119
Robust standard errors clustered on the dyad –in parentheses
Cubic polynomial of years of peace included in model but not reported
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001 in 2-tail test
*: Territorial Dispute is perfectly collinear with the outcome in this sample.
53
Table 20: OLS of Rivalry Initiation 1950-2005
(1) (2)
Count of Rivalries Count of Rivalries
b/se b/se
Revolution 0.050**
(0.016)
SOLS Change 0.019*
(0.009)
Military Capability 0.235*** 0.241***
(0.034) (0.035)
Polity -0.001** -0.001**
(0.000) (0.000)
Territorial Dispute -0.006+ -0.006+
(0.003) (0.004)
Civil War 0.009 0.011
(0.008) (0.009)
Coldwar 0.004 0.006+
(0.003) (0.003)
Systemic Shock -0.001 -0.001
(0.003) (0.003)
Constant 0.008** 0.008**
(0.003) (0.003)
Observations 7210 7210
Robust standard errors clustered on the country –in parentheses
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001 in 2-tail test
Dependent Variable:
Count of rivalries that the state enters into in a given year.
Independent variables:
All control variables are lagged 6 years to ensure that they are measured prior to the 5-year
post-revolution observation window.
Revolution: Defined as in the primary analysis. Takes the value "1" if a revolution has
occurred according to Jeff Colgan’s definition (2010) in the state in question within the last
five years.
Military Capability: Country-year measure of Correlates of War CINC score.
Polity: Country-year measure of Polity2 for the PolityIV dataset.
Territorial Dispute: Takes the value 1 if this country is involved in at least 1 territorial
dispute in the given year according to the Issue Correlates of War dataset.
54
Civil War : Takes the value 1 if this country is involved in a civil war according to the
Correlates of War Intrastate Wars dataset.
Coldwar: Takes the value 1 for years 1950-1991
Systemic Shock: Takes the value 1 for years 1950-1955 (World War 2, system shock) and for
years 1992-2002 (Post Cold-War system shock).
55
ResearchGate has not been able to resolve any citations for this publication.
Book
This book provides the first detailed analysis of international rivalries, the long-standing and often violent confrontations between the same pairs of states. The book addresses conceptual components of rivalries and explores the origins, dynamics, and termination of the most dangerous form of rivalry--enduring rivalry--since 1816. Paul Diehl and Gary Goertz identify 1166 rivalries since 1816. They label sixty-three of those as enduring rivalries. These include the competitions between the United States and Soviet Union, India and Pakistan, and Israel and her Arab neighbors. The authors explain how rivalries form, evolve, and end. The first part of the book deals with how to conceptualize and measure rivalries and presents empirical patterns among rivalries in the period 1816-1992. The concepts derived from the study of rivalries are then used to reexamine two central pieces of international relations research, namely deterrence and "democratic peace" studies. The second half of the book builds an explanation of enduring rivalries based on a theory adapted from evolutionary biology, "punctuated equilibrium." The study of international rivalries has become one of the centerpieces of behavioral research on international conflict. This book, by two of the scholars who pioneered such studies, is the first comprehensive treatment of the subject. It will become the standard reference for all future studies of rivalries.
Article
United Nations (UN) General Assembly votes have become the standard data source for measures of states preferences over foreign policy. Most papers use dyadic indicators of voting similarity between states. We propose a dynamic ordinal spatial model to estimate state ideal points from 1946 to 2012 on a single dimension that reflects state positions toward the US-led liberal order. We use information about the content of the UN’s agenda to make estimates comparable across time. Compared to existing measures, our estimates better separate signal from noise in identifying foreign policy shifts, have greater face validity, allow for better intertemporal comparisons, are less sensitive to shifts in the UN’ agenda, and are strongly correlated with measures of liberalism. We show that the choice of preference measures affects conclusions about the democratic peace.
Article
This article describes the first complete release of the Issue Correlates of War (ICOW) Territorial Claims dataset, which covers all interstate territorial claims between 1816 and 2001. Territory can have substantial tangible and intangible value for states, and competing claims for control of territory represent one of the leading sources of interstate conflict. The dataset identifies 843 territorial claims and includes measures of the salience of the claimed territory, as well as details of the militarization and ending of each claim. Beyond a discussion of the structure and contents of the dataset and the coding procedures that were used to generate it, this article also presents descriptive analyses of the dataset. These analyses highlight important patterns across time and space, including changes in the prevalence, frequency of initiation, salience, militarization, and resolution of territorial claims. Notable patterns include recent declines in the frequency with which claims tend to become militarized and a lower prevalence of tangible salience measures such as natural resources. The regional distribution of claims has also shifted markedly over time, from a historical concentration in Europe towards Asia, where by 2001 claims were far more prevalent than in any other region. The article concludes with suggestions for future research.
Article
Signals from domestic and international actors have been shown to influence the likelihood of coups. Coups remain difficult to predict and consequently leave policy makers in a reactive stance, but little systematic work assesses how these reactions influence long-term outcomes. We examine how reactions from domestic and international actors influence the duration of coup-born regimes, arguing that negative reactions will shorten leadership duration. We further probe these relationships by considering how signaling consistency, Cold War dynamics, and precoup relationships condition the influence of reactions on leadership duration. Tests use events data to capture domestic and international reactions and newly coded information on leadership to capture leader duration. Results indicate that international responses have a profound influence on leadership tenure, especially those from strong actors. We find tentative support that state reactions have the strongest effect during the Cold War, while international organizations matter the most afterward.
Article
This study examines the construction of a new political order in post-revolutionary Iran through the prism of its revolutionary armed forces. It specifically explores the place of the Islamic Revolutionary Guards Corps (IRGC) in this process, and focuses on its role in establishing and maintaining state control. Unlike the regular armed forces, the IRGC was given the additional mandate of ???safeguarding the revolution.??? Exercising this broad responsibility has led to the gradual permeation of the organization???s influence in all sectors of the Iranian state, including the areas of domestic security, ideological promotion, cultural work, industrial development, foreign engagements, and politics. This study argues that military power has been central to the shaping of post-revolutionary Iran, and that the proliferation of military power is directly related to the various conflicts that have and continue to occur in post-revolutionary Iran. These claims are explored within a framework established in recent scholarship on unconventional armed forces and their roles in state formation and politics. Such scholarship has convincingly shown that the roles of unconventional or irregular armed forces in the political development of states have been underappreciated in past scholarship. As military power has been a crucial factor in the making of the Islamic Republic, and as the revolutionary institutions that comprise the core of military power in Iran do not fit the conventional mold, the theoretical framework provided by recent scholarship on irregular armed forces is a valuable tool for understanding the various implications of military influence on the Iranian state. This study also addresses some of the gaps in the scholarship on irregular armed forces and politics by emphasizing the cultural, ideological, and religious dimensions of military power in an Islamic society of the Middle East. By viewing irregular armed forces in Iran as significant political actors, a more coherent and historical perspective about how military power operates within society and why its prevalence continues in Iran is demonstrated.
Article
This article introduces the CHISOLS (Change in Source of Leader Support) dataset, which identifies which leadership changes within countries bring to power a new leader whose primary support is drawn from different societal groups than those who supported her predecessor. The dataset covers all countries of the world with populations greater than 500,000 from 1919 to 2008. We discuss the underlying rationale of our data collection, provide some brief information about the coding rules and procedures, and share some descriptive statistics. We find that changes in sources of leader support are more common in democracies than non-democracies, but also that changes in sources of leader support often occur without irregular leader transitions or large institutional changes, even within non-democracies. CHISOLS can be productively combined with other datasets like POLITY, Archigos, and DPI that provide information about political institutions, modes of leader transition, and placement on a left–right policy continuum, but CHISOLS also provides something new that was not previously available. These data allow researchers to study the extent to which different types of policy change are associated with all leader transitions, with changes in political institutions, or with changes in the set of interests that leaders most closely represent; CHISOLS facilitates comparing the effect of leaders, interests, and institutions on policy change across a wide spatial temporal domain.