Leadership transition, ally restraints, and target state resistance

Understanding how target states respond to international provocations is crucial to the study of crisis management and conflict onset. The primary focus of existing research is on the individual effects of ally influence and domestic politics without considering their interaction. This study addresses this gap by exploring the combined impact of leadership transitions, manners of leader entry, and ally’s importance on target states’ decisions to resist challenges. Utilizing the Truly Dyadic Dyad-Year (TDD) version of the Militarized Interstate Confrontation (MIC) dataset from 1920 to 2014, the Archigos leadership dataset, and logit models for statistical testing, our findings reveal that target states’ leadership transitions and irregular means of leaders entry significantly increase the likelihood of target states’ resistance, particularly when leaders seek to consolidate power. Moreover, although the support of a powerful ally may embolden the target state to resist, the restraining effect of allies during the target state’s leadership transitions becomes more pronounced. Our research highlights the intricate balance between domestic political considerations and external influences, emphasizing that an understanding of the domestic political context is essential for comprehending how states respond to international provocations and the effectiveness of ally restraints.

1 Introduction

The relationship between alliances and the outbreak of war has long been a central concern in the study of interstate conflict. Despite extensive research highlighting the function of alliances in preventing military conflicts, existing studies fall short of fully accounting for the complex causal mechanisms of domestic political factors, particularly the impact of leaders’ decisions.

Traditional research suggests that the impact of alliances on the likelihood of potential challengers initiating militarized interstate disputes primarily due to the expectations of the challenger’s decision-maker (Johnson, 2016). Furthermore, states can enhance the capability and credibility of their extended deterrence threats through strategic alliance design (Johnson et al., 2015). Recent research has begun to focus on the control that allies exert over the target state during conflicts. In international disputes, allies may sometimes leverage their influences to reduce the likelihood that the target state escalates the conflict, as they may seek to avoid being drawn into militarized disputes due to the target state’s actions (Johnson and Leeds, 2011). As Fang et al. (2014) observed, an ally might issue a credible threat that if the target refuses to comply with the demands deemed reasonable by the ally, support in the conflict may be withdrawn, thereby pressuring the target to make concessions to prevent escalation.

However, existing literature frequently overlooks domestic political factors in shaping the target state’s decision-making. This is particularly relevant in the context of leadership transitions, where the target state’s leader may face pressures to consolidate power, affecting their receptiveness to allies’ advice. Additionally, the “rally ‘round the flag” effect suggests that public support tends to increase significantly when the state faces external threats or enters into a state of war (Mueller, 1970).

Existing studies have established that leaders are pivotal causal factors in the initiation of international conflicts (Bak, 2020; Gelpi and Grieco, 2001). Chiozza and Goemans (2011) have argued that leaders, particularly those facing a potential loss of power, may exhibit tougher external stances. Therefore, this research seeks to address the question: how do allies and leadership factors conjointly influence a target state’s decision-making process when confronted with an adversary’s challenge? By examining interactions among the challenger, the target state, and its allies, this study aims to elucidate potential answers from the perspectives of the target state’s leadership dynamics and the critical role of alliances.

The first section of this paper reviews diverse perspectives on ally restraint, target state’s leadership, and target resistance. The second section presents the theories and hypotheses, establishing the basis for the empirical analysis that follows. The third section outlines the research design, detailing the methods and data used to test the hypotheses. The fourth section presents the results of the analysis. The fifth section carries out robustness checks, incorporating alternative model specifications and samples, as well as applying alternative measures for both the dependent and independent variables. Finally, the conclusion synthesizes and highlights the main findings of this study.

2 Ally restraint, leadership, and target resistance

2.1 Alliance: emboldening or restraint effect

Earlier studies on alliances and crises posit that alliance commitments act as costly signals that reduce the risk of conflict by deterring potential challengers (Morrow, 1994, 2000; Smith, 1995). The inherent costs of breaching such commitments lessen the likelihood of allies abandoning their commitments to target states (Leeds and Savun, 2007). But what role does the ally play when deterrence fails and the target state finds itself embroiled in a crisis with a challenger? The literature in this regard diverges into two primary camps: one that emphasizes the emboldening effect; the other that focuses on the restraint effect.

Some scholars have expressed concerns that having the support of allies may, by virtue of their optimistic anticipation of victory, embolden states to escalate conflicts (Smith, 1995). Security commitments, therefore, may inadvertently spark conflicts by emboldening allies, so increasing the probability of wars against non-allied states (Benson et al., 2013; Krebs, 1999). The existing literature further suggests that alliances and defense pacts may be instrumental in a state’s decision to join an ongoing conflict if one of the belligerents is its ally (Leeds, 2003). Morgan and Palmer (2003) have observed that, upon joining an alliance, a state may adopt a more proactive foreign policy and step up its military expenditure, so ultimately increasing the possibility of conflict. Christensen, moreover, through analyzing the extent of alliances’ internal cohesion and the coercive diplomacy of challengers, finds that any divergence in the respective interests or opinions of the ally and target may render regional security somewhat fragile (Christensen, 2011).

Other scholars, argue that allies may exercise restraint on their partners by threatening to weaken or terminate the alliance (Zagare and Marc Kilgour, 2003). Alliances, therefore, may facilitate peaceful resolution of intra-allied conflicts, and prevent members from engaging in reckless behavior (Gelpi, 1999). Moreover, alliance commitments often incorporate escape clauses, or conditions that allow members a certain flexibility regarding the circumstances where under which they render assistance (Leeds, 2003). In asymmetric alliances, for instance, stronger states may limit their commitments to weaker members (Mattes, 2012), while weaker allies may relinquish some of their autonomy in order to gain security protection (Morrow, 1991). An ally’s influence is significantly enhanced when its partners lack alternative alignment options (Crawford, 2003).

That an ally can exert both an emboldening effect—by encouraging the target state to resist disputes—and a restraining effect—by credibly threatening to withhold intervention unless the target state concedes to reasonable demands—is evident (Fang et al., 2014). The inherent contradiction, therefore, lies in whether alliances ultimately increase the risk of conflict by emboldening the target state or reduce the risk of conflict by restraining their partners.

2.2 Ally’s importance, leader and target’s decision

To resolve the contradiction between the emboldening effect and the restraining effect, earlier studies placed primary emphasis on the importance of alliances. If an ally should recommend concession, the target state’s decision on whether or not to follow that suggestion and maintain positive relations may depend on its assessment of the alliance’s importance (Fang et al., 2014). In more recent years, scholars have reflected on the overemphasis of external factors, observing that international interaction of any kind may be described as a comprehensive negotiation involving both the international and domestic arenas. Similarly, a target state’s decision whether or not to resist external threats is influenced by a combination of domestic and international factors, particularly the policy choices of its leader. Despite being acknowledged as a key causal factor in the outbreak of conflict, however, the role of the leader in studies examining the relationship between allies and conflict has been somewhat neglected.

Given that a leader’s legitimacy is fundamentally rooted in domestic politics, it is reasonable to assume that political survival exerts more decisive influence on the target’s resistance than most other international factors, including allies. Earlier studies assume that the policies leaders choose are those that best demonstrate their capability by virtue of increasing domestic political support and thus maximizing their tenure (Levy, 1989; Leeds and Davis, 1997; Bueno de Mesquita et al., 2003; Mansfield and Snyder, 2005). Scholars have also considered leaders’ military experience as a potentially influential factor in regard to conflict (Schultz, 2001; Schultz, 2005; Palmer et al., 2004; Horowitz et al., 2015). By examining the empirical record of target state choices, therefore, this study aims to provide detailed hypotheses based on the theoretical analysis above.

It is essential to elucidate why political survival, or maximizing tenure, plays such a decisive role in the target’s choice to resist. The sponsors’ initiation of challenges provides target state leaders, especially those facing a political existential crisis, with excellent diversionary opportunities. On one hand, the initiator’s provocation is likely to fully mobilize domestic public sentiment in the target state, so enabling its leader to leverage the “rally ‘round the flag” effect (Gaubatz, 1991) and garner public support. On the other hand, successfully resisting the challenger allows leaders to demonstrate their competence, accumulate political prestige, and suppress domestic opposition—a phenomenon known as “gambling for resurrection.”

Of course, the expected term of leadership is itself the result of the combined effects of domestic and foreign factors, but we can nonetheless attempt to address endogeneity here in chronological order. Existing literature has proposed several indicators whereby to measure directly the intensity of domestic crises, namely, the degree of economic recession, the number of political protests (Kisangani and Pickering, 2011), and casualties resulting from domestic unrest. Additionally, two factors—competence and legitimacy—may significantly affect leaders’ sensitivity to domestic crises. When faced with similar domestic crises, leaders lacking competence and legitimacy are those most likely to step down. For example, regarding the winning coalition’s perception of competence, those that have only recently attained governance, in comparison with leaders that have long been in power, have scant opportunity to prove themselves, so may seek a means to rapid elevation of their political prestige. New leaders of countries without formal institutions to ensure the peaceful transfer of leadership power face the constant threats of coups by opposing groups (Chiozza and Goemans, 2011). Such immense pressure on their political survival motivates their “gambling for survival.”

Regarding legitimacy, this is something that leaders who have attained power through unconstitutional means, such as coups or foreign interventions, tend to lack. Such weakness of fundamental legitimacy may amplify the political repercussions of failing to address domestic crises, particularly in countries that frequently experience irregular leadership turnovers. Due to the absence of institutional safeguards, the severe consequences of stepping down that leaders may face motivates their desperate measures to retain power (Chiozza and Goemans, 2011). Recent leadership transitions, therefore, significantly increase the pressure on leaders of political survival.

3 Theories and hypotheses

As discussed above, the allies’ emboldening effect and restraining effect, as well as leadership factors, may all influence the target state’s decision to resist. The central question this paper addresses, therefore, is that of which factor is the more decisive, and under what conditions the influence of allies becomes significant and unavoidable.

Previous studies on extended deterrence have emphasized the importance of allies (Fang et al., 2014). An ally’s ability to effectively restrain a target state depends largely on that state’s decision-makers. Evidently, numerous factors, including the bargaining power of allies, as well as domestic attributes, determine the direction and extent of an ally’s influence on the target state. In a crisis situation, the target state might underscore the importance of a strong ally. Concurrently, to avoid military entanglement, that ally might choose to restrain the target (Snyder, 1984). If, however, the target state perceives such restraint as contradictory to its vital interests, it may not necessarily comply with the ally’s directives during a crisis. Israel’s stance in the Six-Day War serves as an illustrative case. The decision of a target state to stand firm against a challenger, even at the expense of violating the ally’s wishes, indicates that the ally’s importance is not the sole determining factor in regard to the target’s choice during heightened tensions.

Domestic political mechanisms, such as audience costs and leadership factors, however, may play even more crucial roles than allies’ importance. A substantial body of research indeed emphasizes the motivations of leaders, rather than those of states (Horowitz and Stam, 2014; Fuhrmann and Horowitz, 2015; Danneman and Ritter, 2014). Consequently, target state leaders’ decision-making process, particularly in regard to military resistance to provocations from challenger states, often results from a combination of domestic audience factors and ally considerations. Scholars have consistently posited that leaders are inclined to select policies that support their tenure (Bueno de Mesquita et al., 2003; Colaresi, 2004; Mansfield and Snyder, 2005; Lai and Slater, 2006). Unlike challengers that may maximize “rally ‘round the flag” effect by choosing the timing and location of conflicts, targets in international conflicts face higher risks of both regular and irregular removal from office, limited as they are to a choice between standing firm and backing down.

Thus, target leaders can predict the likelihood of victory through careful calculation of their opponents’ strength and determination by observing signals (Beardsley and Lo, 2014). On the one hand, the weakness and resultant concessions of leaders in international crises are detrimental to expanding domestic support. Audience cost theory proposes that governments are vulnerable to replacement by opposition parties if the domestic public perceives them as likely to back down during crises (Fearon, 1994; Schultz, 1998, 1999, 2001; Eyerman and Hart, 1996; Partell and Palmer, 1999). Moreover, when leaders believe that success on the international stage can fend off imminent electoral failure, they tend to use force abroad (Downs and Rocke, 1994; Chiozza and Goemans, 2011).

Therefore, target state leaders must consider the potential reaction of public opinion when deciding whether or not to concede (Licht and Allen, 2018). Particularly when leadership transitions occur within the target state, leaders are likely to adopt strong resistance measures against challenger states, thus either gambling for resurrection or consolidating domestic support, thus to secure a longer tenure. In such scenarios, target state leaders are likely to resist provocations from challengers. Therefore, we propose hypothesis 1 as follows.

Hypothesis 1: In the event of a leadership transition, target states tend to resist provocations from challengers in order to strengthen their domestic political support.

On the other hand, however, in opting for military resistance policies, leaders risk punishment in the event of military failure—generally the more serious consequence, because external failure renders them culpable leaders subject to domestic audience punishment. Therefore, if leaders fear the political penalties for military failure, ally support becomes crucial to their assessment of whether or not they face military defeat, especially for states that are heavily dependent on their allies. If the risk of a leader’s removal is high, therefore, they may exercise greater caution when evaluating the potential outcomes of initiating war, bearing in mind that victors have re-election opportunities, while losers are likely to be summarily ousted (Goemans, 2008).

There hence exists an interaction between a leader’s political survival and the influence of alliances. When the political stability of the target state leader is strong, they are less concerned about political punishment for military failure and more inclined to take firm military actions through ally support to gain domestic support and enhance their reputation. In such cases, the emboldening effect of alliances is more pronounced due to the external support of a strong ally which bolsters the target state leader’s confidence sufficiently to resist militarily. Leaders are more wary about the risk of failure in taking military action when their domestic support is weak or unstable. Fear of domestic punishment thus drives target leaders to seek ally support, thus to avoid defeat in conflicts with challengers (Croco, 2011). If, at such times, the ally, worried about entanglement, restrains the target, especially when that state is highly dependent on it, the leader of the target state must give due priority to the ally’s stance and continued support. In this scenario, therefore, the restraining effect of the ally is more apparent. Therefore, we propose hypothesis 2 as follows.

Hypothesis 2: Alliances exhibit both emboldening and restraining effects, but their specific impacts depend on the domestic politics of the target state, especially factors related to the leader. When the leader is politically stable, a strong ally increases the likelihood of resistance. But when the target state has undergone leadership transition and is thus highly dependent on the ally, it is likely to heed the ally’s call for restraint and to make concessions.

An institutionalized political system of regular procedures plays a crucial role in shaping a leader’s assessment of the domestic audience costs and possibility of political survival (Croco and Weeks, 2016). A leader that comes to power in an irregular manner is likely to stand firm in conflict decision-making, thus to heighten their public reputation. Constitutional entry into power, however, sends a more credible signal to the audience. Citizens of target states are unlikely to support leaders whose power has been attained either through irregular means or those that other states have imposed. Therefore, new leaders who have unconstitutionally attained power must face and deal with the dissent potentially emanating from the opposition party in regard to enacted policies (Licht and Allen, 2018).

We believe that leaders who come to power through irregular means often face legitimacy scrutiny both domestically and internationally. To address these legitimacy deficits, they may resort to military resistance against international challengers. Due to audience costs and the need for public support, engaging in international conflicts can generate a “rally ‘round the flag” effect, temporarily increasing public support and diverting public attention from domestic issues. This can help leaders consolidate their governance by demonstrating their capability to defend the nation’s sovereignty and interests, thereby gaining public approval and reducing internal opposition. Besides, in terms of demonstrating strength and image, by resisting international challengers, these leaders can project an image of strength and decisiveness. This perception is crucial for maintaining leaders’ authority and deterring both domestic and international adversaries.

Therefore, if leaders believe that the benefits of demonstrating strength outweigh the risks of potential military defeat and that inaction or appeasement might embolden both domestic opponents and international challengers, further undermining their authority, then taking a strong stand against international adversaries will be seen as a necessary gamble to maintain and potentially to enhance their position. We believe that leaders who attain power through coups or external intervention inevitably lack legitimacy, and therefore public support. As a result, they are acutely aware of domestic audience costs and may seek to bolster public support through assertive foreign policies. We propose hypothesis 3 as follows.

Hypothesis 3: Target leaders who have attained power through unconstitutional means will likely resist challengers militarily because, acutely aware of audience costs, they are anxious to gain public support.

4 Research design

4.1 Data description

To judge whether or not a leader’s tenure helps us to differentiate between an ally’s restraining or emboldening effect on the target, we test the related data and present the results of a statistical analysis. The basic intuition is that, when deciding whether to concede or resist, the target’s leader must choose between the expected profit of their tenure and the cost of damaging the alliance. Therefore, we have constructed a dataset of disputed directed dyad-years covering the period 1920 to 2014, in which a challenger initiated a dispute against a target whose allies are committed to supporting it in a conflict. To test the hypothesis, the data analyzed in this study are generated from the Militarized Interstate Confrontation (MIC) dataset (Gibler and Miller, 2024), the dyadic version of the Alliance Treaty Obligations and Provisions (ATOP) project (Leeds et al., 2002), and the Archigos dataset (Goemans et al., 2009) on leaders. Each observation consists of a directed dyad between a challenger state and a target state in a given year, derived from the Truly Dyadic Dyad-Year version of the Militarized Interstate Confrontation (MIC) dataset (Gibler and Miller, 2024). This enhances the accuracy of dyadic conflict analysis by eliminating the false positive indicators present in previous datasets (Gibler and Miller, 2024). We also used Militarized Interstate Participants (MIP) data (Gibler and Miller, 2024) to help distinguish between challenger and target states, thus ensuring precise identification and analysis. The ATOP project is used here to establish the impact of allies on target states. The Archigos dataset, moreover, contains information on whether or not a leadership change occurred in the target country in a given year, and the ways in which leaders attained office.

4.2 Dependent variables

According to the hypothesis, the dependent variable in this study is that of whether or not the target state will resist the challenger state when facing potential international conflicts. This variable is binary, and its data source is the Truly Dyadic Dyad-Year version of the Militarized Interstate Confrontation (MIC) dataset (Gibler and Miller, 2024), which comprises non-directed dyad-year data from the period 1816 to 2014. Building on this, we used Militarized Interstate Participants (MIP) data (Gibler and Miller, 2024) to distinguish between challenger and target states in conflicts, thereby generating directed dyadic data of challenger-target dyads from 1920 to 2014. If the target state chose to respond to the challenger state by use of force or war, the dependent variable “resist” is coded as 1; otherwise as 0.

4.3 Independent variables

The key independent variables are derived from the hypotheses. These include the leader’s manner of entry, whether or not the leadership transitioned that year, and the ally’s importance to the target. Interaction terms between leadership transition and ally’s importance will also be included. Indicators of leadership transition and the leader’s entry manner come from the Archigos dataset (Goemans et al., 2009). Leadership transition signifies whether or not the target state experienced a leadership change in the year of challenge, coded as 1 if yes, and 0 otherwise. Leader’s entry identifies the manner whereby a leader came to power, coded as 0 if through regular means, and 1 if power was attained through unconstitutional means or foreign imposition.

To operationalize the ally’s importance to the target, we use an expression commonly employed by other scholars (Fang et al., 2014). The ratio of the ally’s power to the sum of the ally and target’s capabilities is a suitable index to measure the ally’s importance.

$\frac{{\mathrm{cap}}_{\mathrm{a}}}{{\mathrm{cap}}_{\mathrm{a}}+{\mathrm{cap}}_{\mathrm{t}}}$

We use the Composite Index of National Capability (CINC) scores to measure the capabilities of the target state and its allies, derived from version 6.0 of the Correlates of War National Military Capabilities data (Singer, 1988).¹ This approach follows the methodology adopted by many scholars (Gibler, 2008; Kimball, 2006). Here, “cap_t” refers to the target’s CINC score, and “cap_a” to the sum of the allies’ CINC scores. The higher the ally’s national capability, the lower the proportion of the target state’s strength, thus signifying the ally’s high importance to the target. This indicator is recorded in the regression table as “Ally’s importance.”

4.4 Control variables

The first set of control variables relates to the impact of state power on the target state’s conflict decision-making. The first variable is that of whether or not the target state is a major power, referred to as “Target major power,” with data sourced from the State System Membership List.² As major powers typically possess superior capabilities and play significant roles in international conflicts they are both capable and willing to take resistance measures during crises. Their behavior and decisions hence differ from those of smaller states. Including this variable in the model helps us to understand the specific behavioral patterns of major powers in conflicts. The second variable is that of the ratio of strengths between the challenger and target states, referred to as “Relative power.” This variable represents the proportion of the target state’s CINC score relative to the total CINC scores of both the target and the challenger. A considerable strength gap between the target state and the challenger indicates that the target state will be unlikely to take resistance measures that could escalate the conflict.

The level of contiguity between the challenger state and the target state, referred to as “Contiguity,” is another control variable. Geographical proximity intensifies the vulnerability of the target state to military power projection and the risk of conflict overflow from neighboring states (Joyce and Braithwaite, 2013). The data source is the Correlates of War Direct Contiguity Data (Stinnett et al., 2002), whose contiguity codes range from 1 to 5. For this analysis, a value of 6 is assigned to cases where the states are separated by more than 400 miles of water. The full range of values is as follows: 1 = direct land contiguity; 2 = separated by 12 or fewer miles of water; 3 = separated by 24 or fewer (but more than 12) miles of water; 4 = separated by 150 or fewer (but more than 24) miles of water; 5 = separated by 400 or fewer (but more than 150) miles of water; and 6 = separated by more than 400 miles of water.

The subsequent two control variables relate to regime type. The first, “Target democracy,” measures the level of democracy in the target state, sourced from the V-Dem database (Coppedge et al., 2020). The second variable, “Joint democracy,” is a dichotomous indicator of whether or not both states in the dyad are democratic, coded as 1 if both states have Polity IV (Marshall et al., 2017) scores equal to or greater than 6, and otherwise coded as 0. Institutional constraints on leaders’ behaviors might play an influential role in the decision-making process (Horowitz et al., 2005). Previous research shows that foreign policy crises have different impacts on states, depending on their regime type (Debs and Goemans, 2010). Some scholars argue that democratic leaders, by virtue of regular elections, tend to be wary of war outcomes. Others, however, contend that nondemocratic leaders are more vulnerable in this respect, because war outcomes and losses heighten the risk of summary removal from office, and are hence pivotal to their political futures (De Mesquita and Siverson, 1995; Chiozza and Goemans, 2004; Croco, 2011). Debs and Goemans (2010) also find that the impact of crisis outcomes on the tenure of democratic leaders is smaller than that on autocratic leaders. Furthermore, the free press in democratic states provides more information than does the news media in autocratic regimes. In light of these scholars’ arguments, therefore, this study suggests that regime type may be a key determinant in evaluating a leader’s decision to resist.

The hostility level of the challenger is another factor that influences the target state’s decision to resist. We have, therefore, included the variable “Challenger hostility” to measure the highest recorded hostility level of a challenger in the conflict dyad-year. This data is sourced from the “dyhostlev1” indicator in the Truly Dyadic Dyad-Year version of the Militarized Interstate Confrontation (MICTDD) dataset (Gibler and Miller, 2024). The possible values are as follows: 1 = no militarized action; 2 = threat to use force; 3 = display of force; 4 = use of force; and 5 = war.

In addition, the alliance, the rivalry between the challenger and target states, as well as their conflict history, also influence the target state’s conflict decision-making. To measure the impact of these factors, We have included the dummy variable “Alliance” to signify whether or not there is a formal alliance between the challenger and target states in a given year, with data sourced from the ATOP (Alliance Treaty Obligations and Provisions) dataset (Leeds et al., 2002). Another dummy variable, “Strategic rivalry,” signifies whether or not there exists interventionist, ideological, or spatial rivalry between the two states in a specific year, with data sourced from Thompson and Dreyer’s strategic rivalry data (Thompson and Dreyer, 2012). Finally, to control for the impact of conflict history, we include the number of peace years between the challenger and target states, as well as “Peace year²” and “Peace year³,” which represent the square and cube of the peace years, respectively.

5 Results

Table 1 presents the results of a series of logit models examining the factors influencing target state resistance. Model 1 assesses the effects on the likelihood of resistance of target, target leader entry manner, target leadership transition, and the effect of interaction, without any control variables. Model 2 builds upon Model 1 by incorporating all control variables in testing the robustness of the interaction effect between “Ally’s importance” and “Target leadership transition.” Model 3 introduces target state fixed effects to control for unobserved heterogeneity across states. Finally, Model 4 includes year fixed effects to account for temporal variations in the data.

Table 1

Table 1. Ally, target leader, and target resistance.

The results of Model 1 indicate that ally’s importance does not have a significant impact on resistance. However, target leadership transition and target leader entry manner have a significant positive impact on resistance, suggesting that leadership transitions, as well as leaders who come to power through unconstitutional means, are likely to lead to the adoption of resistance measures. This may be due to the tendency of new leaders to demonstrate strength in order to consolidate their domestic position and gain public support. This is particularly true of those that, having attained power through unconstitutional means, feel compelled to adopt a hardline stance in order to prove their legitimacy and decisive leadership. Such leaders might believe that the taking of resistance measures showcases their determination and capability, so enhancing their prestige and influence in both the domestic and international arenas. New leaders might, moreover, proactively address any external threats in efforts to divert domestic attention from any questions arising about their power base and legitimacy, thereby stabilizing their regime. Leaders that attain power through transitions or irregular means, therefore, are more inclined to adopt resistance measures to achieve their political objectives. The interaction between ally’s importance and target leadership transition, however, is both negative and significant, indicating that when a leadership transition occurs, the increased importance of the ally significantly reduces the likelihood of the target state adopting resistance.

The results of Model 2 show that, after controlling for other factors, ally’s importance has a significant positive impact on resistance, signifying that it increases the likelihood of resistance. Target leadership transition and target leader entry manner continue to exhibit significant positive impact, further supporting the conclusion that leaders are more likely to resist if there is a leadership transition or if they have come to power through unconstitutional means. The interaction effect, however, remains negative and significant, indicating that, in the case of a leadership transition, an increase in ally’s importance inhibits resistance. This suggests that the target leadership’s transition has a moderating effect on the impact of ally’s importance. This emboldening effect emanates from the potential military, economic, and diplomatic support of allies that bolsters the target state’s capacity and willingness to resist. In the context of a leadership transition, however, higher importance of the ally may well inhibit the target state’s resistance. The considerable domestic and international pressure, and hence uncertainty, that new leaders initially face may compel them to allocate the bulk of their resources and effort towards maintaining ally relationships. The worry that the high importance of allies could result in either reduced or withdrawn ally support may cause leaders to desist from seeking the back-up necessary for resistance. In the context of leadership transition, therefore, the caution that ally’s importance creates in leaders accentuates the constraining effect of allies and thus lessens the likelihood of resistance.

Model 3 introduces target state fixed effects to control for unobserved heterogeneity across states. The results show that ally’s importance, target leadership transition, and target leader entry manner remain significant predictors of resistance. The interaction effect also remains negative and significant. The inclusion of target state fixed effects suggests that the effects of these variables is consistent across different target states. Model 4 includes year fixed effects to account for temporal variations in the data. The results show that ally’s importance, target leadership transition, and target leader entry manner continue to have a significant positive impact on resistance. The effect of interaction between ally’s importance and target leadership transition, meanwhile, remains negative and significant. The inclusion of year fixed effects suggests that temporal variations do not significantly alter the impact of primary variables, thus signifying consistent effects across different years.

Additionally, other control variables also show significant results across different models. The status of the target state as a major power positively influences the likelihood of resistance, suggesting that major powers, due to their superior capabilities and resources, are those inclined to effectively confront challenges. An increase in the target state’s relative power is also associated with a higher probability of resistance, signifying that stronger states in possession of substantial military and economic capacities are those that feel the most capable of opposing external threats. Territorial contiguity shows mixed significance, but generally, states with contiguous borders are those more likely to engage in conflicts, probably due to the greater opportunities and motivations between neighboring states for border disputes and skirmishes. Higher levels of democracy within the target state are linked to a lesser likelihood of resistance, suggesting that democratic states, aligning with the democratic peace theory, may opt for diplomatic and non-military solutions to international disputes. When both states in the dyad are democracies, however, the likelihood of resistance increases, thus supporting the idea that democratic dyads may experience higher tensions, or feel more compelled to assert their positions in conflicts due to mutual expectations of democratic norms and accountability (Maoz and Russett, 1993; Rosato, 2003; Gleditsch and Hegre, 1997). Higher levels of challenger hostility significantly increase the likelihood of resistance from the target state, so underscoring the importance of perceived threat levels in the target state’s decision-making process. The presence of alliances generally lessens the likelihood of resistance, suggesting that formal alliances might provide security assurances that reduce the need for direct military confrontation. Ongoing strategic rivalries significantly increase the probability of resistance, indicating that long-term adversarial relationships heighten the propensity for conflict. Finally, the number of peace years inversely correlates with the likelihood of resistance, showing that prolonged periods of peace reduce the immediate likelihood of conflict engagement.

When interpreting through logit models the substantive effect of interaction terms, it is often useful to present the results graphically. Figure 1 shows the predicted probabilities of resistance by target states as a function of ally’s importance, separated by whether or not there is a leadership transition. The left panel displays the scenario of no leadership transition; the right panel that which has a leadership transition. In the “No Leadership Transition” panel, the probability of resistance increases in tandem with the ally’s importance. This suggests that when there is no leadership transition, higher importance of the ally is associated with a greater likelihood of resistance. The shaded area represents the 95% confidence intervals, indicating that the relationship is statistically significant. Conversely, in the “Leadership Transition” panel, the probability of resistance decreases as the ally’s importance increases. This finding implies that for new leaders, higher importance of the ally reduces the likelihood of resistance. The broader confidence intervals in this panel suggest greater uncertainty in the estimated probabilities, but the negative trend remains clear.

Figure 1

Figure 1. Effect of ally’s importance on resistance by leadership transition. Shaded areas represent 95% confidence intervals.

In summary, the negative and significant interaction effect between ally’s importance and leader transition highlights that the presence of a new leader moderates the impact of ally’s importance on target’s resistance decisions. Specifically, while high importance of the ally generally encourage resistance, this effect is significantly mitigated when a new leader is in power. This could be due to new leaders facing greater domestic and external pressures and uncertainties, necessitating a more cautious approach to resistance despite high importance of the ally.

6 Robustness checks

To establish the robustness of these findings, we reassess the relationship between ally’s importance, leadership transition, leader entry, and the resistance of target states by employing alternative model specifications, and applying alternative measures for both the dependent variable—target resistance, and the independent variable—ally’s importance.

6.1 Different model specification

In the study of international conflicts, the challenger state’s decision to initiate conflict is a process influenced by selection effects, whose existence signifies that simple regression analysis may not accurately reflect the true relationships between variables. These selection biases may stem from various influences, including political, military, and alliance relations (De Mesquita and Siverson, 1995). To avoid selection bias, therefore, this research employs a two-stage Heckman model. At the selection stage, the dependent variable is dichotomous, indicating whether or not the challenger state attacks the target state in the directed dyad-year. At the outcome stage, the dependent variable is that of whether or not the target state resists (Heckman, 1979). Models in Figure 2 depict the empirical results of both the challenger’s decision whether or not to initiate a conflict, and the target’s choice of whether or not to resist. Thus, we obtain two sets of coefficients related to the challenger and the target.

As Figure 2 illustrates, the empirical results of the Heckman selection model provide robust support for the hypotheses. There are two sets of coefficients: the first section reports the coefficients associated with the variables in the selection equation; the second section reports the coefficients associated with the outcome equation. We additionally included in the selection equation such variables as the challenger’s leadership transition, the challenger’s method of entry into power, and the distance between the capitals of the two countries. The results show that both the target leadership transition and method of the target leader’s entry into power significantly influence conflict decisions. Specifically, the challenger is more likely to initiate a conflict when the target leadership transitions, or when the target leader attains power through irregular means. Furthermore, relative power negatively influences conflict initiation, signifying that the stronger the target state, the lesser the likelihood of the challenger initiating a conflict.

Figure 2

Figure 2. Heckman model results.

The significant positive coefficient of “Rho” in the outcome equation suggests that the Heckman selection model is appropriate here. The key independent variables—“Target leader entry” and “Target leadership transition”—yield results consistent with the models in Table 1, showing significant positive effects. The theoretical analysis above leads us to expect that if a leader comes to power through irregular means, or those imposed by another state, they are more inclined to fight the challenger. Additionally, target states experiencing leadership transitions are those more likely to resist the challenger, because new leaders, facing domestic audience costs, are more inclined to demonstrate a tough stance, while outgoing leaders might engage in “gambling for resurrection” tactic or in diversionary wars.

The results for ally’s importance and its interaction with leadership transition are consistent with those in Table 1. Specifically, higher importance of the ally significantly increases the likelihood of target state resistance, but this effect is considerably weakened in the case of leadership transitions. This aligns with the conclusions of our previous models 1, 2, 3, and 4: higher importance of the ally makes the target state more likely to resist the challenger due to the potential support of strong allies that emboldens the target state’s decision to militarily resist. In the case of a leadership transition in the target state, however, higher importance of the ally makes the target state less likely to resist the challenger, because the constraining effect of allies then becomes more pronounced. During a leadership transition the domestic foundation is unstable, so creating a greater need for external support from allies, and hence to maintain a good relationship with them. Therefore, such states must prioritize the stance of their allies—that of unwillingness to be implicated—which facilitates a more significant constraining effect on the target state.

6.2 Alternative measures of the dependent variable

To ensure the robustness of our research findings, we adopted an alternative measurement of the dependent variable. The original method coded the dependent variable “resist” as 1 if the target state chose to respond to the challenger state with the use of force or war; otherwise as 0. However, given that not all the challenger states’ actions involve explicit military threats or high-intensity military actions, assuming that the target state responds with the use of force or war in every situation is overly stringent. To address this issue, we propose the more nuanced measurement method of broadening the criteria whereby to define resistance. According to this new approach, if the challenger state provokes the target state, the latter’s response is considered resistant if it involves the threat of force or more severe actions. This adjustment aligns more closely with the complexity of international crises and conflicts. Interstate conflict behavior may involve a range of actions that do not necessarily culminate in actual use of force or full-scale war. By expanding the definition of resistance, we may capture a wider array of the target state’s conflict behaviors and responses, thereby providing a more comprehensive understanding of the dynamics involved. This broader approach allows us to reflect more accurately the varied and nuanced nature of state responses to international provocations. The results of the robustness check are presented in Table 2, whose models mirror the structure of those in Table 1 but utilize the revised measurement of resistance. Model 5 includes the primary independent variables and interaction terms. Model 6 adds control variables, Model 7 introduces target country fixed effects, and Model 8 incorporates year fixed effects.

Table 2

Table 2. Robustness check with alternative measurement of target resistance.

In Model 5, which includes the primary independent variables and interaction terms only, the results indicate that ally’s importance does not have a significant impact on resistance. Target leadership transition, however, has a significant positive effect on resistance, but leader entry manner has no significant effect. The interaction term between ally’s importance and leadership transition is negative and significant. These findings are consistent with those in Table 1, suggesting that in case of leadership transition, the increase in the importance of ally reduces the probability of the target state’s resistance. Model 6 introduces control variables. The results show that, controlling for other factors, ally’s importance has a significant positive impact on resistance. Target leadership transition remains a significant positive predictor, and the interaction term remains negative and significant. This reinforces the conclusion of Table 1: that leadership transitions and ally’s importance significantly influence resistance behavior. Model 7 includes target country fixed effects. Its results indicate that, even with the inclusion of fixed effects to control for unobserved heterogeneity between countries, ally’s importance and leadership transition continue to influence resistance significantly. The interaction term remains negative and significant, consistent with the findings in Table 1. Model 8 incorporates year fixed effects to account for temporal variations. The results show that ally’s importance and target leadership transition continue to have a significant positive impact on resistance, while the interaction term remains negative and significant. This consistency across different temporal contexts further supports the robustness of our findings.

Overall, the robustness check using an alternative measurement of the dependent variable strongly supports our initial conclusions. The significant effects of ally’s importance, target leadership transition, and their interaction remain consistent. Specifically, the significant positive effect on resistance of target leadership transition aligns with theoretical expectations, implying that newly appointed leaders, facing both internal and external pressures, are more likely to adopt a confrontational stance aimed at consolidating their power and legitimacy. The negative significant interaction effect between ally’s importance and leadership transition suggests that, in the context of a leadership transition, higher importance of the ally reduces the likelihood of resistance. This is plausible, because new leaders must ensure external support and hence maintain good relations with allies by prioritizing alliance stability and allowing the alliance’s constraining effect to play a more prominent role.

Interestingly, target leader entry is not significant in any of the models. Such insignificance may be due to the broader range of responses that the revised method of resistance measurement encompasses. The impact of the leader’s method of entry may not be significant in lower-intensity conflicts, bearing in mind that its role in high-intensity conflicts stems from the influence it wields on leader legitimacy and domestic support. A leader that comes to power through unconstitutional means, such as a coup or other irregular methods, may feel compelled to adopt a tough foreign policy geared to consolidating their domestic position and gaining them public support. When faced with high-intensity challenges, therefore, to demonstrate their resolve and leadership capabilities they are highly likely to engage in actual military resistance. In lower-intensity conflicts, however, where the challenge does not involve direct military threats or the use of force, the need to consolidate power militarily may be less pressing. Low-intensity conflicts, or provocative actions, are hence more likely to trigger diplomatic, economic, or other non-military responses. In such instances, leaders may elect to address the relevant issues through negotiation, sanctions, or other forms of non-military response, rather than push for military confrontation. In such contexts, therefore, the impact of the leader’s method of entry on resistance behavior is weakened, because non-military means preclude any demonstration of military strength to prove leadership ability or legitimacy.

6.3 Alternative measures of the independent variable

In the context of conflict decision-making, the stance and influence of a state’s most powerful ally often constitutes the most critical factors. States that face external threats generally rely on their most powerful allies for security assurance. Morrow (1991) further emphasizes that, in alliance relations, the military capability and political intentions of the most powerful ally play a decisive role in the decisions of its less powerful allies. In this robustness check, therefore, we have modified the measurement of ally’s importance. Originally defined as the ratio of the sum of the CINC (Composite Index of National Capability) scores of all the target state’s defensive allies to the combined CINC of such allies and the target state, the focus of this measurement has been revised to that of the target state’s most powerful defensive ally, defining ally’s importance as the CINC score of the most powerful ally divided by the sum of the CINC scores of this ally and of the target state. A high value signifies the target state’s stronger dependence on its most powerful ally, and thus that ally’s greater influence on the target state’s decision to resist. Taking into account that economic and trade dependencies may also influence the target state’s decision to resist, we additionally include the variable “Target trade dependency on challenger” which is the ratio of the target state’s total imports and exports with the challenger in a given year to the target state’s total trade in that year. The trade data are sourced from the Correlates of War trade data (Barbieri et al., 2009).

The results of the model based on this revised measurement are presented in Table 3, wherein we can observe, when using the alternative measurement method and adding trade dependency as the control variable, the significance and direction of the primary independent variables are consistent with the findings in Table 1, further validating our initial conclusions.

Table 3

Table 3. Robustness check with alternative measurement of ally’s importance.

In Model 9, which includes only the main independent variables and interaction terms, the results are consistent with those of Model 1. Ally’s importance has a significant positive effect on resistance, as does the target leadership transition. Although target leader entry is not significant in this model, the interaction term between ally’s importance and target leadership transition is negative and significant, showing that higher importance of the ally significantly reduces the likelihood of resistance when a leadership transition has occurred in the target country. Model 10 incorporates all control variables in addition to those in Model 9. The results show that, after controlling for other factors, the effects of ally’s importance, target leadership transition, and their interaction term remain unchanged. Target leader entry becomes significant in this model, possibly due to the model’s enhanced ability to capture the impact of leader entry methods through the inclusion of more control variables. This finding aligns with our theoretical expectation that leaders that attain power through irregular means are those more likely to adopt a hardline stance in response to external pressures, with the aim of consolidating their power and garnering domestic support. Models 11 and 12 further incorporate target country fixed effects and year fixed effects, respectively, based on Model 10. Similarly, Models 13 and 14 add the control variable “Target trade dependency on challenger.” The key findings of the main explanatory variables remain consistent across these models. Although “Target trade dependency on challenger” is not significant, our conclusions remain valid even when this variable is considered.

Therefore, even when the measurement of ally’s importance is changed to reflect the cost imposed by the strongest ally, our findings are still validated. This further confirms the crucial role that the interaction between leadership transition and ally’s importance plays in determining whether or not the target state will take resistance measures. The emboldening effect generated by the target’s strongest ally gives the target state the confidence needed to stand firm when facing conflict provocations. During a leadership transition in the target state, however, the costs imposed by the strongest ally significantly reduce the likelihood of resistance.

6.4 Different samples

The Militarized Interstate Disputes (MIDs) dataset (Maoz et al., 2019) from the Correlates of War (COW) project is the most commonly used database in research on conflict decision-making. This paper, however, opts for the Truly Dyadic Dyad-Year version of the Militarized Interstate Confrontation (MIC) dataset. By eliminating the false positives present in previous datasets (Gibler and Miller, 2024), the MIC dataset provides more accurate and detailed records of conflict events, which significantly enhance the precision of dyadic conflict analysis. Despite the greater accuracy and detail that the MIC dataset provides, however, the MID dataset remains widely used and recognized in the academic community. To ensure the broad applicability and reliability of our findings, therefore, we conducted analyses using the same variables and model specifications as those in Table 1, but with the MID dataset. Meanwhile, research suggests that target states may be more likely to resist in territorial disputes. While the Militarized Interstate Confrontation (MIC) dataset does not directly provide information on dispute types, the dyadic Militarized Interstate Disputes (MID) dataset from the Correlates of War (COW) project includes variables “REVTYPEA/B”, which indicate the type of revisionism pursued by a state.³ Based on MID, we have generated a new variable, “Challenger’s revision type”, in which the value of 1 indicates that the challenger’s revisionist aim is territorial, and 0 otherwise. We have included this as a control variable in our models to account for the potential influence of target states who may be more likely to resist in territorial disputes compared to other types of disputes.

The results of these analyses are presented in Table 4.

Table 4

Table 4. Robustness check using MID dataset.

Based on the results in Table 4 we further validate, using the MID dataset, the robustness of our initial findings. First, ally’s importance is significantly positive across all models. This result is consistent with those of Table 1, signifying that higher importance of the ally increases the likelihood of the target state’s adopting resistance measures. Second, target leadership transition is significantly positive in Model 15, but not in Models 16 to 18. This may be due to the control variables absorbing some of the variation. Although not significant in Models 16 to 18, the positive sign of the coefficient remains, suggesting that target leadership transition is still likely to increase the probability of resistance. The complexity of the models and data characteristics, however, has reduced the significance. Third, the interaction term “Ally’s importance × Target leadership transition” is significantly negative in Model 13, thus aligning with our theoretical expectations. This indicates that higher importance of the ally reduces the likelihood of resistance when a target leadership transition has taken place. This interaction term is not significant in Models 16 to 18, although the negative sign remains. Finally, target leader entry is not significant in Models 16 to 18, and in Model 17, the coefficient sign is reversed to negative. However, this variable remains significantly positive in Model 15.

Models 19 and 20 are built by incorporating the control variable “Challenger’s revision type” to Models 17 and 18. The results show that the effects of ally’s importance, target leadership transition, and the interaction between ally’s importance and leadership transition are consistent with those in Models 17 and 18, maintaining the same positive and negative coefficients. Additionally, the challenger’s revision type is significantly positive, aligning with the theoretical expectation that target states are more likely to resist in territorial disputes. This suggests that leaders are less willing to make concessions to the challenging state in such disputes.

In summary, although after switching to the MID dataset some variables are not significant in certain models, the significant positive impact of ally’s importance and the significant findings of target leadership transition and target leader entry, as well as the interaction term in Model 15, partially validate the robustness of our initial conclusions.

7 Conclusion

We observe, based on the analysis above, that the interplay between domestic political factors and ally influence is crucial to determining the target state’s response to a challenger’s provocation. Our study finds that leadership transitions, methods of leader entry, and ally’s importance have significant impacts on the likelihood of a target state’s resistance. Leaders that have recently come to power, or ascended through irregular means, are those more likely to resist challenge. Because public approval is crucial to their legitimacy and political survival, they set out to consolidate their rule and gain public support. In such scenarios, therefore, leaders tend to adopt assertive foreign policies regardless of potential ally restraint.

However, our findings also show that certain leadership factors play a moderating role in regard to ally influence. The support of a powerful ally may also embolden target states to resist challengers, thus providing leaders with the confidence necessary to undertake military actions, especially when they face no immediate domestic political threats. Conversely, politically vulnerable leaders that depend heavily on their allies are those most likely to accede to ally restraint, thus to secure continued support and avoid isolation. Our robustness checks, including alternative measures for both the independent and dependent variables, adjustments to model specifications, and sample changes, confirm the robustness of these findings. The interaction between domestic political dynamics and external ally influence remains a decisive factor in shaping the target state’s behavior in international conflicts.

In summary, although the focus of this study is on the roles of leadership transitions and ally’s importance, it also emphasizes the intricate balance between domestic political considerations and external influences. Leaders’ concerns about political survival, and the consequent necessity to maintain public support, are pivotal to their decision-making processes. Therefore, understanding the domestic political context is essential for comprehending how states respond to international provocations, and the effectiveness of ally restraint.

Data availability statement

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

Author contributions

HQ: Conceptualization, Data curation, Formal analysis, Methodology, Resources, Software, Supervision, Writing – original draft, Writing – review & editing, Investigation, Project administration, Validation, Visualization. QW: Investigation, Methodology, Software, Validation, Visualization, Writing – review & editing, Data curation, Resources. RL: Methodology, Resources, Software, Writing – review & editing.

Funding

The author(s) declare that no financial support was received for the research, authorship, and/or publication of this article.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpos.2024.1459665/full#supplementary-material

Footnotes

1. ^The Composite Index of National Capability (CINC) score comes from version 6.0 of the Correlates of War National Military Capabilities data. The CINC score is calculated as each state’s share of the world’s total for six composite indicators included in the data for a given year. Theoretically, the CINC score ranges between 0 and 1. A score of 1 would indicate that a state is 100% responsible for the following: (1) all military expenditures in the world, (2) being the only state with a military, (3) all iron and steel production, (4) all of the world’s primary energy consumption, and (5) being the only state in the world with both a population and an urban population.

2. ^Correlates of War Project. 2017. “State System Membership List, v2016.” Online, http://correlatesofwar.org.

3. ^The values of these variables are as follows: 0: Not applicable; 1: Territorial; 2: Policy; 3: Regime/Government; 4: Other; 9: Missing.

References

Bak, D. (2020). Autocratic political cycle and international conflict. Confl. Manag. Peace Sci. 37, 259–279. doi: 10.1177/0738894217741617

Crossref Full Text | Google Scholar

Barbieri, K., Keshk, O. M. G., and Pollins, B. M. (2009). Trading data: evaluating our assumptions and coding rules. Confl. Manag. Peace Sci. 26, 471–491. doi: 10.1177/0738894209343887

Crossref Full Text | Google Scholar

Beardsley, K., and Lo, N. (2014). Third-party conflict management and the willingness to make concessions. J. Confl. Resolut. 58, 363–392. doi: 10.1177/0022002712467932

Crossref Full Text | Google Scholar

Benson, B. V., Bentley, P. R., and Ray, J. L. (2013). Ally provocateur: why allies do not always behave. J. Peace Res. 50, 47–58. doi: 10.1177/0022343312454445

Crossref Full Text | Google Scholar

Bueno De Mesquita, B., Smith, A., Siverson, R. M., and Morrow, J. D. (2003). The logic of political survival. Cambridge, MA: The MIT Press.

Google Scholar

Chiozza, G., and Goemans, H. E. (2004). International conflict and the tenure of leaders: is war still ex post inefficient? Am. J. Polit. Sci. 48, 604–619. doi: 10.1111/j.0092-5853.2004.00090.x

Crossref Full Text | Google Scholar

Chiozza, G., and Goemans, H. E. (2011). Leaders and international conflict. 1st Edn. Cambridge: Cambridge University Press.

Google Scholar

Christensen, T. J. (2011). Worse than a monolith: Alliance politics and problems of coercive diplomacy in Asia. Princeton, NJ: Princeton University Press.

Google Scholar

Colaresi, M. (2004). Aftershocks: postwar leadership survival, rivalry, and regime dynamics. Int. Stud. Q. 48, 713–728. doi: 10.1111/j.0020-8833.2004.00322.x

Crossref Full Text | Google Scholar

Coppedge, M., Gerring, J., Knutsen, C. H., Lindberg, S. I., Teorell, J., Altman, D., et al. (2020). V-Dem Codebook :V10. doi: 10.2139/ssrn.3557877

Crossref Full Text | Google Scholar

Crawford, T. W. (2003). Pivotal deterrence: Third-party statecraft and the pursuit of peace. Ithaca, NY: Cornell University Press.

Google Scholar

Croco, S. E. (2011). The Decider’s dilemma: leader culpability, war outcomes, and domestic punishment. Am. Polit. Sci. Rev. 105, 457–477. doi: 10.1017/S0003055411000219

Crossref Full Text | Google Scholar

Croco, S. E., and Weeks, J. L. P. (2016). War outcomes and leader tenure. World Polit. 68, 577–607. doi: 10.1017/S0043887116000071

Crossref Full Text | Google Scholar

Danneman, N., and Ritter, E. H. (2014). Contagious rebellion and preemptive repression. J. Confl. Resolut. 58, 254–279. doi: 10.1177/0022002712468720

Crossref Full Text | Google Scholar

De Mesquita, B. B., and Siverson, R. M. (1995). War and the survival of political leaders: a comparative study of regime types and political accountability. Am. Polit. Sci. Rev. 89, 841–855. doi: 10.2307/2082512

Crossref Full Text | Google Scholar

Debs, A., and Goemans, H. E. (2010). Regime type, the fate of leaders, and war. Am. Polit. Sci. Rev. 104, 430–445. doi: 10.1017/S0003055410000195

Crossref Full Text | Google Scholar

Downs, G. W., and Rocke, D. M. (1994). Conflict, agency, and gambling for resurrection: the principal-agent problem goes to war. Am. J. Polit. Sci. 38:362. doi: 10.2307/2111408

Crossref Full Text | Google Scholar

Eyerman, J., and Hart, R. A. (1996). An empirical test of the audience cost proposition: democracy speaks louder than words. J. Confl. Resolut. 40, 597–616. doi: 10.1177/0022002796040004004

Crossref Full Text | Google Scholar

Fang, S., Johnson, J. C., and Leeds, B. A. (2014). To concede or to resist? The restraining effect of military alliances. Int. Organ. 68, 775–809. doi: 10.1017/S0020818314000137

PubMed Abstract | Crossref Full Text | Google Scholar

Fearon, J. D. (1994). Domestic political audiences and the escalation of international disputes. Am. Polit. Sci. Rev. 88, 577–592. doi: 10.2307/2944796

Crossref Full Text | Google Scholar

Fuhrmann, M., and Horowitz, M. C. (2015). When leaders matter: rebel experience and nuclear proliferation. J. Polit. 77, 72–87. doi: 10.1086/678308

Crossref Full Text | Google Scholar

Gaubatz, K. T. (1991). Election cycles and war. J. Confl. Resolut. 35, 212–244. doi: 10.1177/0022002791035002004

Crossref Full Text | Google Scholar

Gelpi, C. (1999). “Alliances as instruments of intra-allied control” in Imperfect unions. eds. H. Haftendorn, R. O. Keohane, and C. A. Wallander (Oxford: Oxford University Press), 107–139.

Google Scholar

Gelpi, C., and Grieco, J. M. (2001). Attracting trouble: democracy, leadership tenure, and the targeting of militarized challenges, 1918-1992. J. Confl. Resolut. 45, 794–817. doi: 10.1177/0022002701045006005

Crossref Full Text | Google Scholar

Gibler, D. M. (2008). The costs of reneging: reputation and Alliance formation. J. Confl. Resolut. 52, 426–454. doi: 10.1177/0022002707310003

Crossref Full Text | Google Scholar

Gibler, D. M., and Miller, S. V. (2024). The militarized interstate confrontation dataset, 1816-2014. J. Confl. Resolut. 68, 562–586. doi: 10.1177/00220027221104704

Crossref Full Text | Google Scholar

Gleditsch, N. P., and Hegre, H. (1997). Peace and democracy: three levels of analysis. J. Confl. Resolut. 41, 283–310. doi: 10.1177/0022002797041002004

PubMed Abstract | Crossref Full Text | Google Scholar

Goemans, H. E. (2008). Which way out? The manner and consequences of losing office. J. Confl. Resolut. 52, 771–794. doi: 10.1177/0022002708323316

Crossref Full Text | Google Scholar

Goemans, H. E., Gleditsch, K. S., and Chiozza, G. (2009). Introducing Archigos: a dataset of political leaders. J. Peace Res. 46, 269–283. doi: 10.1177/0022343308100719

PubMed Abstract | Crossref Full Text | Google Scholar

Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica 47:153. doi: 10.2307/1912352

PubMed Abstract | Crossref Full Text | Google Scholar

Horowitz, M., McDermott, R., and Stam, A. C. (2005). Leader age, regime type, and violent international relations. J. Confl. Resolut. 49, 661–685. doi: 10.1177/0022002705279469

Crossref Full Text | Google Scholar

Horowitz, M. C., and Stam, A. C. (2014). How prior military experience influences the future militarized behavior of leaders. Int. Organ. 68, 527–559. doi: 10.1017/S0020818314000046

Crossref Full Text | Google Scholar

Horowitz, M. C., Stam, A. C., and Ellis, C. M. (2015). Why leaders fight. Cambridge: Cambridge University Press.

Google Scholar

Johnson, J. C. (2016). Alliance treaty obligations and war intervention. Confl. Manag. Peace Sci. 33, 451–468. doi: 10.1177/0738894215577557

Crossref Full Text | Google Scholar

Johnson, J. C., and Leeds, B. A. (2011). Defense pacts: a prescription for peace? Defense pacts and peace. Foreign Policy Anal. 7, 45–65. doi: 10.1111/j.1743-8594.2010.00122.x

Crossref Full Text | Google Scholar

Johnson, J. C., Leeds, B. A., and Wu, A. (2015). Capability, credibility, and extended general deterrence. Int. Interact. 41, 309–336. doi: 10.1080/03050629.2015.982115

Crossref Full Text | Google Scholar

Joyce, K. A., and Braithwaite, A. (2013). Geographic proximity and third-party joiners in militarized interstate disputes. J. Peace Res. 50, 595–608. doi: 10.1177/0022343313489587

Crossref Full Text | Google Scholar

Kimball, A. L. (2006). Alliance formation and conflict initiation: the missing link. J. Peace Res. 43, 371–389. doi: 10.1177/0022343306064816

Crossref Full Text | Google Scholar

Kisangani, E. F., and Pickering, J. (2011). Democratic accountability and diversionary force: regime types and the use of benevolent and hostile military force. J. Confl. Resolut. 55, 1021–1046. doi: 10.1177/0022002711414375

Crossref Full Text | Google Scholar

Krebs, R. R. (1999). Perverse institutionalism: NATO and the Greco-Turkish conflict. Int. Organ. 53, 343–377. doi: 10.1162/002081899550904

Crossref Full Text | Google Scholar

Lai, B., and Slater, D. (2006). Institutions of the offensive: domestic sources of dispute initiation in authoritarian regimes, 1950–1992. Am. J. Polit. Sci. 50, 113–126. doi: 10.1111/j.1540-5907.2006.00173.x

Crossref Full Text | Google Scholar

Leeds, B. A. (2003). Do alliances deter aggression? The influence of military alliances on the initiation of militarized interstate disputes. Am. J. Polit. Sci. 47, 427–439. doi: 10.1111/1540-5907.00031

Crossref Full Text | Google Scholar

Leeds, B. A., and Davis, D. R. (1997). Domestic political vulnerability and international disputes. J. Confl. Resolut. 41, 814–834. doi: 10.1177/0022002797041006005

Crossref Full Text | Google Scholar

Leeds, B., Ritter, J., Mitchell, S., and Long, A. (2002). Alliance treaty obligations and provisions, 1815-1944. Int. Interact. 28, 237–260. doi: 10.1080/03050620213653

PubMed Abstract | Crossref Full Text | Google Scholar

Leeds, B. A., and Savun, B. (2007). Terminating alliances: why do states abrogate agreements? J. Polit. 69, 1118–1132. doi: 10.1111/j.1468-2508.2007.00612.x

Crossref Full Text | Google Scholar

Levy, J. S. (1989). “The diversionary theory of war: a critique” in Handbook of war studies. ed. M. I. Midlarsky (Ann Arbor, MI: University of Michigan Press), 259–288.

Google Scholar

Licht, A. A., and Allen, S. H. (2018). Repressing for reputation: leadership transitions, uncertainty, and the repression of domestic populations. J. Peace Res. 55, 582–595. doi: 10.1177/0022343317750215

Crossref Full Text | Google Scholar

Mansfield, E. D., and Snyder, J. (2005). Electing to fight: Why emerging democracies go to war. Cambridge, MA: MIT Press.

Google Scholar

Maoz, Z., Johnson, P. L., Kaplan, J., Ogunkoya, F., and Shreve, A. P. (2019). The dyadic militarized interstate disputes (MIDs) dataset version 3.0: logic, characteristics, and comparisons to alternative datasets. J. Confl. Resolut. 63, 811–835. doi: 10.1177/0022002718784158

Crossref Full Text | Google Scholar

Maoz, Z., and Russett, B. (1993). Normative and structural causes of democratic peace, 1946–1986. Am. Polit. Sci. Rev. 87, 624–638. doi: 10.2307/2938740

Crossref Full Text | Google Scholar

Marshall, M. G., Gurr, T. R., and Jaggers, K. (2017). Polity IV project: Political regime characteristics and transitions, 1800–2017. Center for Systemic Peace. Available at: https://www.systemicpeace.org/inscr/p4manualv2017.pdf (Accessed June 25, 2024).

Google Scholar

Mattes, M. (2012). Reputation, symmetry, and alliance design. Int. Organ. 66, 679–707. doi: 10.1017/S002081831200029X

Crossref Full Text | Google Scholar

Morgan, T. C., and Palmer, G. (2003). To protect and to serve: alliances and foreign policy portfolios. J. Confl. Resolut. 47, 180–203. doi: 10.1177/0022002702251028

Crossref Full Text | Google Scholar

Morrow, J. D. (1991). Alliances and asymmetry: an alternative to the capability aggregation model of alliances. Am. J. Polit. Sci. 35:904. doi: 10.2307/2111499

Crossref Full Text | Google Scholar

Morrow, J. D. (1994). Alliances, credibility, and peacetime costs. J. Confl. Resolut. 38, 270–297. doi: 10.1177/0022002794038002005

Crossref Full Text | Google Scholar

Morrow, J. D. (2000). Alliances: why write them down? Annu. Rev. Polit. Sci. 3, 63–83. doi: 10.1146/annurev.polisci.3.1.63

Crossref Full Text | Google Scholar

Mueller, J. E. (1970). Presidential popularity from Truman to Johnson. Am. Polit. Sci. Rev. 64, 18–34. doi: 10.2307/1955610

Crossref Full Text | Google Scholar

Palmer, G., London, T., and Regan, P. (2004). What’s stopping you? The sources of political constraints on international conflict behavior in parliamentary democracies. Int. Interact. 30, 1–24. doi: 10.1080/725289044

Crossref Full Text | Google Scholar

Partell, P. J., and Palmer, G. (1999). Audience costs and interstate crises: an empirical assessment of Fearon’s model of dispute outcomes. Int. Stud. Q. 43, 389–405. doi: 10.1111/0020-8833.00125

Crossref Full Text | Google Scholar

Rosato, S. (2003). The flawed logic of democratic peace theory. Am. Polit. Sci. Rev. 97, 585–602. doi: 10.1017/S0003055403000893

Crossref Full Text | Google Scholar

Schultz, K. A. (1998). Domestic opposition and signaling in international crises. Am. Polit. Sci. Rev. 92, 829–844. doi: 10.2307/2586306

Crossref Full Text | Google Scholar

Schultz, K. A. (1999). Do democratic institutions constrain or inform? Contrasting two institutional perspectives on democracy and war. Int. Organ. 53, 233–266. doi: 10.1162/002081899550878

PubMed Abstract | Crossref Full Text | Google Scholar

Schultz, K. A. (2001). Looking for audience costs. J. Confl. Resolut. 45, 32–60. doi: 10.1177/0022002701045001002

PubMed Abstract | Crossref Full Text | Google Scholar

Schultz, K. A. (2005). The politics of risking peace: do hawks or doves deliver the olive branch? Int. Organ. 59, 1–38. doi: 10.1017/S0020818305050071

Crossref Full Text | Google Scholar

Singer, J. D. (1988). Reconstructing the correlates of war dataset on material capabilities of states, 1816–1985. Int. Interact. 14, 115–132. doi: 10.1080/03050628808434695

Crossref Full Text | Google Scholar

Smith, A. (1995). Alliance formation and war. Int. Stud. Q. 39:405. doi: 10.2307/2600800

PubMed Abstract | Crossref Full Text | Google Scholar

Snyder, G. H. (1984). The security dilemma in Alliance politics. World Polit. 36, 461–495. doi: 10.2307/2010183

PubMed Abstract | Crossref Full Text | Google Scholar

Stinnett, D. M., Tir, J., Diehl, P. F., Schafer, P., and Gochman, C. (2002). The correlates of war (cow) project direct contiguity data, version 3.0. Confl. Manag. Peace Sci. 19, 59–67. doi: 10.1177/073889420201900203

Crossref Full Text | Google Scholar

Thompson, W. R., and Dreyer, D. (2012). Handbook of international rivalries. Washington, DC: CQ Press.

Google Scholar

Zagare, F. C., and Marc Kilgour, D. (2003). Alignment patterns, crisis bargaining, and extended deterrence: a game-theoretic analysis. Int. Stud. Q. 47, 587–615. doi: 10.1046/j.0020-8833.2003.00280.x

Crossref Full Text | Google Scholar