Me,myself,and allies: Understanding the external sources of power

Abstract

As far back as Thucydides, scholars have hypothesized that power affects the onset of conflict. Despite its importance, power remains a difficult concept to measure, and scholars have primarily relied on material measures that quantify the internal resources available to a state. This concentration on internal sources of power, however, excludes an important power resource available to a state: its external relations. It is reasonable to expect that when a state estimates the power of a potential opponent it looks not only at the internal resources but also at the power of states that would likely join the conflict. In this article, we develop a new measure of external power that explicitly accounts for the external sources of state power. Unlike previous studies that aggregate a state’s expected alliance contributions, our measure is based on the expected contribution of all states, allies and non-allies alike. We conduct a preliminary test of this new measure on dispute onset, and our results provide support for power preponderance over balance of power theories. External power parity contributes to dispute onset rather than deterrence. In addition, we show that examining the combined, rather than individual, effects of external and internal power produces some intriguing results, suggesting that one state’s internal power preponderance can be offset by another state’s preponderance of external power. These results altogether suggest that further studies examining the role of external power can produce fruitful results.

Keywords

alliance conflict joining behavior power

Introduction

The distribution of power affects the outcome of a conflict, so states reasonably examine the balance of capabilities when making decisions on the use of force. However, states do not interact in a vacuum. The international system contains a number of additional states that may intervene on either side of a conflict. As such, states need to take into account the power of potential joiners as well as the odds that they would intervene. The contribution of additional states can dramatically affect the outcome of a conflict (such as the United States entering World War I and II, China entering the Korean War, and so on) and thus a state’s decision must necessarily account for not only the balance of forces between the original pair but also the potential contributions of joiners. This is not much different than Waltz’s (1979) distinction between external and internal balancing, which is essentially a different way of saying that states have both internal and external sources of power.

Despite the importance of external sources of power, most quantitative research almost exclusively uses internal power to determine the distribution of capabilities in a dyad. While it is theoretically important to incorporate both internal and external sources of power, it is also empirically problematic to exclude such a vital part of power. As such, we develop a new measure of external power for each state for the period 1870 to 2001. In addition, as a check on the measure’s validity, we highlight a number of ways in which external power affects the onset of conflict.

Power and conflict

Power has occupied a central role in explaining the onset of conflict as far back as Thucydides (I: 23) who believed that the real cause of the Peloponnesian War was ‘the growth of the power of Athens, and the alarm which this inspired in Lacedaemon’. The centrality of power makes sense in that war centers on the application of power. While power might derive from a number of non-material sources, it is this coercive capacity – that is, the power to hurt – which plays such a critical role in the bargaining process of international politics. This threat of violence is directly related to the power to hurt and often the material aspects of power. States can, and do, extend their power by making commitments to defend other states. The credibility of these commitments is critical in both ‘estimating enemy intentions but [also] influencing them’ (Schelling, 1966: 35; italics in original). These formal, informal, and implied commitments act as the basis of external power.

Thucydides equated the growth of Athenian power with the outbreak of the Peloponnesian War, but this growth was not exclusively internal. One could argue that the development of the Delian League and its transformation into the Athenian Empire was the critical growth in power that inspired fear in the Spartans. In fact, the league was quite useful to the Athenians because it ‘could act swiftly and decisively [and] Athens, moreover, had and used the power to see that league decisions were carried out’ (Kagan, 1969: 43). It is important to note that a concentration only on the internal sources of power would have misrepresented the true power of Athens. In fact, a case could be made that the growth of external sources of power generated more fear in the Spartans than the growth of Athenian internal power.

While the Athenian case is supportive, it also glosses over some important caveats about external sources of power. Perhaps the biggest caveat is that allies are never completely reliable. By the outbreak of the Peloponnesian War, Athens was clearly dominating the Delian League, so defection was unlikely. In most cases, however, the risk that an ally or potential joiner would not intervene is quite acute. Waltz (1979: 168) makes this point when he notes that ‘internal balancing is more reliable and precise than external balancing’. Of course, states cannot always balance internally and so the international system is replete with alliances and potential joiners.

External sources of power have primarily been linked to alliances and alliance politics. Morgenthau (1993 [1948]: 197) was quite specific in delineating the various paths states can use to increase their power:

They can increase their own power, they can add to their own power the power of other nations, or they can withhold the power of other nations from the adversary. When they make the first choice, they embark upon an armaments race. When they choose the second and third alternatives, they pursue a policy of alliances.

Along the same line, Wayman (1984) argued that the balance of power in the international system was a function of both power distribution and alliance clustering. Yet despite the importance of alliances and alliance politics, joining behavior is not limited to allies. In other words, states can enter a conflict without having a formal alliance with either side.

While it is unlikely that any individual case was only affected by external power and potential for external actors to join, there are certainly a number of examples where the potential of states joining a conflict affected decisionmaking. For instance, when Israel contemplated conflict with its Arab neighbors, it wanted to ensure quick victory as Israelis ‘feared that the superpowers would intervene when the Israeli Defense Forces (IDF) were in the process of inflicting a decisive defeat on the Arabs’ (Mearsheimer, 1983: 135). Schelling (1966: 36–37) also highlights this effect by noting that the West did not intervene in Hungary in 1956 ‘because the Soviet Union was strong enough, and likely enough to react militarily, to make Hungary seem not worth the risk’. In addition, during the 1902–03 Venezuelan debt crisis, the United States intervened on behalf of non-allied Venezuela, which in part led to Germany and Britain backing down (Mitchell, 1999). There are certainly additional examples but it seems clear that external interventions matter and states take these into account when making policy decisions.

Measuring external power

External sources of power were measured to a certain extent in the past but we want to move beyond simply aggregating alliance power. As noted above, allies are not guaranteed to keep their commitments and a simple concentration on allies ignores the possibility of non-allies entering a conflict.¹ To develop a more complete measure of external power, we attempt to minimize a priori assumptions about the behavior of potential joiners. We first develop a model to predict dispute joining behavior. After estimating each covariate’s effect on the probability of joining, this information is used to calculate each state’s external power.

Estimating joining behavior

We begin by estimating a model of joining behavior. We first identified all disputes in a given year and created a monadic dispute dataset with separate observations for each of the original disputants. For example, WWII is coded as two observations for each of the original disputants (Poland and Germany). We then add the potential joiners by creating a pair-wise combination of all remaining (non-disputant) states. We consider all states to be potential joiners, and not just formal allies. The unit of analysis is the disputant-potential joiner dyad-year from 1870 to 2001. The dependent variable is dichotomous, coded 1 if the potential joiner aids an original participant in a dispute in a given year and 0 otherwise. We identify joiners based on the MID coding.

As for independent variables, we capture a potential joiner’s ties to the disputant to determine its willingness to intervene.

Alliance ties

Several studies found alliance ties to be associated with a state’s decision to join one side or the other in a dispute. Our model of joining behavior therefore includes a dichotomous measure of Alliance ties, coded 1 if the potential joiner shares an offensive alliance, defensive alliance, or consultation pact with State A and 0 otherwise. The data are from the Alliance Treaty Obligations and Provisions dataset (Leeds, Long & Mitchell, 2000).

Trade dependence

Research on deterrence (Danilovic, 2002; Huth & Russett, 1984) found that a potential defender is more likely to intervene when it has significant interests at stake. Concerning economic ties alone, Aydin (2008) shows that potential joiners are more likely to intervene on the side of states with which they share higher levels of trade. To account for the role of a potential joiner’s economic interests, we include a measure of Trade dependence, calculated as the potential joiner’s total trade (exports + imports) with the disputant (State A) divided by the potential joiner’s total trade. Trade data are derived from the Correlates of War Project Trade Dataset (Barbieri, Keshk & Pollins, 2008).

Joint democracy

A number of studies show that democratic states are more likely to ‘flock’ together (Siverson & Emmons, 1991) in alliances. Democracies have also been found to be more reliable allies (Gaubatz, 1996; Leeds, Mattes & Vogel, 2009). Corbetta (2010) shows that democracies intervene more often on the side of other democracies than non-democracies, although Gartzke & Gleditsch (2004) show that democracies are less reliable allies. To test the effect of joint democracy on joining behavior, we include a measure of Joint democracy. It is coded 1 if State A and the potential joiner are democratic (a Polity2 score of 6 or above); otherwise it is coded 0. Regime type data are from the Polity IV Dataset (Marshall & Jaggers, 2002).

Capabilities-potential joiner

It is relatively straightforward to expect stronger states to be more likely to intervene in disputes. Our model therefore includes the potential joiner’s capabilities, measured using the Correlates of War Composite Indicator of National Capabilities (CINC score).

Alliance portfolio similarity

Altfeld & Bueno de Mesquita (1979) showed that alliance portfolio similarity, as an indicator of shared preferences, is a strong predictor of a state’s willingness to join a dispute. More recently, Corbetta (2010) found this result to hold across a larger temporal and spatial domain. Following these findings, our joining model includes a measure of Alliance portfolio similarity using Signorino & Ritter’s (1999) weighted S-score.²

Contiguity

Consistent with other studies that found that contiguity influences joining behavior (Most & Starr, 1989; Siverson & Starr, 1991), we include a measure of Contiguity. It is coded 1 if State A and the potential joiner share a common land border and 0 otherwise.

Alliance portfolio similarity x Contiguity

In addition to joining disputes, contiguous states are also more likely to fight one another (Diehl, 1985; Bremer, 1992; Vasquez, 1996). In an attempt to more accurately identify when contiguous states would be willing to join a dispute, we created an interaction term between contiguity and the alliance portfolio similarity of State A and the joiner.

The results from the probit model of joining behavior are reported in the Appendix Table A1. The results are consistent with our expectations. States are more likely to intervene on the side of allies, important trading partners, and contiguous states with which they have similar alliance portfolios. Interestingly, however, the negative and statistically significant coefficient for Alliance portfolio similarity indicates that intervention is less likely for non-contiguous states as the similarity of their alliance portfolios increases.³ Intervention is also more likely when the disputant and potential joiners are both democratic and as the power of the potential joiner increases. We also provide the results from a model of joining behavior that includes only the alliance ties between the disputant and potential joiner. The Likelihood Ratio test comparing these two models shows that our full model improves upon the alliance-only model (χ² = 299.19, p = .000), lending validity to our approach to measuring external power.

Calculating external power

Using the estimates from the joining model in Appendix Table A1, each state’s external power in each year was calculated as follows:

E x t e r n a l P o w e r A_{t} = \sum_{i, t = 1}^{n} {P r (J o i n | X)}_{i, t} * C a p_J o i n e r_{i, t}

where Pr(Join|X)_i,t is the unique probability of joining in each ‘State A–Potential joiner’ dyad-year, given the values of the independent variables, and Cap_Joiner_i,t denotes the military capabilities of each potential joiner in that year. In other words, we multiply each potential joiner’s (i’s) probably of joining with its internal military capabilities. This provides us with each state’s potential power contribution to a state in a particular year (t). The state’s External power in year t is then calculated by summing the individual contributions of each potential joiner for that year.⁴

This procedure gives a unique value of a state’s external capabilities per year based on the probability that it will receive assistance in the event of a dispute from an outside state. Our measure of external power ranges from a minimum of essentially zero (2.68 x 10^–08) to a maximum of .065, with a mean of .005. To provide a more detailed illustration of our measure, Table I shows the top 100 states in the year 2000 in terms of their level of external power. For comparison purposes, we also provide each state’s level of internal power (CINC score) and corresponding ranking.

There is a positive but weak correlation (r = .135) between external and internal power, indicating that internally stronger states tend to have higher levels of external power. Interestingly, Canada and Mexico rank #1 and #2, respectively, likely due to their geographic proximity to the United States and, in the former case, its membership in NATO. In the top 20 we find NATO member states surrounded by democratic countries with similar alliance portfolios (France, Germany, Belgium, Netherlands, Italy) or those that are allied to the United States (Japan, South Korea). It is also interesting to note some countries with a greater gap between their internal and external power. China and India, though ranked #2 and #3 in internal power, have substantially lower levels of external power (ranked #62 and #48). Iran drops even further, from #17 in internal power to an external power ranking of #127. Israel’s external power is higher than its internal power which is likely related to its ties to powerful Western states.

Preliminary analysis of external power and MID onset

While the external power data are interesting in their own right, it is also important to see if external power actually has an effect. As such, we provide a preliminary test of the measure in a model of conflict with Dispute onset as a dichotomous dependent variable. It takes a value of 1 if either state in the dyad threatened, displayed, or used force in a given dyad-year, otherwise it

Table I.

State rankings according to their external power, top 100 countries in year 2000

Rank	Country	External power	Internal power	Internal power rank	Rank	Country	External power	Internal power	Internal power rank
1	Canada	.014784	.012364	18	51	Ireland	.003971	.000721	104
2	Mexico	.010139	.015116	15	52	Romania	.003956	.004797	38
3	United States	.009658	.149842	1	53	Sweden	.003937	.003713	47
4	Germany	.009043	.027967	6	54	Haiti	.003901	.000514	117
5	France	.009032	.020791	10	55	Moldova	.003899	.000575	112
6	Russia	.007979	.052663	5	56	Kazakhstan	.003818	.003382	51
7	Italy	.007916	.017998	11	57	Mongolia	.003349	.000263	139
8	South Korea	.007915	.024178	8	58	Kyrgyzstan	.003266	.000405	126
9	Czech Rep.	.007737	.003119	53	59	Ukraine	.003231	.016002	13
10	Spain	.007702	.010669	21	60	Lithuania	.003213	.000579	111
11	Japan	.007523	.052950	4	61	Latvia	.003154	.000379	130
12	Greece	.007452	.003634	48	62	China	.003115	.131174	2
13	Turkey	.007300	.015602	14	63	Tajikistan	.003077	.000403	127
14	Poland	.007270	.008511	24	64	Senegal	.003044	.000638	106
15	Belgium	.007148	.004743	39	65	Belarus	.002997	.002640	57
16	Netherlands	.006931	.005373	35	66	Taiwan	.002920	.011955	19
17	United Kingdom	.006653	.023782	9	67	Croatia	.002911	.000987	85
18	Argentina	.006650	.006280	31	68	Azerbaijan	.002896	.001405	81
19	Venezuela	.006598	.004689	40	69	Thailand	.002881	.007149	27
20	Denmark	.006563	.001415	80	70	Switzerland	.002870	.001795	69
21	Colombia	.006525	.005751	34	71	Georgia	.002841	.000659	105
22	Bolivia	.006515	.000981	87	72	South Africa	.002815	.006985	28
23	Uruguay	.006494	.000551	114	73	Israel	.002811	.004106	44
24	Paraguay	.006492	.000450	122	74	Slovenia	.002758	.000431	124
25	Brazil	.006391	.026478	7	75	New Zealand	.002752	.000978	88
26	Australia	.006370	.007453	26	76	Serbia	.002748	.002245	59
27	North Korea	.006290	.011351	20	77	Cyprus	.002743	.000245	142
28	Portugal	.006282	.001783	70	78	Papua New Guinea	.002740	.000212	145
29	Hungary	.006280	.001966	62	79	Madagascar	.002730	.000842	96
30	Philippines	.006263	.005782	33	80	Mauritius	.002729	.000068	158
31	Guatemala	.006251	.000783	99	81	Botswana	.002690	.000223	144
32	Chile	.006192	.003409	50	82	Armenia	.002441	.000768	101
33	Panama	.006108	.000289	137	83	Syria	.002397	.004116	43
34	Dominican Rep.	.006090	.000949	91	84	Uzbekistan	.002385	.003121	54
35	Costa Rica	.006080	.000318	134	85	Turkmenistan	.002360	.000597	109
36	Honduras	.006079	.000500	118	86	Vietnam	.002345	.007789	25
37	El Salvador	.006067	.000638	107	87	Libya	.002209	.001908	65
38	Trinidad & Tobago	.006063	.000400	128	88	Jordan	.002205	.001457	78
39	Jamaica	.006062	.000198	147	89	Myanmar	.002160	.005045	36
40	Nicaragua	.006040	.000430	125	90	Saudi Arabia	.002155	.009989	22
41	Pakistan	.006004	.014011	16	91	Iraq	.002147	.006457	30
42	Norway	.006001	.001699	72	92	Kuwait	.002145	.001429	79
43	Liberia	.005111	.000281	138	93	United Arab Emirates	.002130	.002087	61
44	Finland	.004838	.002186	60	94	Nepal	.002119	.001140	83
45	Bulgaria	.004424	.001922	63	95	Qatar	.002097	.000943	92
46	Slovakia	.004297	.001657	74	96	Yemen	.002096	.001478	76
47	Guyana	.004184	.000064	160	97	Bahrain	.002093	.000354	132
48	India	.004054	.068318	3	98	Congo	.002073	.000376	131
49	Austria	.004035	.002753	56	99	Algeria	.002068	.004325	41
50	Ecuador	.003988	.001659	73	100	Tunisia	.002065	.000931	93

Table II.

Dispute onset as a function of relative internal and external power, politically relevant dyads 1870–2000

	Model 1	Model 2
Relative internal power	.811** (.058)	–
Relative external power	.309** (.057)	–
Dichotomous categories
Parity conditions
Internal parity-external parity	–	Reference category
Asymmetric internal-external preponderance	–	–.143 (.085)
Preponderance conditions
Internal parity-external preponderance	–	.109 (.102)
Internal preponderance-external parity	–	–.224** (.081)
Symmetric (one-sided) internal-external preponderance	–	–.511** (.079)
Joint democracy	–.015** (.003)	–.016** (.003)
Trade interdependence	–2.729 (2.994)	3.658 (2.229)
Peaceyears	–.085** (.005)	–.088** (.005)
Peaceyears²	.001** (.000)	.002** (.000)
Peaceyears³	–.000** (.000)	–.000** (.000)
Constant	–1.702** (.040)	–.993** (.071)
N	38463	38463
Log-likelihood	–3508.02	–3576.77
χ²	1064.42**	824.75**

Coefficients reported. Robust standard errors clustered on dyads in parentheses. Computed using STATA 12. *p < 0.05, **p < 0.01.

is coded 0. We drop the years from the analysis if there is an ongoing dispute and analyze all politically relevant dyads from 1870 to 2000.⁵ The dependent and independent variables are from the MID dataset (Ghosn, Palmer & Bremer, 2004), downloaded using EUGene (Bennett & Stam, 2000).

Independent variables: Dispute onset model

Relative external power

We use a standard method to transform each state’s external power into a relative (dyadic) measure by dividing the external capabilities of the externally weaker state by that of the externally stronger state. The variable ranges from 0 (external power preponderance for one state) to 1 (external power parity).⁶

Relative internal power

As for internal power, we use the Correlates of War Composite Indicator of National Capabilities (Singer, 1987). Relative Internal Power is measured by dividing the power of the weaker member in the dyad by that of the stronger state. The variable ranges from zero (power preponderance) to one (power parity).

Joint democracy

Given the strength of the findings in support of the democratic peace, we include a variable, Democracy_Low , to capture joint democracy in the dyad. It is measured as the lower democracy score of the two states in the dyad. The democracy scores are from the Polity IV dataset (Marshall & Jaggers, 2002).

Trade dependence

To control for economic interdependence, we use a common measure of trade dependence. Using the data from Oneal & Russett (1997), our models include the lower dependence score of the two states in the dyad. Higher values indicate greater dyadic trade dependence.

Temporal controls

To control for temporal dependence in models with a binary dependent variable, we use the procedure first suggested by Carter & Signorino (2010). Our models include three temporal controls: Peaceyears, Peaceyears ² , and Peaceyears ³ .

Results

Table II reports the results from two multivariate models of dispute onset. In column 1, the positive and statistically significant coefficient for the variable Relative external power

Figure 1.

Predicted probability of a dispute as a function of internal and external power

indicates that dispute onset is more likely under power parity than it is under power preponderance. Our results are therefore more supportive of the power preponderance school.

In Figure 1 we graphed the predicted probability of a dispute as the relative external and internal power variables change from power preponderance to power parity. Consistent with the results in Table II, the figure shows an increase in the probability of dispute onset as the dyad approaches external power parity. Moving from the minimum to maximum values, we see that disputes are more than twice as likely under conditions of external power parity (p ≈ .015) than under external power preponderance (p ≈ .0075). A similar pattern is evident for internal power. However, a comparison of the left- and right-side panels also shows that the magnitude of the effect is larger for internal compared to external power. While this validates the importance of internal power, our findings about the influence of external power confirm our argument that outside contributions to a state’s power do, in fact, matter.

It could also be the case that the influence of external power parity or disparity on dispute onset is conditioned by the balance of internal power. For example, one side’s external power preponderance may be mitigated by the other’s preponderance of internal power (or vice versa). Or the effects of one side’s external power preponderance could be accentuated if that same state is also preponderant in internal power. However, testing this is not straightforward, since a two-way interaction between the two continuous variables for relative internal and external power can obscure some important comparisons. For example, lower values for this interaction would indicate internal and external power preponderance, but would also conflate two competing situations: (1) the same state has a preponderance of both internal and external power, and (2) one side is preponderant in internal power, while the other has a preponderance of external power. Although both cases are represented by lower values on the interaction, they represent two different situations – the former exemplifying internal–external power disparity in one side’s favor, while the latter indicates a condition of power parity with one state’s disadvantage in internal capabilities compensated by its advantage in external power.

To avoid conflating these different situations we created five dichotomous variables representing all possible combinations of internal–external power parity and disparity. We first created two dichotomous variables (Internal power parity and External power parity) that take a value of 1 if the two states in a dyad were roughly similar in their internal and external power, respectively.⁷ Once these variables were coded, we created five dichotomous categories reflecting the different combinations of internal–external power relations as follows:

(C1) Internal parity–external parity: the two sides are roughly equal in their internal and external power.

(C2) Asymmetric internal–external preponderance: one side is stronger in internal power, but the other is preponderant in external power.

(C3) Internal parity–external preponderance: the two states are roughly equal in internal power, but one side has a preponderance of external power.

(C4) Internal preponderance–external parity: one side is preponderant in internal power, but the two sides are roughly equal in external power.

(C5) Symmetric (one-sided) internal–external preponderance: one state is preponderant in internal and external power.

Figure 2.

Relative external–internal power and the predicted probability of disputes: Comparison of combinations of relative internal–external power

The distinction between Symmetric (one-sided) and Asymmetric advantages in internal and external power allows us to distinguish between the two types of power disparities that would be obscured if we were to use an interaction term. As shown in Table II, two of these dichotomous categories (Internal parity–external parity and Asymmetric internal–external preponderance) indicate a type of power parity, and the three remaining categories represent power disparity. If power parity does facilitate conflict, we would expect disputes to be more likely in categories C1–C2 as opposed to C3–C5. The results using the dichotomous categories are reported in Model 2 and the predicted probabilities with 90% confidence intervals are illustrated in Figure 2.

The predicted probabilities in Figure 2 highlight some interesting relationships. Most importantly, the results show that the cases of power parity (C1 and C2) are statistically indistinguishable from one another, even though the distribution of internal and external capabilities that led to parity in power is different. Specifically, the statistically insignificant coefficient for Asymmetric internal–external preponderance shows that these types of dyads are not different in their dispute behavior from those that are equally strong internally and externally (the reference category). Intuitively, this means that the preponderance of one state’s internal power can be offset by the other’s external power advantage, even to the point that it creates a situation that is indistinguishable from the one in which both states are equal in the distribution of internal and external power. Thus, external power not only matters, but under certain conditions it can help compensate for an uneven distribution of internal power.

Figure 2 also shows that disputes are more likely under parity compared to when one side’s internal or external capabilities render them preponderant in power (when one side has a preponderance of both internal and external power, or when an external balance is offset by one state’s preponderance of internal power). In fact, dyads are least likely to experience a dispute when one side has an internal and external power advantage (Symmetric (one-sided) internal–external preponderance). These dyads are almost three times less likely to experience a dispute than when the balance of internal and external capabilities is equal. Once again, this finding runs opposite to the expectations from the balance of power school.

Conclusion

Although it is not unique to argue that power matters in international politics, we believe that previous literature has unduly focused on internal sources of power. The potential of others to join a conflict provides states with power resources outside their own capacity, and the balance of these external power resources in a dyad affects the odds that a conflict will occur. We believe that our measure of external power can be useful for testing any theory that rests on the concept of power. For example, studies testing the relative validity of the balance of power versus power preponderance theories have largely relied on internal power. We provide a first cut at understanding the effect of external power on conflict and our results are more supportive of the power preponderance school. Future tests may analyze the effect of external power on the initiation, reciprocation, and outcome of disputes. This measure can also be adapted to civil conflicts, helping to understand a rebel group’s decision to revolt and/or a government’s willingness to resist. Finally, we conclude with a possible extension. Although we assume that states have an absolute level of external power in each year, it is possible to expect a state’s external power to vary depending on its particular dyadic opponent. Our analysis should therefore be viewed as a first attempt to more accurately measure external power, and the results from our models of dispute onset suggest its usefulness for understanding conflict dynamics.

Footnotes

Replication data

All analyses were conducted using Stata 12. The dataset and do-files for the empirical analysis can be found at .

Acknowledgements

The authors would like to thank the anonymous reviewers and the editor at the Journal of Peace Research for their thoughtful comments on previous versions of this manuscript.

Notes

Appendix

Table A1.

Probit estimates of joining behavior, 1870–2000

	Coefficient (Robust S.E.)	Coefficient (Robust S.E.)
Alliance ties	.665** (.037)	.599** (.047)
Joint democracy	–	.200** (.044)
Dyadic trade with state A/Joiner’s total trade	–	.854** (.149)
Capabilities-potential joiner	–	3.607** (.323)
Contiguity	–	–.164 (.187)
Alliance portfolio similarity	–	–.461** (.066)
Contiguity * alliance portfolio similarity	–	.648* (.220)
Constant	–3.103** (.022)	–3.003** (.042)
N	209148	209148
Log-likelihood	–2279.59	–2129.99
Likelihood ratio test	χ² = 299.19, p = .000
χ²	267.97**	567.16**

* p < .05; ** p < .01.

References

Altfeld

Michael

Mesquita

Bruce Bueno de

(1979) Choosing sides in wars. International Studies Quarterly 23(1): 87–112.

Aydin

Aysegul

(2008) Choosing sides: Economic interdependence and interstate disputes. Journal of Politics 70(4): 1098–1108.

Barbieri

Katherine

Keshk

Omar

Pollins

Brian

(2008) Correlates of War Project Trade Data Set Codebook, Version 2.01 (http://correlatesofwar.org).

Bennett

D Scott

Rupert

Matthew C

(2003) Comparing measures of political similarity. Journal of Conflict Resolution 47(3): 367–393.

Bennett

D Scott

Stam

Allan

(2000) EUGene: A conceptual manual. International Interactions 26(2): 179–204.

Bremer

Stuart A

(1992) Dangerous dyads: Conditions affecting the likelihood of interstate war, 1816–1965. Journal of Conflict Resolution 36(2): 309–341.

Carter

David

Signorino

Curtis S

(2010) Back to the future: Modeling time dependence in binary data. Political Analysis 18(3): 271–292.

Corbetta

Renato

(2010) Determinants of third parties’ intervention and alignment choices in ongoing conflicts, 1946–2001. Foreign Policy Analysis 6(1): 61–85.

Danilovic

Vesna

(2002) When the Stakes Are High: Deterrence and Conflict Among Major Powers. Ann Arbor, MI: University of Michigan Press.

10.

Diehl

Paul

(1985) Contiguity and military escalation in major power rivalries, 1816–1980. Journal of Politics 47(4): 1203–1211.

11.

Gartzke

Erik

Gleditsch

Kristian Skrede

(2004) Why democracies may actually be less reliable allies. American Journal of Political Science 48(4): 775–795.

12.

Gaubatz

Kurt Taylor

(1996) Democratic states and commitment in international relations. International Organization 50(1): 109–139.

13.

Ghosn

Faten

Palmer

Glenn

Bremer

Stuart A

(2004) The MID3 data set, 1993–2001: Procedures, coding rules, and description. Conflict Management and Peace Science 21(2): 133–154.

14.

Huth

Paul

Russett

Bruce

(1984) What makes deterrence work? Cases from 1900–1980. World Politics 36(4): 496–526.

15.

Kagan

Donald

(1969) The Outbreak of the Peloponnesian War. Ithaca, NY: Cornell University Press.

16.

Leeds

Brett Ashley

Long

Andrew

Mitchell

Sara McLaughlin

(2000) Reevaluating alliance reliability: Specific threats, specific promises. Journal of Conflict Resolution 44(5): 686–699.

17.

Leeds

Brett Ashley

Mattes

Michaela

Vogel

Jeremy

(2009) Interests, institutions, and the reliability of international commitments. American Journal of Political Science 53(2): 461–476.

18.

Marshall

Monty G

Jaggers

Keith

(2002) Polity IV dataset, Computer file; version p4v2002. College Park, MD: University of Maryland.

19.

Mearsheimer

John J

(1983) Conventional Deterrence. Ithaca, NY: Cornell University Press.

20.

Mitchell

Nancy

(1999) The Danger of Dreams: German and American Imperialism in Latin America. Chapel Hill, NC: University of North Carolina Press.

21.

Morgenthau

Hans J

(1993 [1948]) Politics Among Nations: The Struggle for Power and Peace, Brief Edition. New York: McGraw Hill.

22.

Most

Benjamin

Starr

Harvey

(1989) Inquiry, Logic, and International Politics. Columbia, SC: University of South Carolina Press.

23.

Oneal

John

Russett

Bruce

(1997) The classic liberals were right: Democracy, interdependence, and conflict, 1950–1985. International Studies Quarterly 41(2): 267–294.

24.

Schelling

Thomas

(1966) Arms and Influence. New Haven, CT: Yale University Press.

25.

Signorino

Curtis

Ritter

Jeffrey M

(1999) Tau-b or not Tau-b: Measuring the similarity of foreign policy positions. International Studies Quarterly 43(1): 115–144.

26.

Singer

J David

(1987) Reconstructing the Correlates of War dataset on material capabilities of states, 1816–1985. International Interactions 14(2): 115–132.

27.

Siverson

Randolph

Emmons

Juliann

(1991) Birds of a feather: Democratic political systems and alliance choices in the 20th century. Journal of Conflict Resolution 35(2): 285–306.

28.

Siverson

Randolph

Starr

Harvey

(1991) The Diffusion of War. Ann Arbor, MI: University of Michigan Press.

29.

Vasquez

John A

(1996) Distinguishing rivals that go to war from those that do not: A quantitative comparative case study of two paths to war. International Studies Quarterly 40(4): 531–558.

30.

Waltz

Kenneth

(1979) Theory of International Politics. New York: McGraw Hill.

31.

Wayman

Frank Whelon

(1984) Bipolarity and war: The role of capability concentration and alliance patterns among major powers, 1816–1965. Journal of Peace Research 21(1): 61–78.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB