Introduction
The chances of war are significantly reduced in the relations of democratic states. Democracies interact more peacefully with each other than do non-democratic or mixed pairs of states.Footnote 1 Theoretical arguments designed to explain why democracies do not fight each other have been based on the Kantian idea of perpetual peace among liberal states, shared norms of compromise and peaceful conflict resolution, and the presence of restraining democratic institutions that prevent inter-democratic wars. Critics of the democratic peace thesis have challenged it on the grounds that some other factors – territorial stability, capitalist economy, interdependence, and globalisation – confound the joint democratic peace.Footnote 2 It has also been argued that inter-democratic peace is a consequence of political or broader cultural similarity rather than joint democracy, and peace among like regimes is not limited to democratic states.Footnote 3 The ‘autocratic peace’ literature, for example, has shown that mixed democratic–autocratic dyads were more conflict prone than either jointly democratic or jointly autocratic dyads.Footnote 4 These arguments suggest a distinct and broader theoretical formulation that makes the democratic peace thesis endogenous to cultural variation in the international system.Footnote 5
Is it joint democracy or cultural similarity that has a pacifying impact on inter-state relations? We purport that the two propositions about the cultural impact are complementary in explaining inter-democratic conduct. Further, we explore the potential of shared emancipative values of personal autonomy, equality of opportunity, respect for individual choices, and voice in community decisionsFootnote 6 to act as an explanatory force of the democratic peace. The culture that embraces these values is not coterminous with democracy, but is both prior to and necessary for the emergence of democratic institutions and norms. The theory posited in this study is one of causal complexity in which the outcome of conflict can result from multiple combinations of conditions.Footnote 7
We conceptualise culture as transmitted patterns of values and beliefs that members of a given society come to share. The benefits of using emancipative values for the purpose of testing the impact of cultural similarity and joint democracy are twofold. First, emancipative values, similarly to many democratic principles, are an outcome of ‘enlightenment humanism’ reflected in the writings of Kant, Hume, Locke, Smith, Madison, Jefferson, Hamilton, and Mill, among others.Footnote 8 Although different from democratic norms, emancipative values are associated with the habits of peaceful resolution of differences and compromise and can, therefore, be used as a proxy in the tests of the normative explanation of democratic peace. Second, emancipative values capture salient cultural variation across states and societies and can, therefore, be used for testing the impact of cultural similarities and differences on peace in state dyads. The data on the emancipative values come from the World Values Survey (WVS), which investigates social, cultural, and political changes worldwide. In addition to the WVS data, this study utilises the data from the Correlates of War Project.Footnote 9
We begin with an overview of the two theoretical paths of the cultural impact on interstate relations and discussion of how cultural similarity and shared emancipative values account for democratic comity. Next, we present the research design of the study and the findings of the statistical analysis. We conclude with a discussion of the results.
Culture and democracy: Causal pathways to inter-democratic peace
According to the normative proposition of the democratic peace theory, liberal ideology or the shared norms of democratic compromise prevent democracies from fighting one another.Footnote 10 Democracies externalise these domestic norms, therefore avoiding conflicts and interventions in relations with other democracies.Footnote 11 This line of reasoning emphasises a distinctive set of norms that guide democratic states’ conduct in relations with each other. The sources of these norms can be found in liberal ideologyFootnote 12 or culture that is central to the normative explanations of democratic peace, also known as the ‘cultural’ accounts of democratic comity.Footnote 13
This emphasis on the internal differences between democratic and autocratic states that bear important implications for their international behaviour also conceals a possibility that it is the similarity of democracies with regard to their respect of individual liberties at home that shape their behaviour abroad. These conventions of mutual respect have formed a cooperative foundation for relations among liberal democracies.Footnote 14 Yet, the same liberal ideology that distinguishes democracies from autocratic states may also be a source of conflict among them, as democratic regimes embark on the liberating crusade to safeguard the rights of citizens who are oppressed by autocratic regimes. Aware of such ‘liberal imperialism’, non-democratic states may respond with force to prevent the possibility of liberal intervention.Footnote 15
Thus, the democratic peace literature offers two culture-centered mechanisms explaining democratic comity. First, cultural predispositions toward peaceful conflict resolution reduce the chance of dispute escalation between two democratic states. Second, cultural affinity eliminates important bones of contention over the management of domestic issues, thus reducing the possibility of conflict emergence in the first place. These two propositions have been used to explain the ‘Janus-faced’ nature of democracies, which are no less likely to fight non-democratic states, but nevertheless avoid wars with each other.Footnote 16
We purport that the two propositions about the cultural impact are complementary in accounting for democratic states’ conduct.Footnote 17 This complementarity extends beyond the ‘division of labor’ whereby cultural similarity reduces possible issues of contention between democracies, while the norms of peaceful conflict resolution prevent an existing dispute’s escalation to war. Rather, we posit that the shared norms of conflict resolution and culture jointly influence each stage of inter-state relations by affecting inferences and preferences of publics and elites. The impact of shared norms of conciliation, for example, can be dampened by the presence of cultural biases, while cultural similarities may inform the perceptions of state regime. Further, we theorise the norms and principles associated with the peaceful conduct of democracies and trace explanations of democratic peace to shared ‘emancipative culture’, that is, a culture that highly values personal autonomy, equality of opportunity, respect for individual choices, and voice in community decisions.Footnote 18 In the remainder of this section we elaborate on these arguments.
Scholars from different disciplines have pointed out strong correlation between societies’ cultural orientations and their institutional formats. Predispositions toward in-group favouritism, kinship ties, and rigid norms, found in ‘collectivistic’Footnote 19 or ‘tight’Footnote 20 cultures tend to favour authoritarian institutions, while ‘loose’ cultures supporting inter-group exchange and individual creativity give rise to liberal institutions and norms.Footnote 21 In these studies on human development, political institutions were found to be interlinked with peoples’ underlying beliefs and values. ‘For democracy only becomes effective after ordinary people have acquired the resources that make them capable to practice freedoms and after they have internalized the values that make them willing to practice freedom.’Footnote 22 According to this scholarship, culture informs the nature of political order, and is also shaped and reinforced by political institutions and norms.
At the societal level, culture performs several interrelated cognitive and evaluative functions.Footnote 23 First, it serves as a communication device that fosters individual and group interactions in the first place providing a framework of shared and collective meanings. In liberal democracies, for example, citizens share beliefs in personal autonomy and inalienable rights. They also hold strong (if simplified) expectations about other countries that inform their perceptions of and interactions with citizens of those states.Footnote 24 Second, socio-cognitive cues linked to the shared cultural understandings enable categorisations and judgements. Learning about human rights abuses in another state, for instance, may trigger its categorisation as a culturally backward ‘other’. Confronting information that cues perceptions of a group-orientated culture may inform expectations of conciliatory behaviour from individuals, groups, and states ascribed with these cultural beliefs.Footnote 25 Cultures also act as motivational devices: cultural values are important sources of individuals’ attitudes that can affect their subsequent decisions and behaviour towards the world around them.
Implicit in the explanations of democratic peace are the propositions about ways, in which national leaders and citizens of democratic states use information about the regime type of a foreign state for making their preferences and decisions about bilateral relations.Footnote 26 Information about a regime type, however, is not always explicitly communicated or accessible to the public. The latter forms perceptions about other states on the basis of snippets of past and present information and the pre-existing sociocultural associations, which trigger schematic and, often, stereotypical inferences about other countries. Individuals use these inferences for drawing associations between their own and foreign states.Footnote 27 Perceiving the target nation as similar or dissimilar carries over to its perception as democratic or non-democratic because individuals can make inferences about one variable on the basis of another.Footnote 28 Further, culturally similar states typically receive more positive associations, while culturally dissimilar states may be labelled with negative out-group labels. This in-group/out-group distinction is further embedded with subjective evaluations about the ‘goodness’ of the in-group members and the ‘bad’ or threatening nature of the out-group states. These subjective judgements are used to justify differential treatment of the in-group and out-group: the in-group members deserve better treatment, while the out-group members can be subjected to violent force.Footnote 29
In this way, perceptions of cultural similarity may prevent conflict in state dyads where one or both nations may lack democratic credentials due to the fact that the leaders and publics of these states identify their nations as part of the same in-group and, therefore, worthy of cooperation. Conversely, cultural biases may impede the construction of normative affinities and, therefore, constitute a barrier to peaceful resolution of disputes involving democratic states. The breakup of Yugoslavia in 1991, for example, was mired in a series of brutal wars and entailed multiple territorial disputes that never escalated into the full-fledged fighting. The two former republics of Yugoslavia, Slovenia and Croatia, declared their independence in June 1991, but Slovenia, which spared a large-scale and protracted war with the Yugoslav People’s Army, was able to establish a multi-party representative democracy earlier than its neighbour, Croatia. The Croatian War of Independence was fought for four years and delayed democratisation by nearly a decade. Throughout the 1990s, Slovenian-Croatian relations deteriorated over the maritime border in the Gulf of Piran.Footnote 30 Despite the politicisation of the conflict by both sides and a stronger military potential of the less-democratic party of the conflict, Croatia, the two states avoided militarisation of the territorial dispute and eventually agreed to submit it to arbitration. While no single factor is sufficient to explain the absence of war between Croatia and Slovenia during that time, the cultural similarities rooted in shared religion (Catholicism), beliefs in greater autonomy and economic prosperity, and animosity toward the Serbs (the shared ‘out-group’) turned these republics into important in-group members.Footnote 31
The democratic peace researchers also pointed out multiple examples of militarised disputes, in which cultural biases washed out the conflict dampening impact of joint democracy.Footnote 32 The Philippine-American War of 1899, for example, which pitted the United States against the First Philippine Republic, was explained by the presence of ‘Western ethnocentric attitudes’ that vitiated democratic constraints on US foreign policy towards a democratic, but culturally dissimilar, adversary.Footnote 33 Cultural differences have reinforced disputes between Israelis and Palestinians as well as the conflict between India and Pakistan.Footnote 34
Not every culture embraces values of peace, toleration of differences, and mediation. From a cultural standpoint, norms that emphasise respect for individual rights, tolerance of differences, rational debate, and negotiation stem from liberal principles about individuals’ autonomy and ability to define and pursue their interests in self-preservation and material well-being.Footnote 35 Freedom is required for the pursuit of these interests, and peace is a prerequisite for freedom. Since violence and coercion are inimical to freedom, individuals are predisposed to value peace.Footnote 36 These liberal predispositions are expected to have a limiting effect by excluding war from the full range of logically possible solutions to an inter-democratic dispute, which gets resolved by peaceful means.Footnote 37 Other cultures, however, may comport with fascist regimes valorising aggressive war or other totalitarian ideologies encouraging contest for authority. Even when shared in a state dyad, these cultures may be conducive to war. Therefore, formal similarity does not fully account for democratic comity.
Previous research has shown that societies embracing the so-called ‘emancipative values’ are more likely to enjoy higher levels of democracyFootnote 38 and interpersonal trust associated with the decline of inter-state conflict.Footnote 39 These societies are also more likely to have higher levels of tolerance towards minoritiesFootnote 40 and experience a reduction in both domestic and international violence.Footnote 41 Emancipative values embrace two orientations: a liberating orientation that emphasises the freedom of choice, and an egalitarian orientation that stresses equality in realising one’s freedoms and choices, thus leading to greater tolerance of dissent and non-conformity.Footnote 42 Societies with prevalent emancipative values have evinced decreased readiness of individuals to scarify their life in wars.Footnote 43 Growing aversion to the loss of human lives has been regarded ‘a natural byproduct of rising pro-choice values: when large parts of a population begin to see life no longer as a source of threats but as a source of opportunities, sacrificing lives increasingly seems an intolerable waste of human potential.’Footnote 44 As a result, individuals willingness to fight in wars dwindles as emancipative values spread. The latter causal mechanism is different from the pacifying impact of democratic norms. A desire for global peace among the like-minded democratic states has, on occasion, given rise to liberating crusades. Emancipative values, on the other hand, spurn any kind of conflict. Societies embracing emancipative norms are reluctant to accept the loss of human life in wars for any cause.
By reducing people’s willingness to fight wars, emancipative values produce a ‘monadic’ impact of emancipative culture. In this causal mechanisms, perceptions of other states as culturally similar are not necessary for triggering the limiting effect of emancipative values on the warlike conduct of states that share them. Still, the limiting effect of emancipative values on warlike conduct will be greatest in the dyads of states that share them. In the mixed-culture dyads, escalation of conflict into war is possible as nations embracing values compatible with war may resort to or threaten military force as a means of conflict resolution prompting self-defence by the states, in which cultural values are loath to military force.
To conclude, cultural similarity in state dyads is linked to socio-cognitive mechanisms of the formation of judgement and perceptions resulting in sentiments of amity or affinity between two states, thus reducing the chances of war between them. Conversely, cultural dissimilarity leads to perceptions of bias that may result in impediments to peaceful conflict resolution. Formal cultural similarity, however, is insufficient to explain differences in the levels of conflict in inter-democratic, mixed, and non-democratic pairs of states. In particular to outcomes of peace, the shared cultures must be based in emancipative values. In addition to, but also separate from, joint democracy has been demonstrated to increase the probability of peace among states both theoretically and empirically. While both cultural similarity and joint democracy are independently theorised to facilitate outcomes of peace, it is also reasonable to assume that the existence of both cultural similarity and joint democracy within a dyad might further reduce the probability of conflict outcomes among the respective states. Figure 1 provides a visual representation of the three causal pathways.
Figure 1 Impact of cultural similarity, emancipative culture, and democracy on conflict.
The theory posited in this manuscript is one of causal complexity in which the outcome of conflict (Y) results from multiple combinations of conditions (X 1 =level of cultural similarity, X 2 =existence of joint democracy; X 3 = shared emancipative culture).Footnote 45 We presented three causal pathways to the outcomes of conflict.Footnote 46 Put succinctly, the premise of conjunctural causation suggests that dyads without shared culture have higher probability for conflict and, as a separate effect, dyads that are not jointly democracy have higher probability for conflict. It also suggests that when both shared culture and joint democracy are missing from the dyad, the probability for conflict is higher.
For many democratic peace scholars, cultural explanations are essentially structural state-level explanations. State culture overrides exogenous differences in leaders’ beliefs, which become theoretically insignificant.Footnote 47 Others, however, have focused on the role of elite level normsFootnote 48 and suggested that differences between leaders in democratic states may condition the impact of democratic culture.Footnote 49 We rely on assumptions and findings of value research that has confirmed an ‘ecological effect’ of values when those are prevalent in the society.Footnote 50 To use the language of statistics, each state, even with heterogeneous cultural traditions, will have central tendencies or ‘means’ of cultural beliefs with a greater or lesser variance characterising the distribution of these beliefs in the population. Prevalent values become intuitively known to large segments of the society through different forms of communication, which, in turn generates a ‘mental climate’ felt by the members of the society. This mental climate constitutes state culture.Footnote 51
As products of their culture, state decision-makers are the bearers of the salient cultural predispositions characterising the rest of the society, regardless of the directional relationship of cultural influences between the public and political elites. Both democratic and authoritarian political systems are sustained by legitimacy. Democratic legitimacy has been associated with various procedural mechanisms, but also a degree of congruence between the fundamental values, worldviews, and patterns of conduct espoused by the political system, including political elites, and by the population.Footnote 52 Authoritarian regimes have used state-controlled public and private informational space to foster certain ways of thinking in the citizens, and create possibilities for their support for and compliance with the policies of ruling administrations.Footnote 53 These discursive legitimation strategies often include appeals to the broader ideology, history, and culture, in addition to the superior decision-making and moral qualities of the ruling elite.Footnote 54 Research has found a high degree of cultural congruence between political elites’ and citizens’ values and opinion in the authoritarian states.Footnote 55
Research design
Measuring culture by emancipative values
This study adopts a widely accepted understanding of culture as shared values, norms, beliefs, and attitudes that relate to social and political relations and context.Footnote 56 Despite variations in the terminology and emphasis found in the wide spectrum of definitions of culture, they all point out the existence of shared meanings that provide a degree of order and direction with regard to the individual and group conception of the relationship with their social and political enforcement.Footnote 57
To measure culture, we chose a set of individual predispositions toward autonomy, choice, equality, and voice known as emancipative values. With their roots in the Enlightenment principles and ideas of humanism, emancipative values are comparable with democratic norms, but not synonymous with those. Some modern democratic states such as South Korea, the Baltic states of Estonia, Latvia, and Lithuania, and the Republic of Georgia, to name but a few, are recognised democracies with less prevalent emancipative values.Footnote 58 Emancipative values are broader in their content than democracy. Their egalitarian dimension, for example, is broadly comparable with not only liberal, but also socialist, communist, libertarian, and other philosophies and traditions. Extreme egalitarianism can lead to mob rule threatening the institutions of democracy, while some non-egalitarian practices, particularly those concerning race and gender, are still common in democratic states.
Emancipative values are also suitable for cross-cultural comparisons, although they are not the only measures of the salient cultural differences and similarities between societies. A different and commonly used cultural typology classifies cultures into collectivistic and individualistic ones.Footnote 59 Other scholars have devised the measures of cultural ‘embeddedness’ and ‘autonomy’Footnote 60 and the ‘tightness’ vs ‘looseness’ of cultures depending on how they tolerate deviant behaviour.Footnote 61 Emancipative values tap the same salient societal differences as other measures of culture, but they have been found to capture these differences better on various indicators of economic development, technological advancement, and democratic achievements.Footnote 62 Emancipative values, therefore, are the preferable measure of culture.
We use the World Values Survey as the main source of data on emancipative values. Our analysis is performed on data from the WVS waves 1981–4, 1990–4, 1995–8, 2000–04, 2005–09, as well as the 2010 data from the WVS wave 2010–14. Not all countries participate in the WVS, and not all countries are included in each subsequent wave. A total of 159 countries are included into the analysis, but not all of these states appear in dyads with one another. Some countries appear in dyadic observations of the sample more often than others.Footnote 63 To increase our sample and thus reliability of our results, values for the emancipatory indices for the first wave of the survey (WVS 1981–4), in which questions used in the construction of indices of emancipative values were either not asked or indices were not calculated, were imputed using multiple imputation techniques.Footnote 64 The underlying logic behind this procedure is the fact that cultural values take time to change, and the aggregate values of indices across the survey waves change at a very slow rate.
Following the exact design created by Welzel in Freedom Rising, we used 12 survey questions to measure the four domains of emancipative values and construct the final index.Footnote 65 The 12 items were normalised into a scale ranging from a minimum of 0 for the least emancipative orientation, to a maximum of 1.0, for the most emancipative orientation. Then, the groups of items characterising various domains of emancipative values were averaged to form sub-indices, which were further averaged into the overall index of emancipative values.Footnote 66
Following the precedent of the study of cultural values at national level and research on national values,Footnote 67 we utilised country means of the emancipative values index. To measure the similarity in the emancipative values in state dyads, we subtracted the means of the emancipative values of State A from the means of the emancipative values of State B, and took the absolute value of the difference. The high values on this variable denote greater cultural ‘distance’ between the states in the dyad and, therefore, lower cultural comparability, whereas the low values indicate high cultural similarities within the country dyad.
We opted for subtraction rather than addition of scores on the value variables within dyads for the following reason. According to the cultural similarity argument, it is the cultural similarity/dissimilarity that drives peace/conflict within state dyads. The addition of scores on the value variables would obscure whether the cultural scores were identical within the dyad. The high cultural values formed by summation would denote the overall higher scores on cultural values within the dyad, but not necessarily their similarity. To illustrate this point, a sum of 1 of the normalised value scores for a state dyad may mean two different things: (1) the two states within the dyad have values of 0.5 on the value dimension, which means they are culturally similar; or (2) one may have a score of 0.1 (0.2, 0.3, etc.) and another one – the score of 0.9 (0.9, 0.7, etc.) meaning that they are culturally dissimilar but still have a joint emancipative value score of 1. In other words, the summation would not allow us to determine the extent to what states within dyads are culturally similar or dissimilar. Subsequently, we opted for the subtraction of cultural scores within state dyads to tap more precisely the cultural distance within the state dyad. To offset heteroskedasticity common to time series data, we normalised all variables by logging them. It also helped to linearise the elastic relationship between the dependent variable and the logged independent variables.Footnote 68 Figures 2 and 3 present descriptive statistics of the untransformed average emancipative scores and distances within state dyads across years.
Figure 2 Mean, minimum, and maximum of emancipative scores by year.
Figure 3 Cultural distances within state dyads by year.
In addition to the cultural similarity variable that we use to test the impact of similarities/differences in state dyads on their propensity to engage in militarised disputes, we created a dummy variable with a value of ‘1’, if both states in the dyad had scores in the 50th percentile of the emancipative values measure, and ‘0’ otherwise. To put it differently, when both states in the dyad had higher than average convictions in the importance of liberty, choice, voice, and equality, the dyad was assigned a value of ‘1’, and ‘0’ otherwise. If the cultural similarity variable is indifferent to highs and lows on emancipative values within the dyad, the dummy variable distinguishes between the pairs of states that share emancipative values and those that share other cultural beliefs. Since emancipative values are comparable with democratic norms, the inclusion of this variable will allow us to test for the impact of normative proposition of the democratic peace theory.
As discussed in the theoretical section, the influence of state similarity and emancipative values on state conduct are intertwined and perceptions of similarity by states with emancipative cultures are not necessary for the emancipative values to exert a limiting impact on these states’ warlike conduct. However, these influences would be felt the most in dyads of states sharing these emancipative values and norms. For the cultural similarity argument, perceptions of other states are important. While our measures of culture do not measure perceptions of other states directly, we assume that prevalent cultures provide sufficient informational cues that will enable individuals from other states to make sense of other cultures and form attitudes toward countries they know little about.Footnote 69
Further, cultural values are a property of individuals. Questions, therefore, can arise about whether the mean scores of the individual responses collected from the nationally representative samples provide a valid estimation of value prevalence at the societal level. From the methodological standpoint, the plausibility of this assumption hinges on the qualities of the distribution of responses: if responses from the members of the same state form a normal distribution with a single peak point around the average value-position of all respondents, then there is a real central tendency around which the individual value positions of all members of a society gravitate. The earlier WVS studies showed that national mean scores of emancipative values of all states included into the WVS data meet this requirement.Footnote 70
Dependent and other independent variables
The outcome (dependent) variable in this study is the existence of a militarised interstate dispute (MID) (following the COW criteria), which serves as a proxy for inter-state war for each state dyad coded on an annual basis as (1) if the dyad experienced at least one MID in that year and (0) if it did not. This is also known as ‘ongoing MIDs’ or ‘conflict involvement’ for the countries in the dyad. Differences in the terminology reflect two approaches to specifying the dependent variables in the studies of international conflict based on time-series cross-sectional data. The first approach, which is the approach used in this study, is to code every time unit (generally years), in which there is conflict between the states in the dyad. The second approach is to code only the onset of an event as one and set the remaining observations during which conflict continues to a zero value.Footnote 71 We include all directed state dyads between years 1981 and 2010. Aware of the problems and limitations in using dyadic modeling (in particular, the temporal dependence of data), we introduced a restricted cubic spline model, further discussed below.Footnote 72
The democratic peace scholarship has examined countless possible covariates influencing the democratic peace. In an effort to keep our statistical models simple, we chose to focus on the explanatory factors that consistently appear in the democratic peace studies. Thus, other independent variables included in the models are the regime type of the dyad, the contiguity of the states, existing alliances within dyads, relative military capabilities, and the presence or absence of major powers.
Much of the democratic peace research suggests that peace is best explained by the level of ‘democraticness’ of the dyad as a whole rather than by the individual polity scores of the states that comprise it.Footnote 73 We echo this finding empirically and follow in the footsteps of our predecessors by including a joint democracy variable measured using the standard Polity IV data.Footnote 74 Polity data provide an 11-point 0 to 10 index of a regime type – DEMOC – based on formal constraints on the executive and institutional support for democracy. We compute joint democracy as a dichotomous variable which is coded ‘1’ when both states in the dyad receive Polity scores of six or more and coded ‘0’ otherwise.Footnote 75
GDP is calculated as within dyad difference in per capita GDP.Footnote 76 It is the difference between the higher and lower monadic scores in per capita GDP (another way to define it is population weighted gross domestic product). GDP values are logged. Contiguity is a dummy variable that is coded ‘1’ if states within the dyad are land contiguous or separated by less than 12 miles of water and coded ‘0’ if the states are not contiguous. We include a dummy variable that is coded ‘1’ if at least one state in a dyad is one of the five post-Second World War major powers (China, France, United States, United Kingdom, and USSR/Russia), and ‘0’ otherwise. To measure relative capabilities of states within a dyad, we used the COW military capabilities index composed of the weighted average of a country’s share of the system’s total population, urban population, energy consumption, iron and steel production, military manpower, and military expenditures.Footnote 77 Capabilities ratio equals the natural log of the ratio of the CINC (Composite Index of National Capability) score. We use a binary indicator to code alliances where ‘1’ denotes state dyads linked by a mutual defence treaty, a neutrality or nonaggression pact, or an entente, and 0 if none of these treaties are present. The coding is based on the COW Alliance Dataset.Footnote 78
All models are executed using a random effects logit model. This model accounts for the binary dependent variable, but also allows for varying slope estimates based on each group unit (dyad) observation. All independent variables were lagged by one year.Footnote 79 We used a restricted cubic spline model in order to flexibly and non-linearly model the onset of war within a dyad of states (our dependent variable) as a function of the specified cultural differences within dyads. This required the use of piecewise cubic polynomials to model observational dependence over time, whose placements were determined based on both the distribution of the data points and on theoretical precedent.Footnote 80 In short, the random effects combined with the cubic spline model loosen the assumption of independence, allowing for flexible modeling of cross-unit and temporal dependence. To examine the independent effects of joint democracy and cultural similarity, we executed models containing these independent variables separately as well as together. To explore links between these two main concepts, we also examined the interaction of these two terms.Footnote 81 In keeping with our arguments about the two causal influences of culture on the inter-state disputes, we substituted the cultural similarity variable with the joint high emancipative values dummy, used as a proxy for the joint normative democratic peace, in some model specifications.
Results
Table 1 reports the findings of the seven random effect logistic regression analyses testing the impact of cultural similarities and joint democracy within state dyads on militarised dispute between these states. Model 1 includes both cultural similarity and joint democracy variables, while Models 2 and 3 report findings from regressions of militarised disputes on cultural similarity or joint democracy respectively. Model 4 includes an interactive term of cultural similarity and joint democracy. In Model 5, we substitute cultural similarity with the high joint emancipative values variable, while keeping joint democracy, which is excluded in Model 6. The last model (Model 7) includes an interactive term of high joint emancipative values and joint democracy. The appendix (Table A1) reports results in odds ratios that allow for comparison of the effect of variables of interest across models.Footnote 82
Table 1 Random effects logistic regression of dyadic analysis.
Note: *p≤0.05 **p≤0.01
We discussed the conceptual distinction between emancipative values and democratic norms above. Statistically, joint emancipative scores and joint democracy scores are not correlated to cause collinearity in the executed models returning a Pearson’s R of 0.01. Among the state dyads with higher than average emancipative values scores, only about 18.2 per cent are jointly democratic dyads.
The cultural similarity variable is positive and statistically significant (p≤0.01) when appears jointly with the joint democracy variable (Model 1) and by itself (Model 2). A positive coefficient on the cultural similarity variable suggests that a higher ‘distance’ on the emancipative value scores of the two states in a dyad is associated with a higher propensity for conflict within the dyad. Reversely, a shorter distance on the emancipative values in a state dyad (that is, greater cultural similarity of states in the dyad) is associated with lower likelihood of militarised disputes within this state dyad. Substantively, when cultural ‘distance’ within a state dyad measured by the difference in aggregate value scores increases by one unit (denoting more dissimilar states), the odds of conflict increase by 21 per cent.
The introduction of the cultural similarity variable did not vitiate the impact of joint democracy on conflict within the dyads of states. The joint democracy variable returned negative and significant coefficients suggesting that two democracies are less likely to experience a militarised dispute than other pairs of states. Substantively, a jointly democratic dyad has an 81.8 per cent decrease in odds of experiencing conflict over non-jointly democratic dyads. This is reflected in Model 3, where joint democracy appears by itself, returning a statistically significant effect (p≤0.01) in an expected direction demonstrating the ability of this model to reproduce the main findings of the democratic peace thesis, namely, jointly democratic dyads are less likely to go to war. The inclusion of an interactive ‘cultural similarity/joint democracy’ term did not wash out the independent impacts of both cultural similarity and joint democracy (Model 4). The interactive term appears to be in the expected direction suggesting that even in the dyads of democratic states, increased cultural difference makes conflict more likely. However, the interactive term did not reach a conventional level of statistical significance (p≤0.05).
The high emancipative value scores dummy returned statistically significant coefficients (p≤0.01) in the expected direction in Models 5 and 6. Dyads where both states are in the 50th percentile for emancipatory scores experience between a 28.1 and 53 per cent decrease in the odds of experiencing MIDs in Models 5 and 6, respectively. Similarly to Model 1, the impact of high emancipative value scores in state dyads does not wash out the impact of the joint democracy variable. Jointly democratic dyads have 74.7 per cent decreased odds of experiencing conflict over non-democratic dyads (Model 5).
In Model 7 containing an interaction term, the direct impact of the high emancipative values becomes insignificant, while joint democracy retains its significant impact. Everything else held constant, jointly democratic dyads have 68.8 per cent decrease in odds of experiencing conflict over non-jointly democratic dyads. The interaction terms is significant as well (p≤0.01) and demonstrates an additional normative impact of emancipative values on joint democracies. Democratic states, which share high emancipative values, have 51 per cent decreased odds of experiencing conflict over non-democratic states that do not share emancipative culture.
Concerns with multicollinearity affecting the results in models with interactive terms are common. One of the possible ‘tests’ for multicollinearity in logistic regression involves the generation of restricted and full models. This is precisely the strategy that we followed in this study: in Models 1, 2, 3, 5, and 6 we added in one variable of interest at a time and then tested the models with the interaction terms watching for major changes in standard errors. We observed virtually no change in standard errors in either Model 4 or Model 7.Footnote 83 An interactive term in Model 4 returned an insignificant coefficient suggesting that state dyads that are culturally similar and democratic are no more or less likely to go to war than dyads that are culturally dissimilar and democratic, or culturally similar and non-democratic. However, two states that are democratic and score above the average on emancipative values are considerably less likely to engage in militarised disputes, holding other variables constant.
Among the control variables, contiguity, alliances, and major powers returned positive and statistically significant results (p≤0.01) consistent with the findings of the democratic peace thesis. State dyads that are contiguous are more likely to experience a MID, as are allied dyads, and dyads in which at least one of the states is a major power. More specifically, using Model 1 as a reference, states that are contiguous are almost 99 per cent more likely to experience a MID than dyads in which the states are not contiguous. An impact of capabilities’ differential is small but highly significant: a one unit increase in the ratio of capabilities of the larger state to the smaller state decreases the odds of war 0.4 per cent, everything else being equal. State dyads with treaties experience a 95 per cent increase in odds of experiencing conflict over states where no treaties exist (Model 1). While this finding contradicts earlier research in democratic peace, it echoes more recent scholarship that suggests states, specifically democratic ones, are more likely to violate the terms of alliance agreements because the costs of maintaining alliances become high as power dynamics shift.Footnote 84 In addition, these results echoes the findings suggesting that democracies are more likely to form alliances with highly autocratic regimes than with other democracies, which, if we assume the validity of democratic peace, that higher frequency of democratic-autocratic alliances would suggest a negative relationship between the existence of an alliance within a dyad as these two regime types are more likely to go to war with one another.Footnote 85
Discussion and conclusion
This study was motivated by an ongoing debate over the sources of inter-democratic peace. Our contributions to the extant democratic peace scholarship lie first in examining whether shared cultural values conceptualised as emancipative orientations of people toward autonomy, freedom, choice, and voice mitigate conflict within state dyads. Second, we explored whether the impact of joint democracy is sustained if we control for cultural similarities and differences in state dyads. The findings of our analysis add support to the conclusions of the two trends of earlier research on democratic peace: one highlighting the importance of liberal ideology and culture for inter-democratic peace, and another one pointing out the impact of political and cultural similarities within state dyads on the propensity of war state dyads.
We were able to reproduce the results of the Democratic Peace scholarship supporting the pacifying impact of democracy on inter-state relations. However, the cultural similarity variable retained its significance when included into the model alongside the joint democracy variable, and after the removal of joint democracy indicating that it does not falsely represent significance that should be attributed to the democracy score. To explicate the effect of cultural similarity and differences on inter-state relations, consider the Greek-Turkish rivalry, one of the oldest enduing conflicts between neighbours worldwide. In July 1974, this rivalry triggered Turkey’s intervention of Cyprus in response to a coup staged by the military Junta in Greece. Although, the Republic of Cyprus saw the restoration of democratic order the same month, Turkey launched another invasion in August 1974 capturing about 40 per cent of Cyprus territory. Turkey had a new democratic government elected in 1973 (POLITY score of 9 during 1973–9), while Cyprus was a constitutional democracy (POLITY score of 10 since 1974). Turkey, however, perceived Greece and Greek Cypriots as the ‘other’, and this out-group classification of Turkey and Turkish Cypriots was shared by Greek Cypriots and Greece. While representing an oversimplification of a complex history of inter-ethnic relations between Turkish and Greek Cypriots and inter-state relations between Turkey and Greece, the perceived ‘otherness’ rooted in differences in religion, language, and lifestyle that were deployed in the nationalist rhetoric and historical narratives of the parties of the conflict contributed to antagonisms between Turkey and Greece. Recall that emancipative values capture cross-national cultural differences as evinced in Turkey’s emancipative score of 0.35 and Cyprus’s of 0.44.Footnote 86
Formal cultural similarity, however, may not suffice to prevent militarised disputes in similar dyads. The Maritime Southeast Asia comprised of several democratic states, including India (the Andaman and Nicobar Islands), Indonesia, East Malaysia, and Philippines, among others, has been mired in low-intensity conflict. Since 2014, Indonesia blew up, sank, or destroyed over three hundred fishing boats from Philippines, Vietnam, and Malaysia in an effort to root out poaching.Footnote 87 Ecuador and Peru, also culturally similar and democratic, fought a brief sequel of the Paquisha war in 1995. Therefore, it is not only the formal similarity that matters in reducing inter-state violence, but also the content of values shared by states. We found that the prevalence of higher than average emancipative values reduces the changes of MID by 42 per cent even when controlling for democracy, which reduces the probability of MID by 0.2. The main effects of the high emancipative value scores, however, was washed out in the interaction model suggesting that a conditional effect of high emancipative scores on democracy: state dyads that are democratic and that share emancipative values they have a lower probability for conflict than jointly democratic states.
The Croatia-Slovenia border conflict discussed above fits this pattern well. During the 1990s, when Slovenia established a multi-party representative democracy (POLITY democracy score of 10 since 1991) while Croatia’s democratisation was delayed (POLITY democracy scores of 0 and 1 between 1991 and 1999), the two countries avoided militarisation of their border dispute. The people of both nations shared predispositions toward autonomy, freedom, choice, and voice captured by higher than average emancipative value scores (Croatia=0.5; Slovenia=0.54). The more prosperous republics of the former Yugoslavia with higher standards of living, these two nations were among the first to express their support for decentralisation and democracy in the Yugoslavia, and, later, declare their independence from the union.Footnote 88
The democratic peace theory has come to have real-world implications. Citing the joint democratic peace proposition, policymakers in the US and other democratic states have drawn a link between their states’ security and the spread of democracy, which has led in turn to the democracy promotion crusade. Democracy promotion strategies, however, have zeroed in on the competitive political process and accountable and transparent government as the primary dimensions of democratisation as opposed to the deeper attitudinal and cultural transformation of the democratised societies.Footnote 89
As demonstrated in this study, political structures are shaped and sustained by the underlying cultural values. Meaningful democracy would not emerge and survive without the spread of predispositions toward autonomy, freedom, choice, and voice in the society. Whether or not newly democratised states stay on the democratic path or regress to authoritarianism, and whether or not they resort to force for resolving disputes with the fellow democracies largely depends on how prevalent the emancipative values are in their societies. The prevalence of emancipative values suggests a possibility for ‘monadic’ democratic peace, whereby countries embracing these value spurn violence as incompatible with humanistic ideals recognising humanity and inimical to their ability to engaging life’s rising opportunities. As a result of the latter, ‘people’s valuation of life changes profoundly: readiness to sacrifice one’s life gives way to an increasing insistence on living it’.Footnote 90
If it is the specific aspects of culture, particularly the values of autonomy, freedom, voice, and choice, rather than democracy, that is an antidote for future wars, it calls for a different set of policy tools and programmes of actions. The rejection of violence can be directly addressed by the educational base and is implicit in furthering human rights and gender equality. The prevention of conflict by tackling root causes can be addressed through sustainable development, among other things. Education and open communication can work together to begin solving problems through negotiation and dialogue.Footnote 91
Acknowledgements
The early work on this project was supported, in part, by the Thompson award from the Department of Political Science at the University of Kansas. The authors express sincere thanks to the anonymous reviewers and the editorial board for valuable feedback on the drafts of the manuscript.
Mariya Omelicheva is Professor of Strategy at National War College, National Defense University. She specialises in counterterrorism and human rights, crime-terror nexus, democracy promotion, and authoritarianism. The views and conclusions contained in this study are those of the author and should not be interpreted as representing the official policies, either expressed or implied, of the Department of Defense or the US Government.
Brittnee Carter is Assistant Professor of Political Science at the University of Kansas.
Appendix
Table A1 Odds ratios of random effects logistic regression of dyadic analysis.
