Democratic peace theory is a theory which posits that democracies are hesitant to engage in armed conflict with other identified democracies.[a] Among proponents of the democratic peace theory, several factors are held as motivating peace between democratic states:
Those who dispute this theory often do so on grounds that it conflates correlation with causation, and that the academic definitions of 'democracy' and 'war' can be manipulated so as to manufacture an artificial trend (Pugh 2005).
Though the democratic peace theory was not rigorously or scientifically studied until the 1960s, the basic principles of the concept had been argued as early as the 1700s in the works of philosopher Immanuel Kant and political theorist Thomas Paine. Kant foreshadowed the theory in his essay Perpetual Peace: A Philosophical Sketch written in 1795, although he thought that a world with only constitutional republics was only one of several necessary conditions for a perpetual peace. Kant's theory was that a majority of the people would never vote to go to war, unless in self-defense. Therefore, if all nations were republics, it would end war, because there would be no aggressors. In earlier but less cited works, Thomas Paine made similar or stronger claims about the peaceful nature of republics. Paine wrote in "Common Sense" in 1776: "The Republics of Europe are all (and we may say always) in peace." Paine argued that kings would go to war out of pride in situations where republics would not (Levy & Thompson 2011; Paine 1945, p. 27). French historian and social scientist Alexis de Tocqueville also argued, in Democracy in America (1835–1840), that democratic nations were less likely to wage war.[b]
Dean Babst, a criminologist, was the first to do statistical research on this topic. His academic paper supporting the theory was published in 1964 in Wisconsin Sociologist (Babst 1964); he published a slightly more popularized version, in 1972, in the trade journal Industrial Research (Babst 1972). Both versions initially received little attention.
Melvin Small and J. David Singer (1976, pp. 50-69) responded; they found an absence of wars between democratic states with two "marginal exceptions", but denied that this pattern had statistical significance. This paper was published in the Jerusalem Journal of International Relations which finally brought more widespread attention to the theory, and started the academic debate. A 1983 paper by political scientist Michael W. Doyle contributed further to popularizing the theory. Rudolph J. Rummel was another early researcher and drew considerable lay attention to the subject in his later works.
Maoz and Abdolali (1989)[incomplete short citation] extended the research to lesser conflicts than wars. Bremer (1992) and Maoz and Russett (1993) found the correlation between democracy and peacefulness remained significant after controlling for many possible confounding variables. This moved the theory into the mainstream of social science. Supporters of realism in international relations and others responded by raising many new objections. Other researchers attempted more systematic explanations of how democracy might cause peace (Köchler 1995), and of how democracy might also affect other aspects of foreign relations such as alliances and collaboration (Ray 2003).
There have been numerous further studies in the field since these pioneering works.[c] Most studies have found some form of democratic peace exists, although neither methodological disputes nor doubtful cases are entirely resolved (Kinsella 2005).
Research on the democratic peace theory has to define "democracy" and "peace" (or, more often, "war").
Democracies have been defined differently by different theorists and researchers; this accounts for some of the variations in their findings. Some examples:
Small and Singer (1976) define democracy as a nation that (1) holds periodic elections in which the opposition parties are as free to run as government parties, (2) allows at least 10% of the adult population to vote, and (3) has a parliament that either controls or enjoys parity with the executive branch of the government.
Doyle (1983)[incomplete short citation] requires (1) that "liberal regimes" have market or private property economics, (2) they have policies that are internally sovereign, (3) they have citizens with juridical rights, and (4) they have representative governments. Either 30% of the adult males were able to vote or it was possible for every man to acquire voting rights as by attaining enough property. He allows greater power to hereditary monarchs than other researchers; for example, he counts the rule of Louis-Philippe of France as a liberal regime.
Ray (1995) requires that at least 50% of the adult population is allowed to vote and that there has been at least one peaceful, constitutional transfer of executive power from one independent political party to another by means of an election. This definition excludes long periods often viewed as democratic. For example, the United States until 1800, India from independence until 1979, and Japan until 1993 were all under one-party rule, and thus would not be counted under this definition (Ray 1995, p. 100).
Rummel (1997) states that "By democracy is meant liberal democracy, where those who hold power are elected in competitive elections with a secret ballot and wide franchise (loosely understood as including at least 2/3 of adult males); where there is freedom of speech, religion, and organization; and a constitutional framework of law to which the government is subordinate and that guarantees equal rights."
The above definitions are binary, classifying nations into either democracies or non-democracies. Many researchers have instead used more finely grained scales. One example is the Polity data series which scores each state on two scales, one for democracy and one for autocracy, for each year since 1800; as well as several others.[d] The use of the Polity Data has varied. Some researchers have done correlations between the democracy scale and belligerence; others have treated it as a binary classification by (as its maker does) calling all states with a high democracy score and a low autocracy score democracies; yet others have used the difference of the two scores, sometimes again making this into a binary classification (Gleditsch 1992).
Several researchers have observed that many of the possible exceptions to the democratic peace have occurred when at least one of the involved democracies was very young. Many of them have therefore added a qualifier, typically stating that the peacefulness apply to democracies older than three years (Doyle 1983[incomplete short citation], Russett 1993, Rummel 1997, Weart 1998). Rummel (1997) argues that this is enough time for "democratic procedures to be accepted, and democratic culture to settle in." Additionally, this may allow for other states to actually come to the recognition of the state as a democracy.
Mansfield and Snyder (2002, 2005), while agreeing that there have been no wars between mature liberal democracies, state that countries in transition to democracy are especially likely to be involved in wars. They find that democratizing countries are even more warlike than stable democracies, stable autocracies or even countries in transition towards autocracy. So, they suggest caution in eliminating these wars from the analysis, because this might hide a negative aspect of the process of democratization (See Owen 2005 for an online description). A reanalysis of the earlier study's statistical results (Braumoeller 2004) emphasizes that the above relationship between democratization and war can only be said to hold for those democratizing countries where the executive lacks sufficient power, independence, and institutional strength. A review (Ray 2003) cites several other studies finding that the increase in the risk of war in democratizing countries happens only if many or most of the surrounding nations are undemocratic. If wars between young democracies are included in the analysis, several studies and reviews still find enough evidence supporting the stronger claim that all democracies, whether young or established, go into war with one another less frequently (Ray 1998, Ray 2003, Hegre 2004), while some do not (Schwartz & Skinner 2002, p. 159).
Quantitative research on international wars usually define war as a military conflict with more than 1000 killed in battle in one year. This is the definition used in the Correlates of War Project which has also supplied the data for many studies on war. It turns out that most of the military conflicts in question fall clearly above or below this threshold (Ray 1995, p. 103).
Some researchers have used different definitions. For example, Weart (1998) defines war as more than 200 battle deaths. Russett (1993, p. 50), when looking at Ancient Greece, only requires some real battle engagement, involving on both sides forces under state authorization.
Militarized Interstate Disputes (MIDs), in the Correlates of War Project classification, are lesser conflicts than wars. Such a conflict may be no more than military display of force with no battle deaths. MIDs and wars together are "militarized interstate conflicts" or MICs. MIDs include the conflicts that precede a war; so the difference between MIDs and MICs may be less than it appears.
Statistical analysis and concerns about degrees of freedom are the primary reasons for using MID's instead of actual wars. Wars are relatively rare. An average ratio of 30 MIDs to one war provides a richer statistical environment for analysis (Mousseau & Shi 1999).
Most research is regarding the dyadic peace, that democracies do not fight one another. Very few researchers have supported the monadic peace, that democracies are more peaceful in general. There are some recent papers that find a slight monadic effect. Müller and Wolff (2004), in listing them, agree "that democracies on average might be slightly, but not strongly, less warlike than other states," but general "monadic explanations is neither necessary nor convincing." They note that democracies have varied greatly in their belligerence against non-democracies.
Some scholars support the democratic peace on probabilistic grounds: since many wars have been fought since democracies first arose, we might expect a proportionate number of wars to have occurred between democracies, if democracies fought each other as freely as other pairs of states; but proponents of democratic peace theory claim that the number is much less than might be expected (Bremer 1992, Bremer 1993, Gelditsch 1992, Doyle 1983[incomplete short citation]). However, opponents of the theory argue this is mistaken and claim there are numerous examples of wars between democracies (Schwartz & Skinner 2002, p. 159).
Historically, cases commonly cited as exceptions include the Sicilian Expedition, the Spanish-American War, and more recently the Kargil War (White 2005). Doyle (1983)[incomplete short citation]) cites the Paquisha War and the Lebanese air force's intervention in the Six-Day War. The total number of cases suggested in the literature is at least 50. The data set Bremer (1993) was using showed one exception, the French-Thai War of 1940; Gleditsch (1995) sees the (somewhat technical) state of war between Finland and UK during World War II, as a special case, which should probably be treated separately: an incidental state of war between democracies during large multi-polar wars (Gowa 1999; Maoz 1997, p. 165). However, the UK did bomb Finland, implying the war was not only on paper. Page Fortna (2004) discusses the 1974 Turkish invasion of Cyprus and the Kargil War as exceptions, finding the latter to be the most significant. However, the status of these countries as being truly democratic is a matter of debate. For instance, in Spain in 1898, two parties alternated in the government in a controlled process known as el turno pacífico, and the caciques, powerful local figures, were used to manipulate election results, and as a result resentment of the system slowly built up over time and important nationalist movements as well as unions started to form. Similarly, the Turkish intervention in Cyprus occurred only after the Cypriot elected government was abolished in a coup sponsored by the military government of Greece.
Limiting the theory to only truly stable and genuine democracies leads to a very restrictive set of highly prosperous nations with little incentive in armed conflict that might harm their economies, in which the theory might be expected to hold virtually by definition.
One advocate of the democratic peace explains that his reason to choose a definition of democracy sufficiently restrictive to exclude all wars between democracies are what "might be disparagingly termed public relations": students and politicians will be more impressed by such a claim than by claims that wars between democracies are less likely (Ray 1998, p. 89).
One problem with the research on wars is that, as the Realist John Mearsheimer (1990, p. 50) put it, "democracies have been few in number over the past two centuries, and thus there have been few opportunities where democracies were in a position to fight one another". Democracies have been very rare until recently. Even looser definitions of democracy, such as Doyle's, find only a dozen democracies before the late nineteenth century, and many of them short-lived or with limited franchise (Doyle 1983[incomplete short citation]; Doyle 1997, p. 261). Freedom House finds no independent state with universal suffrage in 1900 (Democracy's Century 1999).
Wayman (2002), a supporter of the theory, states that "If we rely solely on whether there has been an inter-democratic war, it is going to take many more decades of peace to build our confidence in the stability of the democratic peace".
Many researchers have reacted to this limitation by studying lesser conflicts instead, since they have been far more common. There have been many more MIDs than wars; the Correlates of War Project counts several thousand during the last two centuries. A review (Ray 2003) lists many studies that have reported that democratic pairs of states are less likely to be involved in MIDs than other pairs of states.
Another study (Hensel, Goertz & Diehl 2000) finds that after both states have become democratic, there is a decreasing probability for MIDs within a year and this decreases almost to zero within five years.
When examining the inter-liberal MIDs in more detail, one study (Wayman 2002) finds that they are less likely to involve third parties, and that the target of the hostility is less likely to reciprocate, if the target reciprocates the response is usually proportional to the provocation, and the disputes are less likely to cause any loss of life. The most common action was "Seizure of Material or Personnel".
Studies find that the probability that disputes between states will be resolved peacefully is positively affected by the degree of democracy exhibited by the lesser democratic state involved in that dispute. Disputes between democratic states are significantly shorter than disputes involving at least one undemocratic state. Democratic states are more likely to be amenable to third party mediation when they are involved in disputes with each other (Ray 2003).
In international crises that include the threat or use of military force, one study finds that if the parties are democracies, then relative military strength has no effect on who wins. This is different from when nondemocracies are involved. These results are the same also if the conflicting parties are formal allies (Gelpi & Griesdorf 2001). Similarly, a study of the behavior of states that joined ongoing militarized disputes reports that power is important only to autocracies: democracies do not seem to base their alignment on the power of the sides in the dispute (Werner & Lemke 1997).
According to a 2017 review study, "there is enough evidence to conclude that democracy does cause peace at least between democracies, that the observed correlation between democracy and peace is not spurious" (Reiter 2017).
Most studies have looked only at who is involved in the conflicts and ignored the question of who initiated the conflict. In many conflicts both sides argue that the other side was initiator. Several researchers (as described in Gleditsch, Christiansen & Hegre 2004) have argued that studying conflict initiation is of limited value, because existing data about conflict initiation may be especially unreliable. Even so, several studies have examined this. Reiter and Stam (2003) argue that autocracies initiate conflicts against democracies more frequently than democracies do against autocracies. Quackenbush and Rudy (2006), while confirming Reiter and Stam's results, find that democracies initiate wars against nondemocracies more frequently than nondemocracies do to each other. Several following studies (Peceny & Beer 2003, Peceny & Butler 2004, Lai & Slater 2006) have studied how different types of autocracies with different institutions vary regarding conflict initiation. Personalistic and military dictatorships may be particularly prone to conflict initiation, as compared to other types of autocracy such as one party states, but also more likely to be targeted in a war having other initiators.
One 2017 study found that democracies are no less likely to settle border disputes peacefully than non-democracies (Gibler & Owsiak 2017).
Most of this article discusses research on relations between states. However, there is also evidence that democracies have less internal systematic violence. For instance, one study finds that the most democratic and the most authoritarian states have few civil wars, and intermediate regimes the most. The probability for a civil war is also increased by political change, regardless whether toward greater democracy or greater autocracy. Intermediate regimes continue to be the most prone to civil war, regardless of the time since the political change. In the long run, since intermediate regimes are less stable than autocracies, which in turn are less stable than democracies, durable democracy is the most probable end-point of the process of democratization (Hegre et al. 2001). Abadie (2004) study finds that the most democratic nations have the least terrorism. Harff (2003) finds that genocide and politicide are rare in democracies. Rummel (1997) finds that the more democratic a regime, the less its democide. He finds that democide has killed six times as many people as battles.
Davenport and Armstrong II (2004, p. 1) lists several other studies and states: "Repeatedly, democratic political systems have been found to decrease political bans, censorship, torture, disappearances and mass killing, doing so in a linear fashion across diverse measurements, methodologies, time periods, countries, and contexts." It concludes: "Across measures and methodological techniques, it is found that below a certain level, democracy has no impact on human rights violations, but above this level democracy influences repression in a negative and roughly linear manner." Davenport and Armstrong II (2003) states that thirty years worth of statistical research has revealed that only two variables decrease human rights violations: political democracy and economic development.
Abulof and Goldman add a caveat, focusing on the contemporary Middle East and North Africa (MENA). Statistically, a MENA democracy makes a country more prone to both the onset and incidence of civil war, and the more democratic a MENA state is, the more likely it is to experience violent intrastate strife. Moreover, anocracies do not seem to be predisposed to civil war, either worldwide or in MENA. Looking for causality beyond correlation, they suggest that democracy's pacifying effect is partly mediated through societal subscription to self-determination and popular sovereignty. This may turn "democratizing nationalism" to a long-term prerequisite, not just an immediate hindrance, to peace and democracy (Abulof & Goldman 2015).
These theories have traditionally been categorized into two groups: explanations that focus on democratic norms and explanations that focus on democratic political structures (Gelpi & Griesdorf 2001, Braumoeller 1997). Note that they usually are meant to be explanations for little violence between democracies, not for a low level of internal violence in democracies.
Several of these mechanisms may also apply to countries of similar systems. The book Never at War finds evidence for an oligarchic peace. One example is the Polish-Lithuanian Commonwealth, in which the Sejm resisted and vetoed most royal proposals for war (For a description, see Frost 2000, esp. pp. 9-11, 114, 181, 323), like those of W?adys?aw IV Vasa.
One example from the first group is that liberal democratic culture may make the leaders accustomed to negotiation and compromise (Weart 1998, Müller & Wolff 2004). Another that a belief in human rights may make people in democracies reluctant to go to war, especially against other democracies. The decline in colonialism, also by democracies, may be related to a change in perception of non-European peoples and their rights (Ravlo & Gleditsch 2000).
Bruce Russett (1993, pp. 5-11, 35, 59-62, 73-4) also argues that the democratic culture affects the way leaders resolve conflicts. In addition, he holds that a social norm emerged toward the end of the nineteenth century; that democracies should not fight each other, which strengthened when the democratic culture and the degree of democracy increased, for example by widening the franchise. Increasing democratic stability allowed partners in foreign affairs to perceive a nation as reliably democratic. The alliances between democracies during the two World Wars and the Cold War also strengthened the norms. He sees less effective traces of this norm in Greek antiquity.
Hans Köchler (1995) relates the question of transnational democracy to empowering the individual citizen by involving him, through procedures of direct democracy, in a country's international affairs, and he calls for the restructuring of the United Nations Organization according to democratic norms. He refers in particular to the Swiss practice of participatory democracy.
Mousseau (2000, 2005) argues that it is market-oriented development that creates the norms and values that explain both democracy and the peace. In less developed countries individuals often depend on social networks that impose conformity to in-group norms and beliefs, and loyalty to group leaders. When jobs are plentiful on the market, in contrast, as in market-oriented developed countries, individuals depend on a strong state that enforces contracts equally. Cognitive routines emerge of abiding by state law rather than group leaders, and, as in contracts, tolerating differences among individuals. Voters in marketplace democracies thus accept only impartial 'liberal' governments, and constrain leaders to pursue their interests in securing equal access to global markets and in resisting those who distort such access with force. Marketplace democracies thus share common foreign policy interests in the supremacy--and predictability--of international law over brute power politics, and equal and open global trade over closed trade and imperial preferences. When disputes do originate between marketplace democracies, they are less likely than others to escalate to violence because both states, even the stronger one, perceive greater long-term interests in the supremacy of law over power politics.
Braumoeller (1997) argues that liberal norms of conflict resolution vary because liberalism takes many forms. By examining survey results from the newly independent states of the former Soviet Union, the author demonstrates that liberalism in that region bears a stronger resemblance to 19th-century liberal nationalism than to the sort of universalist, Wilsonian liberalism described by democratic peace theorists, and that, as a result, liberals in the region are more, not less, aggressive than non-liberals.
The case for institutional constraints goes back to Kant (1795), who wrote:
[I]f the consent of the citizens is required in order to decide that war should be declared (and in this constitution it cannot but be the case), nothing is more natural than that they would be very cautious in commencing such a poor game, decreeing for themselves all the calamities of war. Among the latter would be: having to fight, having to pay the costs of war from their own resources, having painfully to repair the devastation war leaves behind, and, to fill up the measure of evils, load themselves with a heavy national debt that would embitter peace itself and that can never be liquidated on account of constant wars in the future.
Democracy thus gives influence to those most likely to be killed or wounded in wars, and their relatives and friends (and to those who pay the bulk of the war taxes; Russett 1993, p. 30). This monadic theory must, however, explain why democracies do attack non-democratic states. One explanation is that these democracies were threatened or otherwise were provoked by the non-democratic states. Doyle (1997, p. 272) argued that the absence of a monadic peace is only to be expected: the same ideologies that cause liberal states to be at peace with each other inspire idealistic wars with the illiberal, whether to defend oppressed foreign minorities or avenge countrymen settled abroad. Doyle (1997, p. 292) also notes liberal states do conduct covert operations against each other; the covert nature of the operation, however, prevents the publicity otherwise characteristic of a free state from applying to the question
Studies show that democratic states are more likely than autocratic states to win the wars. One explanation is that democracies, for internal political and economic reasons, have greater resources. This might mean that democratic leaders are unlikely to select other democratic states as targets because they perceive them to be particularly formidable opponents. One study finds that interstate wars have important impacts on the fate of political regimes, and that the probability that a political leader will fall from power in the wake of a lost war is particularly high in democratic states (Ray 1998).
As described in (Gelpi & Griesdorf 2001), several studies have argued that liberal leaders face institutionalized constraints that impede their capacity to mobilize the state's resources for war without the consent of a broad spectrum of interests. Survey results that compare the attitudes of citizens and elites in the Soviet successor states are consistent with this argument (Braumoeller 1997). Moreover, these constraints are readily apparent to other states and cannot be manipulated by leaders. Thus, democracies send credible signals to other states of an aversion to using force. These signals allow democratic states to avoid conflicts with one another, but they may attract aggression from nondemocratic states. Democracies may be pressured to respond to such aggression--perhaps even preemptively--through the use of force. Also as described in (Gelpi & Griesdorf 2001), studies have argued that when democratic leaders do choose to escalate international crises, their threats are taken as highly credible, since there must be a relatively large public opinion for these actions. In disputes between liberal states, the credibility of their bargaining signals allows them to negotiate a peaceful settlement before mobilization.
An explanation based on game theory similar to the last two above is that the participation of the public and the open debate send clear and reliable information regarding the intentions of democracies to other states. In contrast, it is difficult to know the intentions of nondemocratic leaders, what effect concessions will have, and if promises will be kept. Thus there will be mistrust and unwillingness to make concessions if at least one of the parties in a dispute is a nondemocracy (Levy & Razin 2004).
The risk factors for certain types of state have, however, changed since Kant's time. In the quote above, Kant points to the lack of popular support for war - first that the populace will directly or indirectly suffer in the event of war - as a reason why republics will not tend to go to war. The number of American troops killed or maimed versus the number of Iraqi soldiers and civilians maimed and killed in the American-Iraqi conflict is indicative. This may explain the relatively great willingness of democratic states to attack weak opponents: the Iraq war was, initially at least, highly popular in the United States. The case of the Vietnam War might, nonetheless, indicate a tipping point where publics may no longer accept continuing attrition of their soldiers (even while remaining relatively indifferent to the much higher loss of life on the part of the populations attacked).
Coleman (2002) uses economic cost-benefit analysis to reach conclusions similar to Kant's. Coleman examines the polar cases of autocracy and liberal democracy. In both cases, the costs of war are assumed to be borne by the people. In autocracy, the autocrat receives the entire benefits of war, while in a liberal democracy the benefits are dispersed among the people. Since the net benefit to an autocrat exceeds the net benefit to a citizen of a liberal democracy, the autocrat is more likely to go to war. The disparity of benefits and costs can be so high that an autocrat can launch a welfare-destroying war when his net benefit exceeds the total cost of war. Contrarily, the net benefit of the same war to an individual in a liberal democracy can be negative so that he would not choose to go to war. This disincentive to war is increased between liberal democracies through their establishment of linkages, political and economic, that further raise the costs of war between them. Therefore, liberal democracies are less likely to go war, especially against each other. Coleman further distinguishes between offensive and defensive wars and finds that liberal democracies are less likely to fight defensive wars that may have already begun due to excessive discounting of future costs.
There are several logically distinguishable classes of criticism (Pugh 2005). Note that they usually apply to no wars or few MIDs between democracies, not to little systematic violence in established democracies. (But see List of wars between democracies.)
One study (Schwartz & Skinner 2002) has argued that there have been as many wars between democracies as one would expect between any other couple of states. However, its authors include wars between young and dubious democracies, and very small wars.
Others (Spiro 1994, Gowa 1999, Small & Singer 1976) state that, although there may be some evidence for democratic peace, the data sample or the time span may be too small to assess any definitive conclusions. For example, Gowa finds evidence for democratic peace to be insignificant before 1939, because of the too small number of democracies, and offers an alternate explanation for the following period (see the section on Realist Explanations). Gowa's use of statistics has been criticized, with several other studies and reviews finding different or opposing results (Gelpi & Griesdorf 2001, Ray 2003). However, this can be seen as the longest-lasting criticism to the theory; as noted earlier, also some supporters (Wayman 1998) agree that the statistical sample for assessing its validity is limited or scarce, at least if only full-scale wars are considered.
According to one study (Ray 2003), which uses a rather restrictive definition of democracy and war, there were no wars between jointly democratic couples of states in the period from 1816 to 1992. Assuming a purely random distribution of wars between states, regardless of their democratic character, the predicted number of conflicts between democracies would be around ten. So, Ray argues that the evidence is statistically significant, but that it is still conceivable that, in the future, even a small number of inter-democratic wars would cancel out such evidence.[e]
Douglas M. Gibler and Andrew Owsiak in their study argued peace almost always comes before democracy and that states do not develop democracy until all border disputes have been settled. These studies indicate that there is strong evidence that peace causes democracy but little evidence that democracy causes peace (Gibler & Owsiak 2017). Azar Gat (2017) argues that it is not democracy in itself that leads to peace but other aspects of modernization, such as economic prosperity and lower population growth.
The hypothesis that peace causes democracy is supported by psychological and cultural theories. Christian Welzel's human empowerment theory posits that existential security leads to emancipative cultural values and support for a democratic political organization (Welzel 2013). This is in agreement with theories based on evolutionary psychology.
Several studies fail to confirm that democracies are less likely to wage war than autocracies if wars against non-democracies are included (Cashman 2013, Chapt. 5).
Some authors criticize the definition of democracy by arguing that states continually reinterpret other states' regime types as a consequence of their own objective interests and motives, such as economic and security concerns (Rosato 2003). For example, one study (Oren 1995) reports that Germany was considered a democratic state by Western opinion leaders at the end of the 19th century; yet in the years preceding World War I, when its relations with the United States, France and Britain started deteriorating, Germany was gradually reinterpreted as an autocratic state, in absence of any actual regime change. Shimmin (1999) moves a similar criticism regarding the western perception of Milosevic's Serbia between 1989 and 1999. Rummel (1999) replies to this criticism by stating that, in general, studies on democratic peace do not focus on other countries' perceptions of democracy; and in the specific case of Serbia, by arguing that the limited credit accorded by western democracies to Milosevic in the early '90s did not amount to a recognition of democracy, but only to the perception that possible alternative leaders could be even worse.
Some democratic peace researchers have been criticized for post hoc reclassifying some specific conflicts as non-wars or political systems as non-democracies without checking and correcting the whole data set used similarly. Supporters and opponents of the democratic peace agree that this is bad use of statistics, even if a plausible case can be made for the correction (Bremer 1992, Gleditsch 1995, Gowa 1999). A military affairs columnist of the newspaper Asia Times has summarized the above criticism in a journalist's fashion describing the theory as subject to the no true Scotsman problem: exceptions are explained away as not being between "real" democracies or "real" wars (Asia Times 2006).
Some democratic peace researchers require that the executive result from a substantively contested election. This may be a restrictive definition: For example, the National Archives of the United States notes that "For all intents and purposes, George Washington was unopposed for election as President, both in 1789 and 1792". (Under the original provisions for the Electoral College, there was no distinction between votes for President and Vice-President: each elector was required to vote for two distinct candidates, with the runner-up to be Vice-President. Every elector cast one of his votes for Washington (National Archives and Records Administration n.d.), John Adams received a majority of the other votes; there were several other candidates: so the election for Vice President was contested.)
Sometimes the datasets used have also been criticized. For example, some authors have criticized the Correlates of War data for not including civilian deaths in the battle deaths count, especially in civil wars (Sambanis 2001). Cohen and Weeks (2006) argue that most fishing disputes, which include no deaths and generally very limited threats of violence, should be excluded even from the list of military disputes. Gleditsch (2004) made several criticisms to the Correlates of War data set, and produced a revised set of data. Maoz (1993) made several criticisms to the Polity I and II data sets, which have mostly been addressed in later versions. 
The most comprehensive critique points out that "democracy" is rarely defined, never refers to substantive democracy, is unclear about causation, has been refuted in more than 100 studies, fails to account for some 200 deviant cases, and has been promoted ideologically to justify one country seeking to expand democracy abroad (Haas 2014). Most studies treat the complex concept of "democracy" is a bivariate variable rather than attempting to dimensionalize the concept. Studies also fail to take into account the fact that there are dozens of types of democracy, so the results are meaningless unless articulated to a particular type of democracy or claimed to be true for all types, such as consociational or economic democracy, with disparate datasets.
Recent work into the democratic norms explanations shows that the microfoundations on which this explanation rest do not find empirical support. Within most earlier studies, the presence of liberal norms in democratic societies and their subsequent influence on the willingness to wage war was merely assumed, never measured. Moreover, it was never investigated whether or not these norms are absent within other regime-types. Two recent studies measured the presence of liberal norms and investigated the assumed effect of these norms on the willingness to wage war. The results of both studies show that liberal democratic norms are not only present within liberal democracies, but also within other regime-types. Moreover, these norms show are not of influence on the willingness to attack another state during an interstate conflict at the brink of war (Bakker 2017, 2018).
The peacefulness may have various limitations and qualifiers and may not actually mean very much in the real world.
Democratic peace researchers do in general not count as wars conflicts which do not kill a thousand on the battlefield; thus they exclude for example the bloodless Cod Wars. However, as noted earlier, research has also found a peacefulness between democracies when looking at lesser conflicts.
Liberal democracies have less of these wars than other states after 1945. This might be related to changes in the perception of non-European peoples, as embodied in the Universal Declaration of Human Rights (Ravlo & Gleditsch 2000).
Related to this is the human rights violations committed against native people, sometimes by liberal democracies. One response is that many of the worst crimes were committed by nondemocracies, like in the European colonies before the nineteenth century, in King Leopold II of Belgium's privately owned Congo Free State, and in Joseph Stalin's Soviet Union. The United Kingdom abolished slavery in British territory in 1833, immediately after the Reform Act 1832 had significantly enlarged the franchise. (Of course, the abolition of the slave trade had been enacted in 1807; and many DPT supporters would deny that the UK was a liberal democracy in 1833 when examining interstate wars.)
Hermann and Kegley, Jr. (1995) argue that interventions between democracies are more likely to happen than projected by an expected model. They further argue (Hermann & Kegley, Jr. 1996) that democracies are more likely to intervene in other liberal states than against countries that are non-democracies. Finally, they argue that these interventions between democracies have been increasing over time and that the world can expect more of these interventions in the future (Hermann & Kegley, Jr. 1995, 1996, 1997). The methodology used has been criticized and more recent studies have found opposing results (Gleditsch, Christiansen & Hegre 2004).
Rummel argues that the continuing increase in democracy worldwide will soon lead to an end to wars and democide, possibly around or even before the middle of this century (Democratic Peace Clock n.d.). The fall of Communism and the increase in the number of democratic states were accompanied by a sudden and dramatic decline in total warfare, interstate wars, ethnic wars, revolutionary wars, and the number of refugees and displaced persons (Center for Systemic Peace 2006). One report claims that the two main causes of this decline in warfare are the end of the Cold War itself and decolonization; but also claims that the three Kantian factors have contributed materially (Human Security Report 2005).
Economic historians Joel Mokyr and Hans-Joachim Voth argue that democratic states may have been more vulnerable to conquest because the rulers in those states were too heavily constrained. Absolutist rulers in other states could however operate more effectively (Mokyr & Voth 2010, pp. 25-26).
Democratic peace theory is a well established research field with more than a hundred authors having published articles about it (Rummel n.d.). Several peer-reviewed studies mention in their introduction that most researchers accept the theory as an empirical fact (For example, Kinsella 2005, Owen 2004, Levy & Razin 2004, Mousseau & Shi 2016, Gelpi & Griesdorf 2002).
Imre Lakatos suggested that what he called a "progressive research program" is better than a "degenerative" one when it can explain the same phenomena as the "degenerative" one, but is also characterized by growth of its research field and the discovery of important novel facts. In contrast, the supporters of the "degenerative" program do not make important new empirical discoveries, but instead mostly apply adjustments to their theory in order to defend it from competitors. Some researchers argue that democratic peace theory is now the "progressive" program in international relations. According to these authors, the theory can explain the empirical phenomena previously explained by the earlier dominant research program, realism in international relations; in addition, the initial statement that democracies do not, or rarely, wage war on one another, has been followed by a rapidly growing literature on novel empirical regularities (Ray 2003, Chernoff 2004, Harrison 2010).
Other examples are several studies finding that democracies are more likely to ally with one another than with other states, forming alliances which are likely to last longer than alliances involving nondemocracies (Ray 2003); several studies (including Weart 1998) showing that democracies conduct diplomacy differently and in a more conciliatory way compared to nondemocracies; one study finding that democracies with proportional representation are in general more peaceful regardless of the nature of the other party involved in a relationship (Leblang & Chan 2003); and another study reporting that proportional representation system and decentralized territorial autonomy is positively associated with lasting peace in postconflict societies (Binningsbø 2005).
Many democracies become non-democratic by war, as being aggressed or as aggressor (quickly after a coup), sometimes the coup leader worked to provoke that war.
Carl Schmitt (1985 , Chapt. 1) wrote on how to overrule a Constitution: "Sovereign is he who decides on the exception." Schmitt (2008 , p. 46) again on the need for internal (and foreign) enemies because they are useful to persuade the people not to trust anyone more than the Leader: "As long as the state is a political entity this requirement for internal peace compels it in critical situations to decide also upon the domestic enemy. Every state provides, therefore, some kind of formula for the declaration of an internal enemy." Whatever opposition will be pictured and intended as the actual foreign enemy's puppet.
One general criticism motivating research of different explanations is that actually the theory cannot claim that "democracy causes peace", because the evidence for democracies being, in general, more peaceful is very slight or non existent; it only can support the claim that "joint democracy causes peace". According to Rosato (2003), this casts doubts on whether democracy is actually the cause because, if so, a monadic effect would be expected.
Perhaps the simplest explanation to such perceived anomaly (but not the one the Realist Rosato prefers, see the section on Realist explanations below) is that democracies are not peaceful to each other because they are democratic, but rather because they are similar. This line of thought started with several independent observations of an "Autocratic Peace" effect, a reduced probability of war (obviously no author claims its absence) between states which are both non-democratic, or both highly so (Raknerud & Hegre 1997, Beck & Jackman 1998). This has led to the hypothesis that democratic peace emerges as a particular case when analyzing a subset of states which are, in fact, similar (Werner 2000). Or, that similarity in general does not solely affect the probability of war, but only coherence of strong political regimes such as full democracies and stark autocracies.
Autocratic peace and the explanation based on political similarity is a relatively recent development, and opinions about its value are varied. Henderson (2002) builds a model considering political similarity, geographic distance and economic interdependence as its main variables, and concludes that democratic peace is a statistical artifact which disappears when the above variables are taken into account. Werner (2000) finds a conflict reducing effect from political similarity in general, but with democratic dyads being particularly peaceful, and noting some differences in behavior between democratic and autocratic dyads with respect to alliances and power evaluation. Beck, King, and Zeng (2004) use neural networks to show two distinct low probability zones, corresponding to high democracy and high autocracy.[f] Petersen (2004) uses a different statistical model and finds that autocratic peace is not statistically significant, and that the effect attributed to similarity is mostly driven by the pacifying effect of joint democracy. Ray (2005) similarly disputes the weight of the argument on logical grounds, claiming that statistical analysis on "political similarity" uses a main variable which is an extension of "joint democracy" by linguistic redefinition, and so it is expected that the war reducing effects are carried on in the new analysis. Bennett (2006) builds a direct statistical model based on a triadic classification of states into "democratic", "autocratic" and "mixed". He finds that autocratic dyads have a 35% reduced chance of going into any type of armed conflict with respect to a reference mixed dyad. Democratic dyads have a 55% reduced chance. This effect gets stronger when looking at more severe conflicts; for wars (more than 1000 battle deaths), he estimates democratic dyads to have an 82% lower risk than autocratic dyads. He concludes that autocratic peace exists, but democratic peace is clearly stronger. However, he finds no relevant pacifying effect of political similarity, except at the extremes of the scale.
To summarize a rather complex picture, there are no less than four possible stances on the value of this criticism:
The capitalist peace, or capitalist peace theory, posits that according to a given criteria for economic development (capitalism), developed economies have not engaged in war with each other, and rarely enter into low-level disputes. These theories have been proposed as an explanation for the democratic peace by accounting for both democracy and the peace among democratic nations. The exact nature of the causality depends upon both the proposed variable and the measure of the indicator for the concept used.
A majority of researchers on the determinants of democracy agree that economic development is a primary factor which allows the formation of a stable and healthy democracy (Hegre 2003, Weede 2004). Thus, some researchers have argued that economic development also plays a factor in the establishment of peace.
Mousseau (2000, 2002, 2003, 2005) argues that a culture of contracting in advanced market-oriented economies may cause both democracy and peace. These studies indicate that democracy, alone, is an unlikely cause of the democratic peace. A low level of market-oriented economic development may hinder development of liberal institutions and values. Hegre (2000) and Souva (2004) confirmed these expectations. Mousseau (2005) finds that democracy is a significant factor only when both democracies have levels of economic development well above the global median. In fact, the poorest 21% of the democracies studied, and the poorest 4-5% of current democracies, are significantly more likely than other kinds of countries to fight each other. Mousseau, Hegre, and Oneal (2003) confirm that if at least one of the democracies involved has a very low level of economic development, democracy is ineffective in preventing war; however, they find that when also controlling for trade, 91% of all the democratic pairs had high enough development for the pacifying effect of democracy to be important during the 1885-1992 period and all in 1992. The difference in results of Mousseau (2005) and Mousseau, Hegre, and Oneal (2003) may be due to sampling: Mousseau (2005) observed only neighboring states where poor countries actually can fight each other. In fact, fully 89% of militarized conflicts between less developed countries from 1920 and 2000 were among directly contiguous neighbors (Mousseau 2005, pp. 68-69). He argues that it is not likely that the results can be explained by trade: Because developed states have large economies, they do not have high levels of trade interdependence (Mousseau 2005, p. 70 and footnote 5; Mousseau, Hegre & Oneal 2003, p. 283). In fact, the correlation of developed democracy with trade interdependence is a scant 0.06 (Pearson's r - considered substantively no correlation by statisticians; Mousseau 2005, p. 77).
Both World Wars were fought between countries which can be considered economically developed. Mousseau argues that both Germany and Japan - like the USSR during the Cold War and Saudi Arabia today - had state-managed economies and thus lacked his market norms (Mousseau 2002-2003, p. 29). Hegre (2003) finds that democracy is correlated with civil peace only for developed countries, and for countries with high levels of literacy. Conversely, the risk of civil war decreases with development only for democratic countries.
Gartzke (2005) argues that economic freedom (a quite different concept from Mousseau's market norms) or financial dependence (Gartzke 2007) explains the developed democratic peace, and these countries may be weak on these dimensions too (Gwartney, Lawson & Gartzke 2005). Rummel (2005) criticizes Gartzke's methodology and argues that his results are invalid.
Several studies find that democracy, more trade causing greater economic interdependence, and membership in more intergovernmental organizations reduce the risk of war. This is often called the Kantian peace theory since it is similar to Kant's earlier theory about a perpetual peace; it is often also called "liberal peace" theory, especially when one focuses on the effects of trade and democracy. (The theory that free trade can cause peace is quite old and referred to as Cobdenism.) Many researchers agree that these variables positively affect each other but each has a separate pacifying effect. For example, in countries exchanging a substantial amount of trade, economic interest groups may exist that oppose a reciprocal disruptive war, but in democracy such groups may have more power, and the political leaders be more likely to accept their requests (Russett & Oneal 2001, Lagazio & Russett 2004, Oneal & Russett 2004). Weede (2004) argues that the pacifying effect of free trade and economic interdependence may be more important than that of democracy, because the former affects peace both directly and indirectly, by producing economic development and ultimately, democracy. Weede also lists some other authors supporting this view. However, some recent studies find no effect from trade but only from democracy (Goenner 2004, Kim & Rousseau 2005).
None of the authors listed argues that free trade alone causes peace. Even so, the issue of whether free trade or democracy is more important in maintaining peace may have potentially significant practical consequences, for example on evaluating the effectiveness of applying economic sanctions and restrictions to autocratic countries.
It was Michael Doyle (1983, 1997) who reintroduced Kant's three articles into democratic peace theory. He argued that a pacific union of liberal states has been growing for the past two centuries. He denies that a pair of states will be peaceful simply because they are both liberal democracies; if that were enough, liberal states would not be aggressive towards weak non-liberal states (as the history of American relations with Mexico shows they are). Rather, liberal democracy is a necessary condition for international organization and hospitality (which are Kant's other two articles)--and all three are sufficient to produce peace. Other Kantians have not repeated Doyle's argument that all three in the triad must be present, instead stating that all three reduce the risk of war.
Immanuel Wallerstein has argued that it is the global capitalist system that creates shared interests among the dominant parties, thus inhibiting potentially harmful belligerence (Satana 2010, p. 231).
Toni Negri and Michael Hardt take a similar stance, arguing that the intertwined network of interests in the global capitalism leads to the decline of individual nation states, and the rise of a global Empire which has no outside, and no external enemies. As a result, they write, "The era of imperialist, interimperialist, and anti-imperialist wars is over. (...) we have entered the era of minor and internal conflicts. Every imperial war is a civil war, a police action" (Hardt & Negri 2000).
Many studies (as those discussed in Ray 1998, Ray 2005, Oneal & Russett 2004) supporting the theory have controlled for many possible alternative causes of the peace. Examples of factors controlled for are geographic distance, geographic contiguity, power status, alliance ties, militarization, economic wealth and economic growth, power ratio, and political stability. These studies have often found very different results depending on methodology and included variables, which has caused criticism. DPT does not state democracy is the only thing affecting the risk of military conflict. Many of the mentioned studies have found that other factors are also important.
Several studies have also controlled for the possibility of reverse causality from peace to democracy. For example, one study (Reuveny & Li 2003) supports the theory of simultaneous causation, finding that dyads involved in wars are likely to experience a decrease in joint democracy, which in turn increases the probability of further war. So they argue that disputes between democratizing or democratic states should be resolved externally at a very early stage, in order to stabilize the system. Another study (Reiter 2001) finds that peace does not spread democracy, but spreading democracy is likely to spread peace. A different kind of reverse causation lies in the suggestion that impending war could destroy or decrease democracy, because the preparation for war might include political restrictions, which may be the cause for the findings of democratic peace. However, this hypothesis has been statistically tested in a study (Mousseau & Shi 1999) whose authors find, depending on the definition of the pre-war period, no such effect or a very slight one. So, they find this explanation unlikely. Note also that this explanation would predict a monadic effect, although weaker than the dyadic one[dubious ].
Weart (1998) argues that the peacefulness appears and disappears rapidly when democracy appears and disappears. This in his view makes it unlikely that variables that change more slowly are the explanation. Weart, however, has been criticized for not offering any quantitative analysis supporting his claims (Ray 2000).
Wars tend very strongly to be between neighboring states. Gleditsch (1995) showed that the average distance between democracies is about 8000 miles, the same as the average distance between all states. He believes that the effect of distance in preventing war, modified by the democratic peace, explains the incidence of war as fully as it can be explained.
Supporters of realism in international relations in general argue that not democracy or its absence, but considerations and evaluations of power, cause peace or war. Specifically, many realist critics claim that the effect ascribed to democratic, or liberal, peace, is in fact due to alliance ties between democratic states which in turn are caused, one way or another, by realist factors.
For example, Farber and Gowa (1995) find evidence for peace between democracies to be statistically significant only in the period from 1945 on, and consider such peace an artifact of the Cold War, when the threat from the communist states forced democracies to ally with one another. Mearsheimer (1990) offers a similar analysis of the Anglo-American peace before 1945, caused by the German threat. Spiro (1994) finds several instances of wars between democracies, arguing that evidence in favor of the theory might be not so vast as other authors report, and claims that the remaining evidence consists of peace between allied states with shared objectives. He acknowledges that democratic states might have a somewhat greater tendency to ally with one another, and regards this as the only real effect of democratic peace. Rosato (2003) argues that most of the significant evidence for democratic peace has been observed after World War II; and that it has happened within a broad alliance, which can be identified with NATO and its satellite nations, imposed and maintained by American dominance (see Pax Americana). One of the main points in Rosato's argument is that, although never engaged in open war with another liberal democracy during the Cold War, the United States intervened openly or covertly in the political affairs of democratic states several times, for example in the Chilean coup of 1973, the 1953 coup in Iran and 1954 coup in Guatemala; in Rosato's view, these interventions show the United States' determination to maintain an "imperial peace".
The most direct counter arguments to such criticisms have been studies finding peace between democracies to be significant even when controlling for "common interests" as reflected in alliance ties (Gelpi & Griesdorf 2001, Ray 2003). Regarding specific issues, Ray (1998) objects that explanations based on the Cold War should predict that the Communist bloc would be at peace within itself also, but exceptions include the Soviet Invasion of Afghanistan, the Cambodian-Vietnamese War, and the Sino-Vietnamese War. Ray also argues that the external threat did not prevent conflicts in the Western bloc when at least one of the involved states was a nondemocracy, such as the Turkish Invasion of Cyprus (against Greek Junta supported Cypriot Greeks), the Falklands War, and the Football War. Also, one study (Ravlo & Gleditsch 2000, p. 2) notes that the explanation "goes increasingly stale as the post-Cold War world accumulates an increasing number of peaceful dyad-years between democracies". Rosato's argument about American dominance has also been criticized for not giving supporting statistical evidence (Slantchev, Alexandrova & Gartzke 2005).
Some realist authors also criticize in detail the explanations first by supporters of democratic peace, pointing to supposed inconsistencies or weaknesses.
Rosato (2003) criticizes most explanations to how democracy might cause peace. Arguments based on normative constraints, he argues, are not consistent with the fact that democracies do go to war no less than other states, thus violating norms preventing war; for the same reason he refutes arguments based on the importance of public opinion. Regarding explanations based on greater accountability of leaders, he finds that historically autocratic leaders have been removed or punished more often than democratic leaders when they get involved in costly wars. Finally, he also criticizes the arguments that democracies treat each other with trust and respect even during crises; and that democracy might be slow to mobilize its composite and diverse groups and opinions, hindering the start of a war, drawing support from other authors. Another realist, Layne (1994), analyzes the crises and brinkmanship that took place between non-allied democratic great powers, during the relatively brief period when such existed. He finds no evidence either of institutional or cultural constraints against war; indeed, there was popular sentiment in favor of war on both sides. Instead, in all cases, one side concluded that it could not afford to risk that war at that time, and made the necessary concessions.
Rosato's objections have been criticized for claimed logical and methodological errors, and for being contradicted by existing statistical research (Kinsella 2005). Russett (1995) replies to Layne by re-examining some of the crises studied in his article, and reaching different conclusions; Russett argues that perceptions of democracy prevented escalation, or played a major role in doing so. Also, a recent study (Gelpi & Griesdorf 2001) finds that, while in general the outcome of international disputes is highly influenced by the contenders' relative military strength, this is not true if both contenders are democratic states; in this case the authors find the outcome of the crisis to be independent of the military capabilities of contenders, which is contrary to realist expectations. Finally, both the realist criticisms here described ignore new possible explanations, like the game-theoretic one discussed below (Risse n.d.).
A different kind of realist criticism (see Jervis 2002 for a discussion) stresses the role of nuclear weapons in maintaining peace. In realist terms, this means that, in the case of disputes between nuclear powers, respective evaluation of power might be irrelevant because of Mutual assured destruction preventing both sides from foreseeing what could be reasonably called a "victory". The 1999 Kargil War between India and Pakistan has been cited as a counterexample to this argument (Page Fortna 2004), though this was a small, regional conflict and the threat of WMDs being used contributed to its de-escalation (The Nation 2006).
Some supporters of the democratic peace do not deny that realist factors are also important (Russett 1995). Research supporting the theory has also shown that factors such as alliance ties and major power status influence interstate conflict behavior (Ray 2003).
In the United States, presidents from both major parties have expressed support for the theory. In his 1994 State of the Union address, then-President Bill Clinton, a member of the Democratic Party, said: "Ultimately, the best strategy to ensure our security and to build a durable peace is to support the advance of democracy elsewhere. Democracies don't attack each other" (Clinton 2000). In a 2004 press conference, then-President George W. Bush, a member of the Republican Party, said: "And the reason why I'm so strong on democracy is democracies don't go to war with each other. And the reason why is the people of most societies don't like war, and they understand what war means.... I've got great faith in democracies to promote peace. And that's why I'm such a strong believer that the way forward in the Middle East, the broader Middle East, is to promote democracy." (Office of the Press Secretary 2004)[g]
In a 1999 speech, Chris Patten, the then-European Commissioner for External Relations, said: "Inevitable because the EU was formed partly to protect liberal values, so it is hardly surprising that we should think it appropriate to speak out. But it is also sensible for strategic reasons. Free societies tend not to fight one another or to be bad neighbours" (Patten 1999). The A Secure Europe in a Better World, European Security Strategy states: "The best protection for our security is a world of well-governed democratic states." (EEAS Strategic Planning 2003) Tony Blair has also claimed the theory is correct (The Daily Show with Jon Stewart 2008).
Some fear that the democratic peace theory may be used to justify wars against nondemocracies in order to bring lasting peace, in a democratic crusade (Chan 1997, p. 59). Woodrow Wilson in 1917 asked Congress to declare war against Imperial Germany, citing Germany's sinking of American ships due to unrestricted submarine warfare and the Zimmermann telegram, but also stating that "A steadfast concert for peace can never be maintained except by a partnership of democratic nations" and "The world must be made safe for democracy." (Wilson 1917)[h]R. J. Rummel is a notable proponent of war for the purpose of spreading democracy, based on this theory.
Some point out that the democratic peace theory has been used to justify the 2003 Iraq War, others argue that this justification was used only after the war had already started (Russett 2005). Furthermore, Weede (2004) has argued that the justification is extremely weak, because forcibly democratizing a country completely surrounded by non-democracies, most of which are full autocracies, as Iraq was, is at least as likely to increase the risk of war as it is to decrease it (some studies show that dyads formed by one democracy and one autocracy are the most warlike, and several find that the risk of war is greatly increased in democratizing countries surrounded by nondemocracies). According to Weede, if the United States and its allies wanted to adopt a rationale strategy of forced democratization based on democratic peace, which he still does not recommend, it would be best to start intervening in countries which border with at least one or two stable democracies, and expand gradually. Also, research shows that attempts to create democracies by using external force has often failed. Gleditsch, Christiansen, and Hegre (2004) argue that forced democratization by interventionism may initially have partial success, but often create an unstable democratizing country, which can have dangerous consequences in the long run. Those attempts which had a permanent and stable success, like democratization in Austria, West Germany and Japan after World War II, mostly involved countries which had an advanced economic and social structure already, and implied a drastic change of the whole political culture. Supporting internal democratic movements and using diplomacy may be far more successful and less costly. Thus, the theory and related research, if they were correctly understood, may actually be an argument against a democratic crusade (Weart 1998, Owen 2005, Russett 2005).
Michael Haas has written perhaps the most trenchant critique of a hidden normative agenda (Haas 1997). Among the points raised: Due to sampling manipulation, the research creates the impression that democracies can justifiably fight nondemocracies, snuff out budding democracies, or even impose democracy. And due to sloppy definitions, there is no concern that democracies continue undemocratic practices yet remain in the sample as if pristine democracies.
According to Azar Gat's War in Human Civilization, there are several related and independent factors that contribute to democratic societies being more peaceful than other forms of governments (Gat 2006):
There is significant debate over whether the lack of any major European general wars since 1945, is due to cooperation and integration of liberal-democratic European states themselves (as in the European Union or Franco-German cooperation), an enforced peace due to intervention of the Soviet Union and the United States until 1989 and the United States alone thereafter (Mearsheimer 2010), or a combination of both (Lucarelli n.d.).