Public Opinion and The Democratic Peace

American Political Science Review Vol. 107, No.
4 November 2013
doi:10.1017/S0003055413000488
c American Political Science Association 2013
Public Opinion and the Democratic Peace

MICHAEL R. TOMZ Stanford University
JESSICA L. P. WEEKS University of Wisconsin–Madison
O
ne of the most striking findings in political science is the democratic peace: the absence of war
between democracies. Some authors attempt to explain this phenomenon by highlighting the role
of public opinion. They observe that democratic leaders are beholden to voters and argue that
voters oppose war because of its human and financial costs. This logic predicts that democracies should
behave peacefully in general, but history shows that democracies avoid war primarily in their relations
with other democracies. In this article we investigate not whether democratic publics are averse to war in
general, but whether they are especially reluctant to fight other democracies. We embedded experiments
in public opinion polls in the United States and the United Kingdom and found that individuals are
substantially less supportive of military strikes against democracies than against otherwise identical
autocracies. Moreover, our experiments suggest that shared democracy pacifies the public primarily by
changing perceptions of threat and morality, not by raising expectations of costs or failure. These findings
shed light on a debate of enduring importance to scholars and policy makers.
ew findings in political science have received as gated whether democratic publics are more reluctant
F much attention as the “democratic peace,” the

discovery that democracies almost never fight
other democracies (Doyle 1986; Russett 1993). To
to attack democracies than autocracies.1 Moreover, the
small body of existing work has not accounted for
variables that could confound the relationship between
some, the absence of military conflict among democ- shared democracy and public support for war, nor has
racies is so consistent that it approaches the status of it investigated the mechanisms by which the regime
an “empirical law” (Levy 1988). type of the adversary affects the public mood. Despite
Some authors have attempted to explain the demo- decades of research on the democratic peace, we still
cratic peace by highlighting the role of public opinion. lack convincing evidence about whether and how pub-
They observe that democratic leaders are beholden to lic opinion contributes to the absence of war among
voters and claim that voters oppose war because of democracies.
its human and financial costs. This argument, which We used experiments to shed new light on these
dates to Immanuel Kant, predicts that democracies important questions.2 Our experiments, embedded in
will behave peacefully in general—avoiding war not public opinion polls that were administered to nation-
only against democracies but also against autocracies. ally representative samples of British and American
History shows, however, that democracies frequently citizens, involved a situation in which a country was
fight autocracies. developing nuclear weapons. When describing the sit-
A different possibility is that democratic publics are uation, we randomly and independently varied four
primarily averse to war against other democracies. If potential sources of peace: the political regime, alliance
leaders are responsive to voters and voters are more status, economic ties, and military power of the adver-
reluctant to fight democracies than otherwise equiv- sary. We then asked individuals whether they would
alent autocracies, then public opinion could play an support or oppose a preventive military strike against
important role in the dyadic democratic peace. To the country’s nuclear facilities.
date, however, surprisingly few studies have investi- Participants in our experiments were substantially
less supportive of military strikes against democracies
than against otherwise identical autocracies. Moreover,
Michael R. Tomz is Professor of Political Science, Stanford
University ([email protected]).
because we randomly and independently manipulated
Jessica L. P. Weeks is Assistant Professor of Political Science, the regime type of the adversary, the observed pref-
University of Wisconsin–Madison ([email protected]). erence for peace with other democracies was almost
We are grateful for helpful feedback from Ted Brader, Alexandre certainly causal, rather than spurious. Our findings
Debs, James Fearon, Erik Gartzke, Michael Horowitz, Kosuke Imai, therefore provide empirical microfoundations for the
Robert Johns, Peter Katzenstein, Sarah Kreps, Bethany Lacina, Brad
Laveck, Ashley Leeds, Cliff Morgan, Jim Morrow, Jon Pevehouse,
1 For exceptions, see Geva, DeRouen, and Mintz (1993); Johns and
Lauren Prather, Bill Reed, Bruce Russett, Scott Sagan, Todd Sechser,
Branislav Slantchev, Paul Sniderman, Art Stein, Dustin Tingley, Davies (2012); Mintz and Geva (1993), and Rousseau (2005). See
Rob Van Houweling, Teppei Yamamoto, three anonymous review- also Lacina and Lee (2012), who examine how regime type affects
ers, and the American Political Science Review editorial board. We perceptions of threat, and Geva and Hanson (1999), who focus on
thank seminar participants at Georgetown, Michigan, Rice, Stan- sociocultural similarity.
ford, University of California, Los Angeles, University of California, 2 For other recent examples of experiments about international se-
San Diego, University of Virginia, and Yale, and conference partic- curity, see Herrmann, Tetlock, and Visser (1999); Herrmann and
ipants at Harvard, Princeton, Tel Aviv University, the International Shannon (2001); Berinsky (2007; 2009); Gartner (2008); Baum and
Studies Association, the Midwest Political Science Association, and Groeling (2009); Gelpi, Feaver, and Reifler (2009); Grieco et al.
the European Consortium for Political Research. We are grateful (2011); Horowitz and Levendusky (2011); McDermott (2011); Tin-
to Ray Duch and Iñaki Sagarzazu for invaluable help with the gley (2011); Tingley and Walter (2011); Tomz (2007); Trager and
British survey, and to the National Science Foundation for Grant Vavreck (2011); Levendusky and Horowitz (2012), and Press, Sagan,
SES-0548285. and Valentino (2013).
849
Public Opinion and the Democratic Peace November 2013
hypothesis that the preferences of ordinary voters con- a role in electoral campaigns, and that foreign policy
tribute to peace among democracies. mistakes can hurt leaders at the ballot box (Aldrich,
In addition to estimating the overall effect of democ- Sullivan, and Borgida 1989; Gelpi, Reifler, and Feaver
racy, we investigated the mechanisms through which 2007; Gronke, Koch, and Wilson 2003).
shared democracy reduces public enthusiasm for war. Second, democratic leaders face institutional con-
Democratic publics may feel reluctant to attack other straints on their powers to use force (Morgan and
democracies for a variety of reasons: They may view Campbell 1991), and public opinion affects how tightly
democracies as less threatening (Risse-Kappen 1995; those constraints bind. In many democracies, leaders
Russett 1993), regard democracies as more formidable need legislative authorization for war, but legislative
opponents (Bueno de Mesquita et al. 1999; Lake 1992; approval is less likely to materialize in the face of pub-
Reiter and Stam 2002), or have moral qualms about lic opposition (Hildebrandt et al. 2013; Lindsay 1994).
using force to overturn policies that were freely cho- Moreover, leaders must raise revenues to pay for mil-
sen by citizens in another democracy. Despite vol- itary operations, but legislative bodies are unlikely to
umes of research about the democratic peace, how- levy new taxes, incur new debt, or cut government pro-
ever, little is known about whether these factors grams to finance wars that their constituents oppose
influence the willingness of voters to attack other (Hartley and Russett 1992; Narizny 2003).
democracies. Third, leaders understand that, by remaining pop-
Using a unique experimental design and new tech- ular, they can accomplish more during their time
niques for causal mediation analysis (Imai et al. 2011; in office. In the United States, for example, pop-
Imai, Keele, and Yamamoto 2010), we find that shared ular presidents have more influence over Congress
democracy pacifies the public primarily by changing (Edwards 1997; Howell and Pevehouse 2007; Krosnick
perceptions of threat and morality, not by raising ex- and Kinder 1990). They also wield more international
pectations of costs or failure. Individuals who faced influence, because leaders who enjoy the backing of
democratic rather than autocratic countries were less the public find it easier to persuade other countries
fearful of the country’s nuclear program and harbored that their promises and threats are credible.
greater moral reservations about attacking. Those per- Consistent with these arguments, countless stud-
ceptions, in turn, made citizens more peaceful to- ies have concluded that, in decisions about using
ward democracies. By comparison, respondents did force, democratic leaders pay close attention to pub-
not think that attacking a democracy would result in lic opinion (Baum 2004; Baum and Potter 2008;
substantially higher costs or a lower likelihood of suc- Berinsky 2009; Canes-Wrone 2006; Foyle 1999; Holsti
cess than attacking an autocracy. Thus, our data help 2004; Mueller 1973; Reiter and Stam 2002; Rosenau
arbitrate between competing mechanisms, while also 1961; Russett 1990; Sobel 2001). Studying the pub-
identifying morality as an important but understudied lic can therefore tell us much about leaders’ political
source of peace among democracies. incentives.
In the remainder of the article, we first explain how How might public opinion contribute to the demo-
public opinion could play an important role in the cratic peace? The well-known Kantian argument says
democratic peace. We next revisit existing theories of that voters, who ultimately bear the human and finan-
the democratic peace and derive their implications for cial costs of war, are more war-averse than leaders, who
the preferences and perceptions of citizens. We then do not pay the direct costs of fighting (Rummel 1979).3
explain the merits of an experimental approach to test- As Kant ([1795] 1991, 100) wrote in Perpetual Peace,
ing these implications. The subsequent sections present “If . . . the consent of the citizens is required to decide
our findings about the effect of shared democracy on whether or not war is to be declared, it is very natural
public support for war and the mechanisms behind it. that they will have great hesitation in embarking on so
We conclude by discussing the implications of our find- dangerous an enterprise. For this would mean calling
ings for scholars and policy makers. down on themselves all the miseries of war. . . . But
under a constitution where the subject is not a citizen,
and which is therefore not republican, it is the simplest
PUBLIC OPINION AND THE DEMOCRATIC thing in the world to go to war.”
PEACE As others have pointed out, this argument implies
a monadic democratic peace, in which democracies
The leaders who make the ultimate decisions about are more restrained from using force overall (Rosato
war and peace in democracies have powerful incen- 2003). Historically, however, democracies have only
tives to respect the opinions of citizens. Public opinion been more peaceful in their relations with other democ-
matters for several reasons. First, leaders who disap- racies. In the remainder of this article, we investigate
point or anger their constituents risk being removed whether democratic publics distinguish between au-
from office. Although early research claimed that pub- tocratic and democratic opponents and are primarily
lic opinion on foreign policy was incoherent (Almond averse to war against democracies, thus potentially con-
1960) and that politics “stopped at the water’s edge” tributing to the dyadic democratic peace.
(Wildavsky 1966), this view has been supplanted by
numerous studies showing that mass opinion is coher- 3 Similarly, Russett (1993, 38–39) casts public opinion as a monadic
ent and influential. Leaders know that citizens care constraint on leaders, although one that can be shrugged off in situ-
about foreign policy, that foreign policy often plays ations of high threat.
850
American Political Science Review Vol. 107, No. 4
SHARED DEMOCRACY AND PUBLIC attacked by other democracies (Dixon 1994; Doyle
SUPPORT FOR WAR: 1986; Maoz and Russett 1993; Owen 1994; Risse-
CAUSAL MECHANISMS Kappen 1995; Russett 1993).
Perceptions of threat also play a crucial role in
Most theories of war presume that, before engaging “institutional” theories. Some argue that democratic
in violence, leaders and their constituents weigh the institutions reduce fear by constraining the executive,
pros and cons. Perceptions about the advantages and thereby slowing the process of mobilization and low-
disadvantages of military action are crucial in classic ering the likelihood of surprise attack (Russett 1993).
texts about war (Jervis 1976; Morgenthau 1948; Thucy- Others claim that democratic institutions contribute
dides 1954), modern game-theoretic models (Bueno to peace by conveying information about intentions
de Mesquita et al. 1999; Fearon 1995; Kydd 2005), psy- (Fearon 1994; Schultz 2001), thereby increasing the
chologica ltheories of conflict (Hermann and Kegley likelihood that inter-democratic disputes will be re-
1995; Herrmann et al. 1997), and constructivist theo- solved through peaceful bargains, rather than unneces-
ries, which argue that the costs and benefits of war are sary military conflicts. Finally, democracy could reduce
socially constructed (Finnemore 2003; Wendt 1999). fear by creating expectations of shared interests. Oneal
From this body of theory, we highlight four inputs and Russett (1999), for example, argue that democratic
into citizens’ calculations about the merits of going to institutions increase “affinity,” measured by similarity
war. First, individuals form perceptions of how threat- of voting patterns in the UN. If democracies believe
ening other countries are. Individuals who feel threat- they have common interests, they may not feel threat-
ened may support an attack in the interest of self- ened by each other.
preservation (Jervis 1978; Kydd 2005). Next, voters In sum, a number of prominent theories of the demo-
could be deterred by the costs of war and the likeli- cratic peace suggest that citizens in democracies may
hood of success. All else equal, voters view the use view other democracies as less threatening than au-
of force as more attractive when they think the eco- tocracies. Testing whether this is true is crucial to es-
nomic, diplomatic, and human costs of war will be low tablishing why democracy might lead to peace. It is
and when they expect military operations to succeed also important to see how much of the effect of shared
(Gelpi, Feaver, and Reifler 2006). Finally, moral con- democracy is driven by threat perception, as opposed
siderations could influence decisions about whether to to other mechanisms.
fight (Herrmann and Shannon 2001; Price 1998; Welch
1993).
Existing theories of the democratic peace, we show Deterrence (The Costs of Fighting and
later, can be categorized according to the claims they Likelihood of Success)
make about how the regime type of the adversary af-
fects these four inputs into the war calculus. Viewing Whereas some theories of the democratic peace gener-
the leading theories of the democratic peace this way, ate predictions about threat perception, others imply
it becomes clear that they have important but largely that democracy affects two other inputs into the war
untested implications for the preferences and beliefs calculus: the costs of fighting and the likelihood of
of individuals.4 By highlighting the micro-level impli- success. For example, Lake (1992), Reiter and Stam
cations of different theories, we lay the foundation (2002), and Bueno de Mesquita et al. (1999) argue
for our experimental analysis of individual attitudes that wars against democracies are especially costly, be-
toward the use of force. cause democratic leaders are better able to mobilize
resources for war and have strong incentives to win
the wars they start. Autocrats, by contrast, are said to
Threat Perception be less forbidding opponents, either because they have
The first input into the war calculus is threat perception. difficulty mobilizing resources or because they think
Many theories of the democratic peace suggest that they can lose wars without suffering much domestic
democracies view other democracies as less threaten- punishment and therefore choose to spend fewer re-
ing than autocracies (i.e., less likely to have malicious sources on the war effort.
intent and to take military action). Following this logic, citizens may be deterred from
Perceptions of threat play a crucial role in “norma- using force against democracies because they antici-
tive” theories of the democratic peace. These theories pate high costs of war and a low probability of victory.
begin with the premise that citizens in democracies are If we found little evidence that democracy affected
normatively opposed to violence. People in democra- perceptions of cost and success, this would contradict
cies solve domestic disagreements peacefully and apply the idea that democracies are deterred from attacking
the same nonviolent norms internationally, at least in other democracies because they view them as particu-
relations with democratic states. Democracies expect larly formidable adversaries.
other democracies to externalize peaceful norms in the
same way and therefore trust that they will not be
Morality
4 For an important exception to scholars’ inattention to the
Finally, shared democracy could produce peace by rais-
individual-level implications of democratic peace theories, see ing moral concerns about using military force: Per-
Hermann and Kegley (1995). haps democracies avoid attacking other democracies
851
because they believe it would be morally wrong.5 In students—with a total of 117 respondents across the
his influential interpretation of Kant’s Perpetual Peace three groups. The investigators described a crisis in
([1795] 1991), Michael Doyle writes, “[D]omestically which one hypothetical country had invaded another,
just republics, which rest on consent, presume foreign and they randomly varied whether the invader was a
republics to be also consensual, just, and therefore stable democracy with a newly elected parliament or
deserving of accommodation” (Doyle 1983, 230). We a military dictatorship with a puppet parliament and
advance a similar hypothesis. The foreign and domestic fierce police. Respondents were then asked to express
policies of democracies reflect the will of the people. their level of approval for various policy options, in-
Knowing this, people in democracies will feel morally cluding whether to use military force to stop the in-
reluctant to overturn policies that the citizens of other vader. In each of the three samples, subjects were more
democracies have chosen freely. Coercively interfer- likely to favor using force when the invader was a dic-
ing with another democracy would, by this argument, tatorship than when it was a democracy.
count as an illegitimate assault on the freedom and self- Rousseau (2005) ran a similar experiment on 141
determination of individuals. In contrast, democratic American college students. Each student played the
publics might have fewer moral qualms about using role of chief political advisor to the president of a
force to reverse the will of a dictator who has imposed fictional, democratic country that was involved in a
foreign and domestic policies without popular consent. territorial dispute with a southern neighbor. Rousseau
If morality is an independent driver of the democratic randomly varied three features: the southern neigh-
peace, we would expect to find a moral aversion to at- bor’s political regime (democratically elected govern-
tacking fellow democracies, separate from perceptions ment versus single-party dictatorship), the balance
of threat, cost, and success. of military forces (strong versus weak), and the do-
In sum, different theories have distinct implications mestic political position of the president the stu-
for how and why shared democracy could affect public dent was advising (strong versus weak). The stu-
beliefs and preferences. These theories not only sug- dents were then asked whether they would advise
gest that public opinion should be less inclined to use the president to use military force to settle the dis-
force against a democracy than against an autocracy, pute. Participants were less likely to recommend us-
but also propose different reasons why actors would ing military force against a democracy than against a
hold these preferences. Through experiments, we not dictatorship.
only test whether democratic publics are reluctant to Finally, Johns and Davies (2012) conducted ex-
attack fellow democracies but also adjudicate among periments on nationally representative samples in
various causal mechanisms. Britain and the United States. In their scenario,
the British/American government had uncovered evi-
dence that a country was secretly developing nuclear
PAST EXPERIMENTAL APPROACHES TO weapons, which it intended to use against its neigh-
THE DEMOCRATIC PEACE bors. The study randomized three features: regime type
Survey experiments offer unique advantages for study- (democratically elected president vs. unelected dicta-
ing how the regime type of an adversary affects popular tor), whether the foreign country was predominantly
support for war and thus whether public opinion could Christian or Islamic, and how many civilian casualties
contribute to the democratic peace. By assigning key would result from air strikes against the nuclear pro-
explanatory variables (such as the political regime of duction facilities. The study found higher public sup-
the target state) randomly, we avoid problems of endo- port for air strikes against the unelected dictator than
geneity and spurious correlation. With experiments, we against the democratically elected president and for
can also guard against collinearity and omitted variable strikes against an Islamic as opposed to a Christian
bias by independently varying factors that would other- country.
wise coincide in observational data. Finally, by measur- These studies, although path breaking, are open to
ing how the target’s political regime affects perceptions several critiques. First, with the exception of Johns and
of threat, cost, success, and morality, our experiments Davies (2012), the studies were carried out on small
have distinct advantages for shedding light on causal samples, usually of university students. This limits our
pathways. ability to generalize about how the target’s regime type
To date, only a handful of studies have used survey affects preferences in society as a whole.
experiments to investigate the democratic peace. In a Second, existing studies did not control for other fac-
pioneering article, Mintz and Geva (1993) carried out a tors that could explain the correlation between shared
survey experiment on three small samples—American democracy and peace. Skeptics of the democratic peace
college students, American adults, and Israeli college have pointed out that, for much of history, democracy
has overlapped with other potential sources of peace.
This fact has raised the suspicion that the democracy–
peace correlation found in observational studies might
5 Discussions of morality are surprisingly rare in existing scholarship
be spurious. For example, democracies might have
on the democratic peace. If morality is mentioned, it is usually in shared political interests that fortuitously coincided
passing; for example, Doyle (1983, 230) and Russett (1993, 40). Even
those who have argued that democracies may follow a “logic of
with democracy, particularly during the Cold War
appropriateness” in their dealings with fellow democracies have not (Farber and Gowa 1995; Gartzke 1998). Alternatively,
elaborated a moral argument in detail (Harrison 2004). the inter-democratic peace might be due to capitalism,
852
not democracy (Gartzke 2007).6 Another possibility adults in October–December 2010, before and after
is that the absence of war among democracies is a the U.S. congressional elections.
consequence of the post–World War II distribution Participants in both studies were told the following:
of material power, particularly American hegemony “There is much concern these days about the spread of
(Rosato 2003). nuclear weapons. We are going to describe a situation
Similar concerns about spurious correlation cast the [U.K./U.S.] could face in the future. For scientific va-
doubt on existing experimental studies of the demo- lidity the situation is general, and is not about a specific
cratic peace. When respondents read that the coun- country in the news today. Some parts of the descrip-
try was a democracy, for example, they might have tion may strike you as important; other parts may seem
assumed that the country was also an ally, a ma- unimportant. After describing the situation, we will ask
jor trading partner, or a powerful adversary.7 Thus, your opinion about a policy option.” Respondents then
previous studies cannot tell us whether the effect of received a series of bullet points with details about the
democracy was due to democracy itself or to other situation. The first bullet point explained, “A country
pacifying factors that are known to coincide with is developing nuclear weapons and will have its first
democracy. nuclear bomb within six months. The country could
Third, although previous studies showed evidence then use its missiles to launch nuclear attacks against
that the regime type of the adversary affected support any country in the world.”
for the use of force, they did not investigate the mecha- U.K. respondents received information about three
nisms behind this pattern.8 Democracy could decrease factors: the country’s political regime, military al-
the willingness to attack for a variety of reasons, but liances, and military power. We randomly and inde-
past studies were not designed to parse different expla- pendently varied these factors, each of which had two
nations. levels. Thus, in half the interviews, the country had
Our research builds on previous experiments in sev- signed a military alliance with the United Kingdom,
eral important ways. First, by carrying out surveys on but in the other half the country had not. Likewise, half
larger, more representative samples, we are able to the respondents read that the country “is a democracy
quantify the effect of democracy across the population, and shows every sign that it will remain a democracy,”
as well as on the subgroup of politically interested and whereas the other half read that the country “is not a
active voters who are most likely to affect policy. Sec- democracy and shows no sign of becoming a democ-
ond, we randomly varied whether the country in our racy.” Finally, we told participants that the country’s
scenario was an ally, a major trading partner, and/or nonnuclear forces were either “as strong” or “half as
a strong military power. This allows us to distinguish strong” as Britain’s.
the effect of democracy from potential confounders, The U.S. survey was nearly identical but random-
while also estimating the roles of alliances, trade, and ized information about trade. Respondents learned
power as potentially independent sources of peace. whether the country had, or did not have, high lev-
Third, we designed our experiments to illuminate not els of trade with the United States. As in Britain,
only whether but also why shared democracy produces we also varied whether the country was a democracy
peace. and whether it had signed a military alliance with the
United States. Unlike in Britain, we held the country’s
conventional military strength constant at half the U.S.
EXPERIMENTAL DESIGN AND level, because it seemed unrealistic to portray an ad-
PROCEDURES versary that was at conventional parity with the United
We fielded two major surveys: one in the United King- States. Thus, each study involved three random fac-
dom and one in the United States. YouGov, an internet- tors, resulting in fully crossed 2 × 2 × 2 experimental
based polling firm, administered the U.K. study to 762 designs.
adults in April–May 2010, just before the British na- We concluded with several bullet points that were
tional election, and fielded the U.S. study on 1,273 identical for everyone. Respondents were told that
“the country’s motives remain unclear, but if it builds
nuclear weapons, it will have the power to blackmail
or destroy other countries.” Additionally, they learned
6 Though see Russett (2010) and Dafoe (2011).
that the country had “refused all requests to stop its
7 In the literature on experiments, this problem is called “information nuclear weapons program.” Finally, the scenario ex-
leakage.” The survey instrument used by Johns and Davies (2012) plained that “by attacking the country’s nuclear devel-
leaked additional information by telling British and American re- opment sites now,” they could “prevent the country
spondents that their government favored air strikes and was making from making any nuclear weapons.” After presenting
the case to the United Nations. By implying that leaders deemed
it wise to attack, even though the adversary was democratic, these
this information, we asked whether respondents would
phrases may have reduced the estimated effect of democracy. favor or oppose using their country’s armed forces to
8 One possible exception is Rousseau (2005), who explored moral attack the nuclear development sites. The full text is
reservations about attacking democracies. Rousseau asked whether available on the authors’ websites.
respondents would support the use of force if it could be kept secret, The U.S. study contained two additional features
asserting that only moral qualms could explain the reluctance to use
covert force against democracies. However, other mechanisms, such
that were not part of the British survey. First, the U.S.
as a reduction in threat perception when the target is a democracy, survey measured each person’s perceptions of threat,
would predict the same response. cost, success, and morality, with the goal of shedding
853
TABLE 1. The Effect of Democracy on Willingness to Strike
United Kingdom United States United States

(between) (between) (within)
Not a democracy 34.2 53.3 50.0

Democracy 20.9 41.9 38.5
Effect of democracy −13.3 −11.4 −11.5
95% C.I. (−19.6 to −6.9) (−17.0 to −5.9) (−14.7 to −8.3)
Note: The table gives the percentage of respondents who supported military strikes when the target
was a democracy and when it was not. The difference is the estimated effect of democracy. In the
United Kingdom, we obtained between-subject estimates by comparing 364 cases in which the target
was a democracy, versus 398 cases in which it was not a democracy. In the United States, we obtained
between-subject estimates by comparing 639 cases in which the target was a democracy, versus 634 in
which it was not. The United States within-subject estimates were based on 972 respondents, each of
whom assessed two scenarios, one in which the target was a democracy and another in which the target
was not a democracy. 95% confidence intervals appear in parentheses.
light on causal mechanisms.9 To gauge perceptions of not only when the country was a democracy but also
threat, we asked which of the following events had when it was an autocracy.
more than a 50% chance of happening if the United
States did not attack: The country would build nuclear EVIDENCE ABOUT THE MAIN EFFECT OF
weapons, threaten to use them against another country, DEMOCRACY
threaten to use them against the United States or a U.S.
ally, launch a nuclear attack against another country, or Table 1 summarizes the overall effect of democracy on
launch a nuclear attack against the United States or a support for military strikes. We generated between-
U.S. ally. Respondents could select as many events as subject estimates for the United Kingdom and the
they thought probable or indicate “none of the above.” United States by comparing the average responses
To assess perceptions of cost and success, we asked among people who read about a democracy to those
which, if any, of the following events would have more who read about an autocracy. We also generated
than a 50% chance of happening if the United States within-subject estimates for the United States, where
did attack: The country would respond by attacking people completed two questionnaires, by noting how
the United States or a U.S. ally, the U.S. military would each person’s preferences changed when we switched
suffer many casualties, the U.S. economy would suf- the adversary from democracy to autocracy and vice
fer, U.S. relations with other countries would suffer, versa.10
or the United States would prevent the country from As Table 1 shows, citizens in both countries were
making nuclear weapons in the short and/or the long much less willing to attack another democracy than
run. Finally, to measure perceptions of morality, we to attack an otherwise equivalent autocracy. Approx-
asked whether it would be “morally wrong for the U.S. imately 34.2% of respondents in the U.K. supported
military to attack the country’s nuclear development a military strike when the country was not a democ-
sites.” racy versus 20.9% when the country was a democracy.
The U.S. study was unique in another way: We inter- Thus, democracy reduced support for a military strike
viewed participants twice, before and after the Novem- by more than 13 percentage points, with a 95% con-
ber 2010 election. The postelection questionnaire, ad- fidence interval of −19.6 to −6.9. The baseline level
ministered after a delay of about four weeks, repeated of militarism was much higher in the United States,
the scenario from the preelection questionnaire, but where at least half the respondents wanted to strike an
switched the political regime of the target: People who autocracy. Nonetheless, democracy exerted a similarly
had previously been asked to consider a democracy large effect in the United States: The between-subject
were asked about an autocracy, and vice versa. All
other features of the adversary, including its alliance 10 When computing within-subject effects, we assumed that there was
status, trade relations, and military power, were held no carryover (Jones and Kenward 2003), meaning that the particular
constant across both waves. The U.S. study was there- treatment a respondent received during the first interview did not
fore a crossover experiment. Of the 1,273 people who influence the attitudes he or she expressed in the second interview.
completed the preelection survey, 972 (76%) com- This assumption is reasonable given the long washout period be-
pleted the postelection survey as well. For each of those tween interviews. The time between interviews ranged between 5
and 51 days, with a median delay of 27 days. Moreover, we employed
individuals, we measured perceptions and preferences a counterbalanced design: Half the respondents received the autoc-
racy scenario before the democracy scenario, whereas the other half
received the treatments in the opposite order. Carryover effects in
one direction could therefore offset carryover effects in the opposite
9 For examples of studies using observational data to parse causal direction. Finally, as Table 1 shows, the within-subject estimates were
mechanisms about the democratic peace, see Lektzian and Souva nearly identical to the between-subject estimates, further increasing
(2009), Maoz and Russett (1993), and Schultz (1999). confidence that carryover was not a problem in this study.
854
and within-subject estimates concurred that democracy and/or high trade with the United States. The effect of
reduced enthusiasm for a military strike by about 11.5 democracy was at least as large given those scenarios as
percentage points. In both countries, democracy pro- when the target was, like Iran or North Korea, neither
duced substantively large and statistically significant an ally nor a major trading partner.
effects on preferences. Third, the effect of democracy did not weaken when,
It bears emphasizing that, because of randomization, in a follow-up experiment, we located the country that
the political regime of the target in our experiment was was pursuing nuclear weapons on a different continent
independent of alliances, power, and trade. Thus, the from Iran or North Korea. For this follow-up study,
effects in Table 1 were not spurious. Our experiments we recruited 2,393 U.S. adults via an online service
revealed the independent contribution of democracy, called Amazon Mechanical Turk (MTurk) and inter-
above and beyond the effects of alliances, power, and viewed them between October 2010 and November
trade. 2011. MTurk subscribers are younger, more likely to
Democracy reduced support for strikes not only on be female, and more liberal than the national popula-
average but also for each combination of alliances, tion. Nevertheless, Berinsky, Huber, and Lenz (2012)
power, and trade. However, one must tread carefully show that experiments on MTurk produce roughly the
here, because subdividing the data in this way results same treatment effects as experiments on nationally
in small cell sizes. In the United Kingdom, for example, representative samples.
we had 762 observations in total, implying fewer than Some participants in our MTurk experiment re-
100 cases of democracy on average (and fewer than 100 ceived no information about the country’s location;
cases of autocracy on average) for each combination of others were told that the country was in Africa. When
power and alliances. Nevertheless, the estimated effect we did not specify the location of the target, democracy
of democracy always exceeded 8 percentage points, re- reduced support for a military strike by 11.7 percentage
gardless of whether the target was militarily strong or points, essentially the same as the 11.5-point effect in
weak and whether it had or had not signed an alliance our nationally representative sample. When we told
with Britain.11 In the United States, democracy always respondents that the country was in Africa, the effect of
reduced support for military strikes by at least 6 per- democracy was 15 percentage points, somewhat larger
centage points, no matter the combination of alliances than the effect for a generic country but not statistically
and trade.12 different at conventional confidence levels. Thus, using
In our experiment we did not name the country that MTurk, we replicated the core findings in Table 1 and
was developing nuclear weapons, nor did we identify confirmed that our conclusions did not change when
its location. We intentionally omitted this information we specified a location for the target that excluded
in order to test general hypotheses about the effects countries such as Iran or North Korea.13
of democracy, rather than claims about specific lead- In addition to showing the importance of democ-
ers, nations, or regions. Nonetheless, one might won- racy, our experiments revealed the effects of alliances,
der whether participants reacted strongly because they power, and trade (Table 2). As expected, respondents
assumed the autocracy in our study was either Iran were less willing to strike allies than non-allies, but the
or North Korea. In January 2002, President George existence of an alliance reduced support for military
W. Bush claimed that both countries were sponsoring action by only 5.7% in Britain and 5.1% in the United
terrorism and pursuing weapons of mass destruction, States, effects that were not distinguishable from zero
and he dubbed them—along with Iraq—as the “axis of with 95% confidence.
evil.” If respondents thought we were asking about Iran There was also relatively scant evidence that the
or North Korea when we described a nondemocratic respondents took the conventional military power of
proliferator, they might have been especially inclined the adversary into account. In the United Kingdom,
to strike. where we varied military power, 29% of the public
This possibility seems unlikely for three reasons. wanted to strike a country half as strong as Britain,
First, we told respondents that our scenario was “not whereas 26% stood ready to attack when the target was
about a specific country in the news today.” Second, at conventional parity with Britain. The effect, there-
most respondents received additional information that fore, was 3 percentage points. In summary, enthusiasm
distinguished the target from Iran or North Korea. In for attacking was lower against strong adversaries than
the U.S. study, for example, three-quarters of partic- against weak ones, but the difference was small and not
ipants read that the country had a military alliance statistically significant.
Finally, our experiments provided micro-level ev-
11 The estimated effects of democracy (and 95% confidence inter-
idence about the commercial peace. In the United
vals) were −21.9 (−35.0 to −8.8) percentage points when the target States, where our vignette included information about
was a weak non-ally, −10.9 (−23.1 to 1.8) when the target was a weak
ally, −8.1 (−20.9 to 5.2) when the target was a strong non-ally, and
−12.6 (−24.8 to −0.8) when the target was a strong ally. 13 We also confirmed that our findings were not sensitive to the
12 The within-subject estimates (and 95% confidence intervals) were order in which the questions were posed by fielding a follow-up
−6.3 (−12.6 to 0) when the target was a non-ally with low levels of study that measured perceptions (mediators) before, rather than
bilateral trade, −15.7 (−22.0 to −9.8) when the target was a non-ally after, asking whether respondents would support a military strike.
with high bilateral trade, −7.1 (−13.5 to −0.4) when the target was When we administered this questionnaire to 797 members of MTurk
an ally with low bilateral trade, and −17.1 (−24.1 to −10.1) when in February 2011, the effect of democracy did not budge: Support for
the target was an ally with high bilateral trade. The between-subject a strike remained 11.7 percentage points lower when the potential
estimates were very similar in magnitude. target was a democracy.
855
TABLE 2. The Effect of Alliances, Power, TABLE 3. The Effect of Democracy on

and Trade Perceptions of Threat
United Kingdom United States If the U.S. did not

attack, the country Belief if Effect of
No military alliance 30.7 50.2 would . . . Autocracy Democracy
Military alliance 25.1 45.1
Effect of alliance −5.7 −5.1 Build nuclear weapons 75 −3∗
95% C.I. (−12.0 to 0.6) (−10.7 to 0.5) Threaten to use nukes 52 −14∗
Half as strong 29.3 vs. another country
As strong 26.3 Threaten to use nukes 45 −11∗
Effect of strength −3.0 vs. U.S. or U.S. ally
95% C.I. (−9.4 to 3.2) Launch a nuclear attack 34 −8∗
vs. another country
No high trade 50.3 Launch a nuclear attack 30 −6∗
High trade 45.1 vs. U.S. or U.S. ally
Effect of high trade −5.2
95% C.I. (−10.6 to 0.2) Average 47 −9∗
Note: The table gives the estimated effects (with 95% Note: The first column gives the percentage of respondents
confidence intervals) of alliances, power, and trade on who thought the event had more than a 50% chance
of happening when the country was an autocracy. The
support for a military strike. In the United Kingdom, the
second column shows how much that percentage changed
sample sizes were 371 for no alliance, 391 for alliance, 382
when the same respondents considered an identical
for half as strong as British conventional forces, and 380 for
as strong as British conventional forces. In the United States scenario involving a democracy. Each row is based on 972
respondents. Asterisks denote effects that were significant
the sample sizes were 634 for no alliance, 639 for alliance,
at the .05 level.
612 for no high trade, and 661 for high trade.
trade, 45% of the public endorsed preventive strikes Our investigation proceeded in three steps. First,
against major trading partners. In contrast, 50% ap- we estimated the effect of democracy on each of the
proved of attacking targets that did not trade exten- mediators. This step required no elaborate statistical
sively with the United States. The swing in public opin- modeling. We simply computed how each person’s per-
ion was, therefore, 5 percentage points—less than half ceptions of threat, costs, success, and morality changed
the effect of democracy—and not statistically signif- when we switched the adversary from democracy to au-
icant at the .05 level. In short, our studies provided tocracy. Second, we estimated the effect of each media-
experimental evidence for the democratic peace, while tor on support for a military strike. This step was more
also estimating the influence of alliances, power, and intricate because we observed the mediators instead
trade on attitudes toward military intervention. of randomizing them. We used probit regressions to
estimate the contribution of each mediator, controlling
not only for other mediators but also for variables that
EVIDENCE ABOUT CAUSAL MECHANISMS could confound the estimated relationship between the
We designed the U.S. survey to shed light not only on mediators and the outcome. Finally, we combined the
the effect of democracy but also on the mechanisms findings from these two steps to infer how much of
through which it operates. Earlier, we identified four the total effect of democracy (given in Table 1) was
pathways through which the target’s regime type could transmitted via each of the mediators.
affect the inclination to strike: by changing percep-
tions of threat, costs, success, and/or morality. We refer The Effect of Democracy on Each of the
to these perceptions as mediators, because they could Mediators
mediate the relationship between the treatment vari-
able (democracy) and the final outcome (support for a We begin by discussing how democracy affected each
military strike). of the four mediators. Table 3 summarizes the impact of
To facilitate the analysis of causal mechanisms, we democracy on the first mediator, perceptions of threat.
ran a panel study in which people were interviewed We asked what would happen if the United States did
twice. For every individual who completed both waves not strike. The first column shows what participants
of the panel, we observed the final outcome and the me- expected when the scenario involved an autocracy; the
diators not only when the adversary was a democracy second column tells how expectations changed given an
but also when it was an autocracy. Had we run a purely identical scenario involving a democracy. An asterisk
cross-sectional study, with each individual randomly indicates that the effect was statistically significant at
assigned to either a democracy or an autocracy, half the the .05 level.
measures of outcomes and mediators would have been The first row, labeled “build nuclear weapons,”
missing. By eliminating problems of missing data and shows that democracy did not substantially affect be-
allowing within-subject comparisons, our unique panel liefs about whether the country would finish building
design made it easier to uncover causal mechanisms. nuclear weapons. Three-quarters of the mass public
856
predicted that the autocracy would build a bomb, but

the percentage who expected the democracy to go TABLE 4. The Effect of Democracy on
nuclear was only 3 points lower. This effect was sta- Perceptions of Cost, Success, and Morality
tistically significant but substantively small. Hence, in If the U.S. did Belief if Effect of
our experiment, democracy played only a minor role attack . . . Autocracy Democracy
in allaying fears that the country would build nuclear
weapons. Cost
This weak result may have stemmed from the in- The country would attack 39 0
formation we provided. Respondents read that the U.S. or U.S. ally
country had already refused all requests to cancel its The U.S. military would 32 1
nuclear program, perhaps encouraging respondents to suffer many casualties
conclude that both types of countries were equally The U.S. economy would 31 0
likely to cross the nuclear threshold. Future surveys suffer
U.S. relations with other 49 4∗
could introduce more uncertainty about the country’s countries would suffer
intentions and test whether, given those conditions, Average 38 1
people think nuclearization is more likely under au- Success
tocratic regimes than under democratic ones. It would prevent nukes in 66 −5∗
The next two lines in Table 3 summarize beliefs about the near future
nuclear threats. In the sample as a whole, 52% thought It would prevent nukes in 30 −5∗
that the autocracy would not only build the bomb but the long run
would also threaten to use it against another country. Average 48 −5∗
When those same respondents considered an equiv- Morality
alent democracy, anticipation of nuclear threats was It would be immoral 31 7∗
14 points lower. Similarly, 45% predicted that an au- Note: For our measures of cost and success, the first column
tocracy would issue nuclear threats against the United gives the percentage of respondents who thought the event
States or its allies; those fears dropped by 11 percentage had more than a 50% chance of happening when the country
points when the country was a democracy. was an autocracy. The second column shows how much that
percentage changed when the same respondents considered
Moving further down the table, we see that democ- an identical scenario involving a democracy. Each row is
racy also reduced fears of an actual nuclear attack. based on 972 respondents. Asterisks denote effects that were
Around one-third of respondents thought the autoc- significant at the .05 level.
racy would not only obtain nuclear weapons but would
also launch them against a foreign target. Substantially
fewer thought that the democracy would use its nuclear
arsenal. Here, the effect of democracy was 6–8 percent-
age points. The bottom row of Table 3 gives the mean not oppose attacking democracies because they expect
of the five items. On average, democracy reduced per- democracies to fight more effectively.
ceptions of threat by 9 points. In summary, democracy Next, we studied how democracy affected a third me-
mattered not by lowering the expected probability of diator: beliefs about the probability of success of mil-
building nuclear weapons, but by changing perceptions itary action (bottom half of Table 4). Given an autoc-
about how the country would use them. racy, around two-thirds of respondents thought a U.S.
Although democracy reduced perceptions of threat, strike would prevent nuclear proliferation in the near
it had surprisingly little effect on the second causal future, and roughly a third thought the strike would
pathway: expectations about the costs of fighting stop proliferation over the long run. Respondents were
(Table 4). We asked what would happen if the United somewhat less sanguine about striking democracies:
States struck the country’s nuclear facilities. Thirty- Expectations of success were around 5 points lower
nine percent of the public thought that the autocracy against democratic targets than against autocratic
would retaliate against the United States or a U.S. ally, ones.
but they did not think a democratic target would be- Finally, democracy had a pronounced effect on the
have much differently. Similarly, around a third of re- fourth mediator: the moral intuitions of respondents.
spondents said that the U.S. military would suffer many About a third deemed it immoral to strike an au-
casualties and that the U.S. economy would decline as tocracy, but when respondents read about a democ-
a result of the strike. These perceptions did not depend racy, the moral reluctance to strike grew by 7 points.
on whether the target was a democracy or an autocracy. Thus, our survey provides micro-level evidence that
Democracy did affect forecasts about the cordial- democracy affects the moral calculation for war. Other
ity of U.S. relations with other countries, however. factors equal, people have more moral reservations
Roughly half of the respondents thought that strik- about attacking a democracy than about attacking an
ing an autocracy would hurt U.S. relations with other autocracy.
nations. That prediction was 4 percentage points more In summary, democracy affected some but not all of
common in scenarios involving democratic targets. In the hypothesized mediators. Democracy substantially
general, however, the effect of democracy on the pre- reduced perceptions of threat, but had almost no effect
dicted cost of fighting was weak: only 1 point on on the expected cost of launching a preventive military
average. These findings suggest that U.S. citizens do strike. Respondents were a bit less optimistic about
857
their chances of success against a democracy, and they

were more likely to regard attacking as immoral when TABLE 5. The Effect of Mediators on Support
the adversary was a democracy than when it was an for a Military Strike
autocracy. Variable Coefficient z-stat
The Effect of the Mediators on Support for a Mediators

Military Strike Threat 0.30 13.3∗
Cost −0.21 7.2∗
Next, we estimated the effect of each mediator on sup- Success 0.23 4.8∗
port for military strikes. Having observed the medi- Morality −1.12 12.2∗
ators instead of randomizing them, we needed a sta- Treatments
tistical model with control variables. Given the binary Democracy −0.18 2.9∗
Ally −0.06 0.7
nature of our dependent variable—1 if the respondent
Trade −0.05 0.7
supported a strike and 0 otherwise—we used probit Controls
regression. Militarism −0.02 0.2
The key independent variables for these analy- Internationalism 0.02 0.4
ses were the four mediators: threat, cost, success, Republican 0.10 1.4
and immorality. To summarize perceptions of threat, Ethnocentrism 0.09 1.1
we counted the number of adverse events (listed in Religiosity −0.03 0.5
Table 3) that respondents marked as probable if the Male 0.05 0.6
United States did not strike the country’s nuclear facil- White −0.18 1.8
ities. Threat ranged from 0 to 5, with a mean of 2.1. Sim- Age 0.01 2.2∗
Education −0.06 1.5
ilarly, we summarized perceptions of cost by counting
Intercept −0.30 1.5
the number of negative consequences—military retali-
ation, high casualties, economic damage, and deterio- Note: The table gives the estimated coefficients and z-statistics
rating relations—that the respondent anticipated if the from a probit regression. The dependent variable is coded 1
United States carried out the operation. Cost ranged if the respondent supported a military strike and 0 otherwise.
from 0 to 4, with a mean of 1.5. To gauge perceptions Each of the 972 respondents appears in the sample twice,
once when treated with democracy and once when treated with
of success, we asked whether respondents thought the autocracy. Z-statistics are computed from standard errors that
mission would stop the country from building nuclear were clustered by respondent. An asterisk indicates that the
weapons. Success was 2 if respondents thought the mis- estimated coefficient is statistically significant at the .05 level.
sion would succeed both in the short and in the long
run, 1 if it would prove efficacious only in the short run,
and 0 if it had less than a 50–50 chance of working even
in the near term: The mean was 0.9. Finally, Immorality shifted each mediator from its minimum to its maxi-
was coded 1 if respondents thought it would be morally mum, while holding the other variables at their means.
wrong to strike (35%) and 0 otherwise. The effects were massive. If perceptions of threat rose
We then added dummy variables for each of the ran- from low to high, support for military action would
domized treatments: Democracy, Ally, and Trade. Fi- increase by 54 points. Similarly, a groundswell of opti-
nally, we included demographic and attitudinal control mism about the chances of success would boost support
variables that, if omitted, could confound the relation- by 18 points. Conversely, if the expected cost changed
ship between the mediators and support for the use from low to high, the popularity of military action
of force. For example, we controlled for whether the would decline by 30 points. Finally, if people came to
respondent was Male (50%) and White (78%). We also view the operation as immoral, enthusiasm would drop
controlled for the respondent’s Age in years (mean by 39 points.
of 53) and level of Education. Finally, to control for
baseline attitudes toward the use of military force, we Overall Estimates of Causal Mechanisms
included indices of Militarism, Internationalism, Reli-
giosity, Ethnocentrism, and identification with the Re- We have now estimated the effect of democracy on
publican Party. Each of these indices had a mean of each mediator and the effect of each mediator on sup-
0 and a standard deviation of about 0.8; details on port for military strikes. By joining these parts of the
the construction of these variables are available in the causal chain, we can see how perceptions of threat,
Online Appendix. cost, success, and morality mediate the relationship
Table 5 confirms that, when deciding whether to use between democracy and strikes (Imai et al. 2011; Imai,
military force, people weighed the threat the adversary Keele, and Yamamoto 2010) and thereby assess the
posed, the expected cost of taking military action, the mechanisms linking democracy and public support for
perceived likelihood of success, and the morality of em- war.
ploying violence. Threat, Cost, Success, and Immorality Recall that every individual received two scenarios,
all worked in the hypothesized directions and were one in which the target was a democracy and another
statistically significant at the .05 level. in which the target was an autocracy. The role of any
To judge the importance of these four variables, we particular mediator can be quantified by measuring
simulated how support for a strike would change if we a person’s willingness to strike when the mediator
858
takes on its democracy value, and subtracting that

same person’s willingness to strike when the mediator TABLE 6. Estimates of Causal Mechanisms
takes on its autocracy value, with all other factors held Average Effect via % of Total Effect
constant. Mediator this Mediator of Democracy
More precisely, for each individual i who completed
both waves of the survey, let Ti be a treatment indicator Threat −4.0∗ 34
that takes a value of 1 when i was asked about a democ- Cost −0.4 4
racy, and 0 when i was asked about an identical scenario Success −0.7∗ 6
involving an autocracy. Use Yi (t) to denote i’s support Morality −1.7∗ 15
for a military strike under treatment condition Ti = t.
Note: The table decomposes the total effect of democracy into
Because each panelist considered both a democracy various pathways or mechanisms. It shows the percentage of
and an autocracy, we observed both Yi (1) and Yi (0) for the total effect that was transmitted by threat, cost, success,
every i. and immorality. Estimates are based on data from 972
Our analysis focused on four mediators, which we respondents. Asterisks indicate effects that are statistically
index as k = {1, 2, 3, 4}. For each person in our panel, significant at the .05 level.
let Mik (1) represent the value of mediator k when the
target is a democracy, and let Mik (0) represent the 2. For each i,
value of that same mediator when the country is an
autocracy. Because of the special design of our survey, a. Use the parameter estimates from the pro-
we observed both Mik (1) and Mik (0) for every i and bit model to predict the probability of sup-
every k. porting a military strike, given Ti = 1; Mi =
For any given individual, the effect of democracy Mik (0) , Mi−k (1) ; and Xi , where Mik (0) is the ob-
transmitted via mediator k is served value of mediator k when i read about an
autocracy, and Mi−k (1) are the observed values of
νki = Yi 1, Mik (1) , Mi−k (1) − Yi 1, Mik (0) , Mi−k (1) . the other mediators when i read about a democ-
racy. Denote this prediction as π̂i .
(1) b. Draw Ỹi ∼ Bernoulli(π̂i ).
c. Compute ν̃ki = Yi (1) − Ỹi .
The first term on the right-hand side is i’s support n
for a military strike when the target is a democ- 3. Compute the sample-wide average, ν̃k = 1
n i=1 ν̃i .
k
racy, mediator k takes on its democracy value, and

all the other mediators (−k, meaning “not k”) take This algorithm produces one sample-wide estimate
on their democracy values. The second term is iden- for each of the four mediators. One can approximate
tical, except that mediator k takes on its autocracy the sampling distributions of the ν̃k’s by repeating the
value. algorithm many times, with each iteration based on a
νki is the difference between an observable different bootstrap resample of the original data.14
quantity
and a counterfactual
one. The minuend, Using this algorithm, we estimated how much of the
Yi 1, Mik (1) , Mi−k (1) , simplifies to Yi (1), which we total effect of democracy (from Table 1) was transmit-
measured for every person in our panel. The subtra- ted via each of the four mediators. Recall that, in our
hend, Yi 1, Mik (0) , Mi−k (1) , in contrast, is hypotheti- panel study, democracy reduced support for a military
cal. It represents the preference i would have expressed strike by 11.5 percentage points. Table 6 shows that
if he or she were considering a democracy but perceived democracy exerted about 34% of this effect by chang-
mediator k as if the country had been an autocracy. ing perceptions of threat and an additional 15% by
Because the subtrahend is a counterfactual, νki is not altering perceptions of morality. The mediatory roles
directly observable. of cost and success were much weaker; cost in particular
Fortunately, one can estimate νki and the sample- was not statistically significant.
We found little evidence that democracy promotes
wide average, νk = n1 ni=1 νki, by applying the following
algorithm: peace by changing perceptions of cost and success.
This does not mean that citizens disregarded the ex-
pected cost of fighting and the probability of success.
1. Using all 2n cases (because each of the n re-
On the contrary, Table 5 showed that respondents were
spondents received both the democracy and the
much less enthusiastic about military action when they
autocracy treatment), estimate a probit model of
support for a military strike. In this model, Yi ∼
Bernoulli(πi ) and πi = (αTi + βMi + γXi ), where 14 Our algorithm relies on the standard assumption of sequential
is the cumulative normal distribution, Ti is the ignorability, meaning that the mediators—threat, cost, success, and
treatment indicator with coefficient α, Mi is a vector morality—were as if randomly assigned, given the covariates in
of mediators with coefficients β, and Xi is a vector of Table 5 and the randomized democracy/autocracy treatment. Our
control variables with coefficients γ. Xi includes not algorithm also presumes that the mediators operate independently,
instead of influencing each other. Later we relax this assumption
only demographic and attitudinal variables but also by considering how the conclusions might change if perceptions
indicator variables for ally and trade. We estimated of morality were a consequence of beliefs about threat, cost, and
this model in Table 5. success.
859
thought strikes would be costly or unsuccessful. Rather, tant. Indeed, respondents did not seize on every piece
the reason that cost and success did not mediate the of information we supplied; the conventional military
effect of democracy is because democracy had rela- power of the adversary, for example, did not signifi-
tively small effects on perceptions of costs and success cantly influence attitudes toward war. We suspect that
(Table 4). the effect of democracy on opinion may be stronger,
Morality proved to be a far more important me- rather than weaker, in real-life crises, where the stakes
diator. But did people regard preventive strikes as are higher and individuals are repeatedly exposed to
morally wrong because they thought the target posed information about regime type.16
little threat, the attack would involve significant costs, A second question is whether, in actual crises, leaders
and/or military action would fail? To answer this ques- could shape public opinion by spreading false infor-
tion, we carried out a more complicated analysis in mation about the regime type of the potential adver-
which we modeled morality not only as an indepen- sary. Proponents of war might try to mischaracterize a
dent force but also as a potential consequence of the democratic target as an autocratic state, whereas op-
other mediators. Having estimated this more compli- ponents of war might speak about an autocracy as if it
cated model, we credited morality as a mediator only were a democracy. If leaders could easily manipulate
to the extent that democracy changed perceptions of beliefs about regimes in foreign countries, this would
morality directly. Where democracy influenced moral- weaken the link between a country’s true regime type
ity indirectly—by altering other mediators that, in turn, and public support for a strike. However, free speech
affected morality—we allocated credit to the other me- and freedom of the press limit the extent to which pol-
diators and not to morality itself.15 Even with this con- icy makers can make fallacious claims about political
servative method of scoring, morality mediated more systems in other nations. Although elites may be able to
than 10% of the total effect of democracy on support influence public perceptions of regime type, especially
for war. when foreign regimes fall into a “gray area” between
democracy and autocracy, they probably cannot dupe
the public into thinking that highly democratic leaders
IMPLICATIONS FOR THE DEMOCRATIC are tyrants and vice versa.
PEACE Third, skeptics might wonder whether people re-
Using survey experiments, we found clear micro-level sponded differently to our hypothetical scenarios than
evidence of a democratic peace. Individuals in the they would have responded to specific countries. We
United States and United Kingdom were substantially consciously avoided naming countries because we
less willing to attack democracies than to attack oth- wanted to learn about the general effects of democracy,
erwise equivalent autocracies. Moreover, the target’s rather than specific countries or leaders. We also wor-
regime type mattered mainly by altering perceptions of ried that, by asking respondents to compare a named
threat and morality, rather than by raising expectations autocracy with a named democracy, we would lose ex-
of costs or failure. Thus, our data showed a strong causal perimental control, because countries differ on many
relationship between democracy and peace in public dimensions other than regime type. Nevertheless, to
opinion, while also illuminating several mechanisms make our vignette more concrete, recall that we ran a
that drive this relationship. We now consider several follow-up experiment that located the target in Africa;
questions about the interpretation of our findings. it confirmed that the effect of democracy remained the
same. This discovery corroborates previous research,
which found little difference in public reactions to
Generalizability hypothetical versus real scenarios and to generic ver-
First, critics might wonder whether our experiments sus actual countries (Berinsky 2009, 124; Herrmann,
exaggerated the importance of democracy by making Tetlock, and Visser 1999; Horowitz and Levendusky
the regime of the adversary salient. We believe that this 2011, 531–32).
concern is misplaced. In actual crises, politicians and Fourth, can we extrapolate from surveys in the
the media make information about democracy signifi- United States and Britain to draw more general con-
cant to voters. Before the U.S. invasion of Iraq in 2003, clusions about the democratic peace? It bears noting
for example, Saddam Hussein was constantly portrayed that the effects of democracy were similar among U.K.
as the dictator—not the elected leader—of Iraq. Policy and U.S. respondents, despite large differences in the
makers and newscasters often use evocative language militarism of the British and American electorates. This
(“tyrant,” “dictator”) to describe regime type, whereas suggests that our findings generalize to countries with
our experiments deliberately used neutral terms (“not varying attitudes about military action. Moreover, as
a democracy”). Moreover, we explicitly invited respon- two of the most influential democracies in the world,
dents to disregard any facts they regarded as unimpor- the United Kingdom and United States are important
in and of themselves. Their willingness to use force
against autocracies and their comparative reticence to
15 Alternatively, one could give morality credit as a mediator, not
only when democracy affects morality directly but also when democ-
racy affects morality indirectly via changes in perceptions of threat, 16 Of course, politicians sometimes offer justifications for using force
cost, and success. Had we taken this approach, we would have in- or remaining at bay, whereas our scenario did not provide arguments
ferred an even larger role for morality and concomitantly smaller for or against attacking. Future experiments could expose respon-
roles for threat, cost, and success. dents to a variety of arguments about the merits of war.
860
use force against democracies have profound effects Importantly, the swing was even larger among the
on international relations in general. Nevertheless, it most politically engaged segments of the population.
would be useful to replicate our experiments in a wider We examined the opinions of politically attentive U.S.
range of countries and regions to see how our findings citizens, who follow government and public affairs most
extend to other cultural, political, economic, and secu- of the time (63% of the sample). Within this politically
rity contexts. attentive group, democracy reduced enthusiasm for
Fifth, contemporary surveys do not tell us whether war by 14.5 percentage points versus 5.0 points among
citizens held similar views in earlier time periods. Have citizens who were less politically aware. A similar pat-
citizens of democracies always had pacific beliefs about tern emerged in the United Kingdom. There, people
other democracies, or did those views develop over somewhat or very interested in politics (71% of the
time? Modern-day surveys cannot answer this ques- sample) were 17.4 points less willing to strike a demo-
tion, but future research could trace when and how cratic opponent than an autocratic one. By contrast,
pacific beliefs about fellow democracies emerged (Kahl the democracy effect was 3.4 points among respon-
1998; Owen 1997; Williams 2001). Indeed, such re- dents who were “not much” or “not at all” interested
search might shed light on why the democratic peace in politics.17 Thus, the regime type of the adversary ex-
is stronger now than it was in the nineteenth and early erted a large effect on the people most likely to follow
twentieth centuries (Gowa 1999; McDonald 2009). At politics.
a minimum, our experiments show that the democratic Regime type also mattered greatly for voters. In the
peace is alive and well in two of the most important United States, the pacifying effect of democracy was
democratic military powers in the world, and they shed 12.1 points among respondents who said they voted in
light on the mechanisms through which democracy con- the 2008 election (85% of the sample), versus 5.7 points
tributes to a preference for peace in the contemporary among respondents who did not. In the United King-
era. dom, respondents were asked whether they intended
Finally, it is worth considering whether our choice to vote in the future. Among people who said they
of a scenario involving nuclear proliferation affected were very likely to vote if a general election were held
our conclusions about how and why democracy affects tomorrow (80% of the sample), democracy depressed
the willingness to strike. One possibility is that regime support for war by 15.3 points, versus 6.7 points among
type matters when the stakes are low, but not when people less likely to vote.18
the stakes are high, because in high-threat situations Moreover, democracy had a large effect on citizens
one must always assume the worst about the inten- who went beyond voting to participate more actively
tions of other countries (Mearsheimer 2003). From in politics. Building on the work of Verba, Schlozman,
that perspective, we chose a hard test of democratic Brady, and Nie (1993), we classified someone in the
peace theory. If democracy matters even in a situation United States as a political activist if, in the past year,
of existential threat, it might matter even more when he or she had worked for a political campaign, donated
the stakes are lower. An alternative view is that, by money to a campaign, put up a political sign such as a
choosing a high-stakes issue, we created more room for lawn sign or a bumper sticker, or attended local politi-
democracy to decrease perceptions of threat, and hence cal meetings. The effect of democracy on these activists
willingness to strike. If we had instead chosen an issue (46% of the sample) averaged 16.5 points, compared
in which the baseline threat was minimal, there would with 11.5 points among those who were not as active.19
have been little opportunity for democracy to further The British survey contained different measures of
reduce threat, and the role of other mediators might political activism, but our overall conclusion was simi-
have been more prominent. Future research could in- lar. We coded British respondents as highly active if, in
vestigate these questions by applying our experimental the previous day, they had worn a badge or sticker for
template to other types of issues. a candidate, discussed a candidate with someone else,
gone to hear a candidate speak, visited the website
of a candidate or political party, or watched a video
of a candidate on the internet (24% of the sample).
Political Consequences
Members of this highly energized subgroup were 26.7
A different set of questions involves the political sig- points more reluctant to attack a democracy than to
nificance of the effects uncovered in our experiments.
First, were the effects we observed large enough to
be politically consequential? We believe so. In our ex- 17 In both countries, the effect of democracy on the attentive public
periments, democracy reduced willingness to strike by was statistically different from zero at the .05 level, and the gap
about 11–13 percentage points. Shifts of that magni- between attentive and inattentive citizens was statistically significant
at the .10 level or better. Throughout this section, we report between-
tude would change the nature of political debate. They subject estimates for the United States; within-subject estimates were
could also make the difference between a majority and very similar.
a minority. In the United States, for example, a majority 18 In both countries, the effect of democracy on voters was statis-
of the mass public favored a preventive strike when the tically significant, but the difference between voters and nonvoters
target was an autocracy, but only a minority wanted to was not.
19 Data on political activism were only available for people who
strike the democracy. This is important because democ- participated in both waves of the survey. The effect of democracy on
racies typically do not go to war in the face of public each group was statistically significant, but the difference between
opposition (Reiter and Stam 2002). groups was not.
861
attack an autocracy. The effect among more typical targets. Public opinion may therefore foster a special
citizens was 8.9 percentage points.20 Thus, in both the zone of peace among democracies.
United States and the United Kingdom, the regime Moreover, our experimental approach allows us to
type of the adversary not only affected public opinion conclude with confidence that the effect of democracy
in the aggregate but also had a particularly profound is genuinely causal. Democracy affects preferences in-
effect on the most politically active segments of the dependent of confounders such as alliances, power, and
population. trade. Although our experiments confirm the intuition
A second issue is how to interpret the finding that a of skeptics that at least part of the peace among democ-
sizable number of our American respondents—around racies is due to shared interests, military power, and
42%—indicated that they would support an attack on economic ties, we nonetheless find clear evidence that
a fellow democracy in order to stop it from acquiring democracy has an independent effect on support for
nuclear weapons. This evidence might be interpreted war.
as disconfirming the democratic peace, by showing that Our experiments also reveal the mechanisms
a large percentage of U.S. citizens are willing to attack a through which democracy dampens support for war.
democracy. We think this would be a mistaken interpre- The finding that democracies view other democracies
tation for two main reasons. First, our experiments pre- as less threatening, which in turn reduces support for
sented respondents with a highly threatening scenario: using force, accords with major works on the demo-
a country that refused to stop its nuclear program. Our cratic peace that emphasize threat perception (Risse-
analysis confirms that high levels of threat increase Kappen 1995; Russett 1993). Understanding how and
support for military strikes (Table 5); support for war why democracies trust fellow democracies, but not au-
would be much lower in a less threatening scenario. tocracies, is an important avenue for future research
Second, critics of the democratic peace rightly point out (Kahl 1998; Williams 2001). We also found that per-
that democracy coincides with other pacifying factors, ceptions of cost do not explain the public aversion
such as alliances, trade, and power. The combination to fighting democracies, and that expectations about
of those forces can lead to major swings in prefer- success explain only a small amount of the effect.
ences. In the United States, around 55% of respon- Finally, we found that morality plays an important
dents wanted to strike an autocracy that was neither role in the democratic peace. The regime type of the
an ally nor a major trading partner. In contrast, only target affects moral calculations, which in turn change
30% were willing to attack a democracy that was also preferences about the use of force. Surprisingly few
an ally and a trading partner. Thus, if the United States scholars have explored morality as a potential source
typically entered alliances with other democracies and of the democratic peace. This should be a major topic
traded extensively with them, while forgoing alliances for future research.
and trade with autocracies, respondents would be 25 There are numerous opportunities for follow-up
points less willing to attack democracies than to at- studies. For example, the experiments in this article
tack autocracies.21 In the United Kingdom, around cannot distinguish between “normative” and “struc-
43% of respondents supported attacking a militarily tural” theories, both of which predict that democracy
weak, autocratic non-ally, whereas less than 17% sup- reduces threat perception. Do democracies seem less
ported attacking a militarily powerful democratic ally, a threatening because people think democracies will ex-
swing of 26 points. These are consequential differences ternalize their domestic values of peaceful coexistence,
indeed. because they believe democratic institutions will slow
or prevent the march to war, or both? Researchers
could address this question by randomizing informa-
CONCLUSION tion about the normative and structural attributes of
The fact that democracies almost never fight each other regimes. One could also test whether the perceived
is one of the most striking findings in political science. credibility of threats and promises varies by regime
Yet scholars continue to debate whether the relation- type (Fearon 1994; Schultz 2001; Slantchev 2005; 2011),
ship between democracy and peace is causal and what whether people think democracies would be more will-
mechanisms explain it. This article uses experiments, ing to make peaceful bargains (Debs and Goemans
embedded in public opinion polls, to shed new light on 2010), and whether democracy leads to perceptions of
the democratic peace. shared preferences (Oneal and Russett 1999).
Our research supports the hypothesis that peace Moreover, our study provides a template for re-
among democracies could be due, at least in part, search on issues other than nuclear proliferation. His-
to public opinion. Countless studies have shown that torically, no issue has driven countries toward war more
democratic leaders are responsive to public opinion on often than national boundaries (Ghosn, Palmer, and
matters of foreign policy; we demonstrate that the pub- Bremer 2004). Although shared democracy may pre-
lic discriminates between democratic and autocratic vent territorial disputes from escalating, some have
claimed that the militarization of territorial disputes
20 The effect of democracy on each group was statistically different
should not vary by regime type (Gibler 2007). Coun-
tries also spar over domestic policies such as respect
from zero, and the difference between the groups was statistically
significant at the .05 level. for human rights and cracking down on terrorism.
21 Based on the within-subject analysis. The between-subject esti- Democracy may contribute to peace in these instances,
mate was 22 points. but because changing the domestic policies of another
862
country involves interference in that country’s internal References

affairs, democracy could matter mainly by affecting
beliefs about the morality of intervention, rather than Aldrich, John H., John L. Sullivan, and Eugene Borgida. 1989.
by reducing perceptions of threat. Researchers could “Foreign Affairs and Issue Voting.” American Political Science
design experiments to assess these predictions. Review 83(1): 123–41.
Almond, Gabriel A. 1960. The American People and Foreign Policy.
Future studies could also explore policy responses New York: Praeger.
other than military force. Countries can deal with inter- Baum, Matthew A. 2004. “How Public Opinion Constrains the Use of
national disputes in a variety of ways, including diplo- Force: The Case of Operation Restore Hope.” Presidential Studies
macy and mediation (Dixon 1994), appealing to in- Quarterly 34(2): 187–226.
Baum, Matthew A., and Tim Groeling. 2009. “Shot by the Messenger:
ternational law and organizations (Russett and Oneal Partisan Cues and Public Opinion Regarding National Security
2001), or applying economic sanctions. New surveys and War.” Political Behavior 31(2): 157–86.
could ask respondents to evaluate a wide range of mil- Baum, Matthew A., and Philip B. K. Potter. 2008. “The Relationships
itary and nonmilitary options, thereby revealing how between Mass Media, Public Opinion, and Foreign Policy: Toward
democracy affects not only decisions about war but a Theoretical Synthesis.” Annual Review of Political Science 11:
39–65.
also the use of nonviolent foreign policy tools and the Berinsky, Adam J. 2007. “Assuming the Costs of War: Events, Elites,
willingness to escalate from peaceful measures to vio- and American Public Support for Military Conflict.” Journal of
lent ones. Politics 69(4): 975–97.
Finally, our study focused on the public in the United Berinsky, Adam J. 2009. In Time of War. Chicago: University of
Chicago Press.
Kingdom and the United States. Researchers could Berinsky, Adam J., Gregory A. Huber, and Gabriel S. Lenz. 2012.
replicate our experiments in other countries to see “Evaluating Online Labor Markets for Experimental Research:
how our findings generalize to other parts of the world. Amazon’s Mechanical Turk.” Political Analysis 20: 351–68.
Additionally, researchers could interview government Bueno de Mesquita, Bruce, James D. Morrow, Randolph M. Siver-
officials and other policy elites to assess how democracy son, and Alastair Smith. 1999. “An Institutional Explanation of
the Democratic Peace.” American Political Science Review 93(4):
affects the preferences and perceptions of the people 791–807.
who ultimately decide whether to go to war. Some Canes-Wrone, Brandice. 2006. Who Leads Whom? Presidents, Policy,
scholars have found that elites and the public have and the Public. Chicago: University of Chicago Press.
similar beliefs about foreign policy issues,22 whereas Dafoe, Allan. 2011. “Statistical Critiques of the Democratic Peace:
Caveat Emptor.” American Journal of Political Science 55(2): 247–
others have concluded that elites and masses respond 62.
differently to international cues, with elites demon- Debs, Alexandre, and H. E. Goemans. 2010. “Regime Type, the Fate
strating more complex strategic thinking.23 Future of Leaders, and War.” American Political Science Review 104(3):
studies could assess whether policy makers respond 430–45.
to democracy differently than the masses. Dixon, William J. 1994. “Democracy and the Peaceful Settlement of
International Conflict.” American Political Science Review 88(1):
For decades, U.S. and foreign leaders have cited 14–32.
the democratic peace when analyzing foreign affairs Doyle, Michael W. 1983. “Kant, Liberal Legacies, and Foreign
and justifying efforts to spread democracy around Affairs.” Philosophy & Public Affairs 12(3): 205–35.
the globe. This topic remains critical today, given the Doyle, Michael W. 1986. “Liberalism and World Politics.” American
Political Science Review 80(4): 1151–69.
tremendous pressure for democracy around the world. Edwards, George. 1997. “Aligning Tests with Theory: Presidential
By providing micro-level evidence about the demo- Approval as a Source of Influence in Congress.” Congress and the
cratic peace and its causes, this article has important Presidency 24(2): 113–30.
implications for policy makers as well. For example, Farber, Henry S., and Joanne Gowa. 1995. “Polities and Peace.”
policy makers who want to cooperate with autocracies International Security 20(2): 123–46.
Fearon, James D. 1994. “Domestic Political Audiences and the Es-
must recognize that citizens in democracies distrust dic- calation of International Disputes.” American Political Science
tators and have fewer moral reservations about using Review 88(3): 577–92.
force against them. On the flip side, our findings suggest Fearon, James D. 1995. “Rationalist Explanations for War.” Interna-
that democratization may have security benefits. When tional Organization 49(3): 379–414.
Finnemore, Martha. 2003. The Purpose of Intervention. Ithaca, NY:
a democracy and an autocracy take identical actions, Cornell University Press.
the democracy receives the benefit of the doubt from Foyle, Douglas C. 1999. Counting the Public In: Presidents, Pub-
citizens in fellow democracies, whereas the autocracy lic Opinion, and Foreign Policy. New York: Columbia University
is viewed with much greater suspicion. Advocates of Press.
democracy promotion could therefore cite our findings Gartner, Scott Sigmund. 2008. “The Multiple Effects of Casualties on
Public Support for War: An Experimental Approach.” American
as evidence that joining the club of democracies confers Political Science Review 102(1): 95–106.
significant benefits not only at home but also abroad. Gartzke, Erik. 1998. “Kant We All Just Get Along? Opportunity,
Willingness, and the Origins of the Democratic Peace.” American
Journal of Political Science 42(1): 1–27.
22 See for example Herron and Jenkins-Smith (2002). Other scholars Gartzke, Erik. 2007. “The Capitalist Peace.” American Journal of
have found a match between elite and mass opinions about the use Political Science 51(1): 166–91.
of force after adjusting for demographic differences between the Gelpi, Christopher, Peter D. Feaver, and Jason Reifler. 2006. “Suc-
two populations. See, for example, Wittkopf and Maggiotto (1983) cess Matters: Casualty Sensitivity and the War in Iraq.” Interna-
and Wittkopf (1987). Holyk (2011) finds that leaders and masses tional Security 30(3): 7–46.
have comparable foreign policy opinion structures, meaning that the Gelpi, Christopher, Peter Feaver, and Jason A. Reifler. 2009. Paying
correlations between foreign policy opinions and other variables are the Human Costs of War: American Public Opinion and Casualties
about the same for the two groups. in Military Conflicts. Princeton, NJ: Princeton University Press.
23 For a helpful overview of this literature, see Hafner-Burton, Gelpi, Christopher, Jason Reifler, and Peter Feaver. 2007. “Iraq the
Hughes, and Victor (2013). Vote: Retrospective and Prospective Foreign Policy Judgments
863
on Candidate Choice and Casualty Tolerance.” Political Behavior Jervis, Robert. 1976. Perception and Misperception in International
29(2): 151–74. Politics. Princeton, NJ: Princeton University Press.
Geva, Nehemia, Karl R. DeRouen, and Alex Mintz. 1993. “The Jervis, Robert. 1978. “Cooperation under the Security Dilemma.”
Political Incentive Explanation of ‘Democratic Peace’: Evidence World Politics 30(2):167–214.
from Experimental Research.” International Interactions 18(3): Johns, Robert, and Graeme Davies. 2012. “Democratic Peace or
215–29. Clash of Civilizations? Target States and Support for War in Britain
Geva, Nehemia, and D. Christopher Hanson. 1999. “Cultural Sim- and the United States.” Journal of Politics 74(4): 1038–52.
ilarity, Foreign Policy Actions, and Regime Perception: An Ex- Jones, Byron, and Michael G. Kenward. 2003. Design and Analysis
perimental Study of International Cues and Democratic Peace.” of Cross-over Trials. 2nd ed. New York: Chapman & Hall.
Political Psychology 20(4): 803–27. Kahl, Colin H. 1998. “Constructing a Separate Peace: Construc-
Ghosn, Faten, Glenn Palmer, and Stuart A. Bremer. 2004. “The tivism, Collective Liberal Identity, and Democratic Peace.”
MID3 Data Set, 1993–2001: Procedures, Coding Rules, and Security Studies 8(2–3): 94–144.
Description.” Conflict Management and Peace Science 21(2): Kant, Immanuel. [1795] 1991. Kant’s Political Writings. Edited by
133–54. Hans Reiss. 2nd ed. Cambridge: Cambridge University Press.
Gibler, Douglas M. 2007. “Bordering on Peace: Democracy, Terri- Krosnick, Jon A., and Donald R. Kinder. 1990. “Altering the Foun-
torial Issues, and Conflict.” International Studies Quarterly 51(3): dations of Support for the President through Priming.” American
509–32. Political Science Review 84(2): 497–512.
Gowa, Joanne S. 1999. Ballots and Bullets: The Elusive Democratic Kydd, Andrew H. 2005. Trust and Mistrust in International Relations.
Peace. Princeton, NJ: Princeton University Press. Princeton, NJ: Princeton University Press.
Grieco, Joseph M., Christopher Gelpi, Jason Reifler, and Peter D. Lacina, Bethany, and Charlotte Lee. 2012. “Culture Clash or Demo-
Feaver. 2011. “Let’s Get a Second Opinion: International Institu- cratic Peace?: Results of a Survey Experiment on the Effect of
tions and American Public Support for War.” International Studies Religious Culture and Regime Type on Foreign Policy Opinion
Quarterly 55(2): 563–83. Formation.” Foreign Policy Analysis 9(2): 143–70.
Gronke, Paul, Jeffrey Koch, and J. Matthew Wilson. 2003. “Fol- Lake, David A. 1992. “Powerful Pacifists: Democratic States and
low the Leader? Presidential Approval, Presidential Support, and War.” American Political Science Review 86(1): 24–37.
Representatives’ Electoral Fortunes.” Journal of Politics 65(3): Lektzian, David, and Mark Souva. 2009. “A Comparative Theory
785–808. Test of Democratic Peace Arguments, 1946–2000.” Journal of
Hafner-Burton, Emilie M., Alex Hughes, and David G. Victor. 2013. Peace Research 46(1): 17–37.
“The Cognitive Revolution and the Political Psychology of Elite Levendusky, Matthew S., and Michael C. Horowitz. 2012. “When
Decision Making.” Perspectives on Politics 11(2): 368–86. Backing Down Is the Right Decision: Partisanship, New Informa-
Hartley, Thomas, and Bruce Russett. 1992. “Public Opinion and the tion, and Audience Costs.” Journal of Politics 74(2): 323–38.
Common Defense: Who Governs Military Spending in the United Levy, Jack S. 1988. “Domestic Politics and War.” Journal of Interdis-
States?” American Political Science Review 86(4): 905–15. ciplinary History 18(4): 653–73.
Harrison, Ewan. 2004. “State Socialization, International Norm Dy- Lindsay, James M. 1994. Congress and the Politics of U.S. Foreign
namics and the Liberal Peace.” International Politics 41(4): 521–42. Policy. Baltimore: Johns Hopkins University Press.
Hermann, Margaret G., and Charles W. Kegley. 1995. “Rethinking Maoz, Zeev, and Bruce Russett. 1993. “Normative and Structural
Democracy and International Peace: Perspectives from Political Causes of Democratic Peace, 1946–1986.” American Political Sci-
Psychology.” International Studies Quarterly 39(4): 511–33. ence Review 87(3): 624–38.
Herrmann, Richard K., and Vaughn P. Shannon. 2001. “Defending McDermott, Rose. 2011. “New Directions for Experimental Work
International Norms: The Role of Obligation, Material Interest, in International Relations.” International Studies Quarterly 55(2):
and Perception in Decision Making.” International Organization 503–20.
55(3): 621–54. McDonald, Patrick J. 2009. The Invisible Hand of Peace: Capitalism,
Herrmann, Richard K., Philip E. Tetlock, and Penny S. Visser. 1999. the War Machine, and International Relations Theory. New York:
“Mass Public Decisions to Go to War: A Cognitive-Interactionist Cambridge University Press.
Framework.” American Political Science Review 93(3): 553–73. Mearsheimer, John J. 2003. The Tragedy of Great Power Politics.
Herrmann, Richard K., James F. Voss, Tonya Y. E. Schooler, and New York: W. W. Norton.
Joseph Chiarrochi. 1997. “Images in International Relations: An Mintz, Alex, and Nehemia Geva. 1993. “Why Don’t Democracies
Experimental Test of Cognitive Schemata.” International Studies Fight Each Other? An Experimental Study.” Journal of Conflict
Quarterly 41(3): 403–33. Resolution 37(3): 484–503.
Herron, Kerry G., and Hank C. Jenkins-Smith. 2002. “U.S. Percep- Morgan, T. Clifton, and Sally Howard Campbell. 1991. “Domestic
tions of Nuclear Security in the Wake of the Cold War: Comparing Structure, Decisional Constraints, and War.” Journal of Conflict
Public and Elite Belief Systems.” International Studies Quarterly Resolution 35(2): 187–211.
46(4): 451–79. Morgenthau, Hans J. 1948. Politics among Nations. New York:
Hildebrandt, Timothy, Courtney Hillebrecht, Peter M. Holm, and Knopf.
Jon Pevehouse. 2013. “The Domestic Politics of Humanitarian Mueller, John E. 1973. War, Presidents, and Public Opinion. New
Intervention: Public Opinion, Partisanship, and Ideology.” Foreign York: Wiley.
Policy Analysis 9(3): 243–66. Narizny, Kevin. 2003. “Both Guns and Butter, or Neither: Class
Holsti, Ole R. 2004. Public Opinion and American Foreign Policy. Interests in the Political Economy of Rearmament.” American
rev. ed. Ann Arbor: University of Michigan Press. Political Science Review 97(2): 203–20.
Holyk, Gregory G. 2011. “Individual-level Predictors of Leader and Oneal, John R., and Bruce M. Russett. 1999. “Is the Liberal Peace
Public Support for the Use of Force.” International Journal of Just an Artifact of Cold War Interests? Assessing Recent Cri-
Public Opinion Research 23(2): 214–29. tiques.” International Interactions 25(3): 213–41.
Horowitz, Michael C., and Matthew S. Levendusky. 2011. “Drafting Owen, John M. 1994. “How Liberalism Produces Democratic
Support for War: Conscription and Mass Support for Warfare.” Peace.” International Security 19(2): 87–125.
Journal of Politics 73(2): 524–34. Owen, John M. 1997. Liberal Peace, Liberal War: American Politics
Howell, William G., and Jon C. Pevehouse. 2007. While Dangers and International Security. Ithaca, NY: Cornell University Press.
Gather: Congressional Checks on Presidential War Powers. Prince- Press, Daryl G., Scott D. Sagan, and Benjamin A. Valentino. 2013.
ton, NJ: Princeton University Press. “Atomic Aversion: Experimental Evidence on Taboos, Traditions,
Imai, Kosuke, Luke Keele, Dustin Tingley, and Teppei Yamamoto. and the Non-use of Nuclear Weapons.” American Political Science
2011. “Unpacking the Black Box of Causality: Learning about Review 107(1): 188–206.
Causal Mechanisms from Experimental and Observational Stud- Price, Richard. 1998. “Reversing the Gun Sights: Transnational Civil
ies.” American Political Science Review 105(4): 765–89. Society Targets Land Mines.” International Organization 52(3):
Imai, Kosuke, Luke Keele, and Teppei Yamamoto. 2010. “Identifi- 613–44.
cation, Inference, and Sensitivity Analysis for Causal Mediation Reiter, Dan, and Allan C. Stam. 2002. Democracies at War. Princeton,
Effects.” Statistical Science 25(1): 51–71. NJ: Princeton University Press.
864
Risse-Kappen, Thomas. 1995. “Democratic Peace—Warlike Democ- Foreign Policy since Vietnam: Constraining the Colossus. New
racies?” European Journal of International Relations 1(4): 491– York: Oxford University Press.
517. Thucydides. 1954. History of the Peloponnesian War. Transl. Rex
Rosato, Sebastian. 2003. “The Flawed Logic of Democratic Peace Warner. New York: Penguin.
Theory.” American Political Science Review 97(4): 585–602. Tingley, Dustin. 2011. “The Dark Side of the Future: An Experimen-
Rosenau, James N. 1961. Public Opinion and Foreign Policy: An tal Test of Commitment Problems in Bargaining.” International
Operational Formulation. New York: Random House. Studies Quarterly 55(2): 521–44.
Rousseau, David L. 2005. Democracy and War: Institutions, Norms, Tingley, Dustin, and Barbara F. Walter. 2011. “The Effect of
and the Evolution of International Conflict. Stanford, CA: Stanford Repeated Play on Reputation Building: An Experimental Ap-
University Press. proach.” International Organization 65(2): 343–65.
Rummel, R. J. 1979. War, Power, Peace. Vol. 4 of Understanding Tomz, Michael. 2007. “Domestic Audience Costs in International
Conflict and War. Beverly Hills, CA: Sage. Relations: An Experimental Approach.” International Organiza-
Russett, Bruce M. 1990. Controlling the Sword: The Democratic Gov- tion 61(4): 821–40.
ernance of National Security. Cambridge, MA: Harvard University Trager, Robert F., and Lynn Vavreck. 2011. “The Political Costs of
Press. Crisis Bargaining: Presidential Rhetoric and the Role of Party.”
Russett, Bruce M. 1993. Grasping the Democratic Peace: Principles American Journal of Political Science 55(3): 526–45.
for a Post-Cold War World. Princeton, NJ: Princeton University Verba, Sidney, Kay Lehman Schlozman, Henry Brady, and Norman
Press. H. Nie. 1993. “Citizen Activity: Who Participates? What Do They
Russett, Bruce M. 2010. “Capitalism or Democracy? Not So Fast.” Say?” American Political Science Review 87 (2): 303–18.
International Interactions 36(2): 198–205. Welch, David A. 1993. Justice and the Genesis of War. New York:
Russett, Bruce M., and John R. Oneal. 2001. Triangulating Peace: Cambridge University Press.
Democracy, Interdependence, and International Organizations. Wendt, Alexander. 1999. Social Theory of International Politics.
New York: W. W. Norton. Cambridge: Cambridge University Press.
Schultz, Kenneth. 1999. “Do Democratic Institutions Constrain or Wildavsky, Aaron. 1966. “The Two Presidencies.” Society 4(2): 7–14.
Inform?” International Organization 53(2): 233–66. Williams, Michael C. 2001. “The Discipline of the Democratic
Schultz, Kenneth A. 2001. Democracy and Coercive Diplomacy. New Peace.” European Journal of International Relations 7(4): 525–53.
York: Cambridge University Press. Wittkopf, Eugene R. 1987. “Elites and Masses: Another Look at
Slantchev, Branislav. 2005. “Military Coercion in Interstate Crises.” Attitudes toward America’s World Role.” International Studies
American Political Science Review 99(4): 533–47. Quarterly 31(2): 131–59.
Slantchev, Branislav. 2011. Military Threats: The Costs of Coercion Wittkopf, Eugene R., and Michael A. Maggiotto. 1983. “Elites and
and the Price of Peace. New York: Cambridge University Press. Masses: A Comparative Analysis of Attitudes toward America’s
Sobel, Richard. 2001. The Impact of Public Opinion on U.S. World Role.” Journal of Politics 45(2): 303–34.
865

Public Opinion and The Democratic Peace

Uploaded by

Copyright:

Available Formats

Public Opinion and The Democratic Peace

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Public Opinion and The Democratic Peace

Uploaded by

Copyright:

Available Formats

American Political Science Review Vol. 107, No.

Public Opinion and the Democratic Peace

F much attention as the “democratic peace,” the

TABLE 1. The Effect of Democracy on Willingness to Strike

United Kingdom United States United States

Not a democracy 34.2 53.3 50.0

TABLE 2. The Effect of Alliances, Power, TABLE 3. The Effect of Democracy on

United Kingdom United States If the U.S. did not

predicted that the autocracy would build a bomb, but

their chances of success against a democracy, and they

The Effect of the Mediators on Support for a Mediators

takes on its democracy value, and subtracting that

racy, mediator k takes on its democracy value, and

country involves interference in that country’s internal References

You might also like