Journal of Social Issues, Vol. 63, No. 2, 2007, pp.

421--440

Subjective Well-Being and Peace

Ed Diener∗
University of Illinois and the Gallup Organization

William Tov
University of Illinois at Urbana-Champaign

Hierarchical generalized linear modeling was employed to examine the relations

between person-level subjective well-being (SWB) and peace-relevant attitudes,
and how these relations vary across nations in the World Values Survey. Person-
level SWB was associated with more confidence in the government and armed
forces, greater emphasis on postmaterialist values, stronger support for democ-
racy, less intolerance of immigrants and racial groups, and greater willingness to
fight for one’s country. These associations were moderated at the nation level by
liberal development, violent inequality, gross domestic product, and nation-level
SWB. The moderator effects indicate that happy people are not completely blind
to the conditions of their society and that their endorsement of peace attitudes is
sensitive to whether the conditions for peace do exist.

Will a happy citizenry create more peace? Although it is often assumed that
peace establishes the conditions for happiness, we entertain the possibility that
increasing happiness fosters or sustains the conditions for peace. These hypotheses
are not mutually exclusive. Instead, they may play complementary roles in a process
linking the well-being of a nation to the well-being of its people. The concept of
peace has broad appeal because it connects with the human concern for security.
However, as de Rivera (2004) notes, the United Nations’ program of action for a
“culture of peace” likens peace to an ideal society, one in which there is education,
equality, tolerance, and freedom to communicate and participate in governance—
in short, a society that is not merely safe, but flourishing. To this list of bases

∗ Correspondence concerning this article should be addressed to Ed Diener, Department

of Psychology, University of Illinois, 603 E. Daniel Street, Champaign, IL 61820 [e-mail:
[email protected]].
We thank Carol Nickerson, Evelyn Au, and several reviewers for providing us with helpful com-
ments. This work was supported by a National Science Foundation Graduate Fellowship awarded to
William Tov.
421

C 2007 The Society for the Psychological Study of Social Issues
422 Diener and Tov

for a culture of peace, we propose an essential addition: the subjective well-being

(SWB) of the citizenry.
Our use of the term SWB refers to a stable, overall sense of well-being and to
emotional and cognitive components that are relatively enduring rather than mo-
mentary. The emotional component includes how frequently one experiences posi-
tive emotions like happiness, whereas the cognitive component includes judgments
of one’s satisfaction with life. The term “happiness” can refer either to momentary
positive feelings or to long-term well-being. However, for ease of exposition, we
will sometimes use the term “happy people” primarily to refer to individuals who
have a stable, long-term sense of well-being and are not simply in a temporary
positive state. In this article, we consider the relation between individual SWB and
attitudes that may be relevant to various conceptions of peace. We refer to these
attitudes (such as confidence in government) as peace-relevant attitudes or simply,
peace attitudes. We do not mean to imply that these attitudes cohere tightly within
individuals as part of an overall “peaceful personality.” Rather, our interest in these
attitudes is in their implications for a culture of peace. In such a society, depending
on how one conceives it, individuals might endorse some of these attitudes and not
others. While SWB concerns individuals’ own sense of wellness, peace attitudes
concern how individuals perceive the sociopolitical climate of their society.

Peace Attitudes Arising from SWB

At the individual level, SWB may foster peace attitudes by influencing the
way people perceive and relate to others. A large body of psychological research
suggests that positive emotions predispose individuals to trust and cooperate with
others, and to engage in prosocial behaviors (for reviews, see Lyubomirsky, King,
& Diener, 2005; Tov & Diener, in press). Much of this research is experimental,
revealing that people who are induced to feel positive moods are more likely than
those who are not to trust and help others, to display greater liking toward others,
and to prefer cooperative strategies. In contrast, when people perceive threat or ex-
perience fear and anger, they are more likely to endorse punitive measures (Rucker,
Polifroni, Tetlock, & Scott, 2004) and are less politically tolerant (Skitka, Bauman,
& Mullen, 2004; see Conejeros and Etxebarrı́a this issue). Because prolonged pe-
riods of anger and anxiety in a society can lead to instability, it is important for a
population to recover from events that threaten security, and positive emotions have
been found to predict resilience and recovery in such cases (e.g., the September
11th attacks; Fredrickson, Tugade, Waugh, & Larkin, 2003).
Although the preceding research has emphasized momentary positive emo-
tions, similar effects have been observed among individuals who are generally
happy and satisfied with their lives (Lyubomirsky et al., 2005). Taken together,
this literature suggests several possible links between SWB and peace attitudes.
For instance, participatory forms of government require individuals to trust that
their fellow citizens will not abuse civil and political liberties that are granted.
Peace and SWB 423

Thus, by facilitating trust and cooperation, SWB may have important implications
for tolerance, as well as support for democracy and individual freedom. If individ-
ual well-being is linked to various peace attitudes, this would support the notion
of SWB as a basis for a culture of peace.

Peace, Prosperity, and SWB at the Nation Level

Our investigation of peace attitudes concerns subjective perceptions of society

—that is, peace in the minds of the people. However, peace can also apply to objec-
tive conditions in a society. Previously, de Rivera (2004) conducted a factor anal-
ysis on societal indicators relevant to cultures of peace and arrived at four “peace
factors”: liberal development (an indicator of economic strength and democratic
institutions), violent inequality (which reflects homicide rates and unequal income
distribution), violent means (the extent of military spending and use), and nurtu-
rance (which includes tolerance and education spending). Using a multinational
sample, we employ hierarchical generalized linear modeling (HGLM) to explore
how these nation-level peace factors relate to (1) overall levels of peace attitudes
in a society, and (2) how strongly these attitudes relate to SWB at the level of the
person. The first question asks how prevalent these attitudes are in various types
of nations; the second question asks how likely happy people are to endorse an
attitude if they are in one type of nation versus another.
Although gross domestic product (GDP) per capita is a component of the
liberal development factor (de Rivera, 2004), we also examine specifically how
national wealth relates to peace attitudes. Insofar as wealth enables people to
meet basic needs, high GDP per capita may imply that a society is relatively
stable and secure. However, national wealth is often associated with greater po-
litical freedom, equality, and greater societal levels of SWB (Diener, Diener, &
Diener, 1995). These conditions overlap greatly and are difficult to disentangle.
Are people happier simply because they have met their basic needs? Or are people
happier because other correlates of development (e.g., freedom and equality) pro-
vide an opportunity structure for individuals to pursue their goals and aspirations?
It seems unlikely that increasing wealth alone would ensure peace in the face of
large income disparities and growing discontentment. The direction of influence
may work both ways: rising prosperity can increase SWB, but rising SWB can
influence productivity (Diener & Seligman, 2004; Lyubomirsky et al., 2005) and
nurture economic development and political stability (Inglehart & Klingemann,
2000).
Thus, in the interest of developing a culture of peace, factors other than eco-
nomic indicators are necessary. (Diener, 2000; Diener & Tov, in press) argued
that societal indicators of SWB be used with economic measures to inform policy
decisions. To further discuss this, we examine how GDP and nation-level SWB
relate to peace attitudes. Although the two are strongly correlated, they may not
always have the same consequences for a culture of peace. The extent to which
424 Diener and Tov

they do or do not, has important implications for the utility of subjective measures
in supplementing economic indicators. Nation-level SWB is also of interest as
a reflection of the emotional climate within a country (Basabe, Paez, Valencia,
Rimé, & Diener, 2002) and is therefore in keeping with the theme of this special
issue.
Method

We obtained data from the 1995 and 1999/2000 waves of the World Values
Survey (WVS; World Values Survey Organization, 2005). Data from 28 societies
could not be included in our analyses either because peace factor scores were not
available, or because too many cases were missing from the variables of interest.
The final data set consisted of 51,929 adults (age 18 or older) from 51 nations (see
Appendix). Although few African nations are represented, the sample still varied
greatly in terms of geography, economics, and political structure, enabling us to
explore the crossnational effects of GDP, SWB, and the peace factors on peace
attitudes.

Measures

Person-Level Criterion Variables.

Confidence in Parliament, Civil Service, and the Armed Forces. The con-
fidence that people have in sociopolitical institutions helps to sustain peace in a
society. Therefore, we examined confidence in parliament and civil service. Re-
spondents indicated their level of confidence on a four-point scale with 1 = “a
great deal,” 2 = “quite a lot,” 3 = “not very much,” and 4 = “none at all.” Because
peacefulness is commonly associated with nonviolence, we also considered con-
fidence in the armed forces. One might expect peace advocates to downplay the
importance of military might. On the other hand, confidence in the armed forces
might also reflect an overall sense of security. For the purposes of calculating cu-
mulative probabilities (see below), we considered responses of 2 or less to reflect
confidence in a given institution.

Endorsement of Army Rule, Autocracy, and Democracy. The WVS assessed

respondents’ attitudes toward rule by the military (army rule), rule by a strong
all-powerful leader (autocracy), and rule by democracy. Responses were made on
a four-point scale with 1 = “very good,” 2 = “fairly good,” 3 = “fairly bad,” and
4 = “very bad.” For army rule and autocracy, responses of 2 or less were counted
as endorsement of these political systems. By the same criterion, democracy was
endorsed by the vast majority of respondents (92%). Therefore, our results focus
only on those who fully endorsed democracy (i.e., a response of 1). If a culture
of peace supports freedom and participation (de Rivera, 2004), then individuals in
such societies should endorse democracy and oppose restrictive forms of govern-
ment such as army rule or autocracy.
Peace and SWB 425

Postmaterialist Values. According to Inglehart (2000), greater economic de-

velopment leads to a shift in people’s concerns from basic needs and security
(materialist values) to having greater freedom and control over one’s life (post-
materialist values). Postmaterialist values seem to accord with cultures of peace,
where people are both secure and free to focus on their individual goals. In the
WVS, respondents were asked which of the following four should be their na-
tion’s first and second priorities: giving people more say in government decisions;
protecting freedom of speech; maintaining order in the nation; or fighting rising
prices. The first two goals reflect postmaterialist concerns whereas the latter two
reflect materialist concerns. Respondents who selected two postmaterialist goals
were coded as “1”; those who selected only one were coded as “2”; and those who
did not select any postmaterialist goals were coded as “3.”

Racial Intolerance. Respondents indicated whether they objected to living

near people who were of a different race or immigrants/foreign workers. As
a measure of racial and ethnic intolerance, respondents who objected to living
near both groups were coded as “1,” those who objected to living near one of
the groups were coded as “2,” and those who did not object to living near ei-
ther were coded as “3.” Responses of 2 or less were counted as reflecting racial
intolerance.

Restrictions on Immigration. Attitudes toward immigration policy were mea-

sured on a four-point scale with the following value labels: 1 = “Prohibit people
coming here from other countries”; 2 = “Place strict limits on the number of
foreigners who can come here”; 3 = “Let people come as long as there are jobs
available”; and 4 = “Let anyone come who wants to.” Thus, lower numbers reflect
a preference for greater restrictions on immigration. Responses of 2 or less were
counted as having a restrictive attitude toward immigration.

Willingness to Fight for Country. As a measure of attitudes toward the use

of violence, we examined respondents’ willingness to fight for their country if it
went to war (1 = yes, 0 = no). However, because of the framing of this question,
responses could be interpreted more specifically as an acceptance of violence only
as a last resort. Also, data were only available for 36 countries (see Appendix).
Despite its limitations, examining the willingness to fight could refine the “culture
of peace” construct and its implications for individual attitudes.

Person-Level Predictor: Person-Level SWB. As an indicator of overall SWB,

we averaged two questions regarding how happy respondents generally felt and
how satisfied they were with their lives. Scores ranged from 1 (not at all happy
and dissatisfied) to 7 (very happy and satisfied). In our HGLM analyses, each
respondent’s SWB score was used as a predictor at the person-level of analysis.
426 Diener and Tov

Nation-Level Predictors.

Peace Factor Scores. We used each nation’s factor scores on the four peace
factors (liberal development, violent inequality, violent means, and nurturance; de
Rivera, 2004) as predictors of between-nation differences in peace attitudes.
Gross Domestic Product Per Capita. We obtained per capita gross domestic
product (GDP) data from the Penn World Tables (Heston, Summers, & Aten,
2002) in constant 1996 dollars. Because of the wide variance in GDP, all analyses
employed the natural log of GDP per capita (log GDP).
Nation-Level SWB. For each nation, mean SWB was computed by averaging
across all respondents within the same nation.

Data Analytic Strategy

Because all peace attitudes were measured using at most four response cat-
egories, the data were often skewed and random effects were not normally dis-
tributed. In such cases, ordinal logistic regression models are preferable. Treating
the criterion variables as ordinal requires a nonlinear approach (HGLM; however,
see Basabe and Valencia in this issue for a complementary analysis using col-
lective level correlations of logged variables). Instead of predicting a raw score
(Ŷ ij ), ordinal HGLM (Raudenbush & Bryk, 2002) generates a regression equation
that can be used to predict the log-odds (ηij ) of giving a certain response. Log-
odds can then be converted into probabilities via the following formula: exp(ηij )/[1
+ exp(ηij )]. In the present study, we will often present cumulative log-odds and
hence, cumulative probabilities (e.g., the probability that a respondent will select a
2 or lower on a four-point scale). At the person-level, the log-odds that person i in
country j will endorse a peace attitude is predicted from his or her level of SWB:
Person−Level Model : ηij = β0 j + β1 j (SWBij − NSWB j )
where β 1j represents the effect of SWB on the likelihood that person i in nation
j will endorse an attitude. Each person’s score (SWBij ) is centered on his or her
nation’s average (NSWBj ) so that the intercept (β 0j ) represents the average log-
odds that people in nation j endorse a given attitude.
Preliminary tests of random-coefficients models (Raudenbush & Bryk, 2002)
indicated that the 51 nations in our sample varied significantly in terms of the
average log-odds of endorsing peace attitudes (i.e., the size of the intercept β 0j ), as
well as how strongly person-level SWB was associated with log-odds of endorse-
ment (i.e., the size of the slope β 1j ). Therefore, we sought to determine whether
nation-level characteristics were associated with crossnational differences in the
intercept and slope, as presented in the following equations:
Nation−Level Intercept Model: β0j = γ00 + γ01 W1 j + u 0 j
Nation−Level Slope Model: β1 j = γ10 + γ11 W1 j + u 1 j
Peace and SWB 427

In the nation-level intercept model, γ 00 is the grand intercept (average log-odds of

endorsement across all 51 nations), and γ 01 represents the effect of a nation-level
predictor W 1j (e.g., GDP) on the size of the intercept (β 0j ) in nation j. In the
nation-level slope model, γ 10 is the grand slope (average effect of person-level
SWB on log-odds of endorsement across all 51 nations), and γ 11 is a moderator
effect (the effect of a nation-level predictor (W 1j ) on the size of the slope (β 1j )
in nation j). All nation-level predictors were grand-mean centered before entering
them into the models. Finally, u0j and u1j are residuals representing variance in
the intercept and slope, respectively, that remain unaccounted for by nation-level
predictors.

Results

Hierarchical Models Using Peace Factors

Interpreting the Results of HGLM Analyses. We explored the relation between

nation-level peace factors, person-level SWB, and person-level peace attitudes us-
ing four types of hierarchical models. For each model, a peace factor score (e.g.,
liberal development) was used as the nation-level predictor (W 1j ). These models
(Models 1–4) are presented in Table 1 with the unit-specific estimates of the grand
intercept and unstandardized/standardized coefficients (see Raudenbush & Bryk,
2002, for a discussion of unit-specific estimates). As an example, take Model 1
in which liberal development is used as a nation-level predictor. In the first col-
umn, we see that γ 01 = −.18 (p < .10), indicating that liberal development is
associated with a lower log-odds (at the nation-level) of being confident in civil
service. This suggests that societal levels of confidence tend to be lower when
liberal development is high. We can compute the average log-odds of confidence
when liberal development is high (e.g., liberal development factor score = 1) as
β 0j = γ 00 + γ 01 W 1j = −.25 + (−.18)(1) = −.43. Thus, in a country where lib-
eral development is high, the overall probability of confidence in civil service is
equal to exp(−.43)/[1 + exp(−.43)] = .39, or 39%. In Table 1, we also see that on
average, person-level SWB is associated with greater log-odds of confidence in
civil service (γ 10 = .12). However, the effect of liberal development on the slope
(γ 11 = −.00) is not significant. This suggests that the strength of association be-
tween person-level SWB and confidence in civil service (β 1j ) does not vary across
nations as a function of how liberally developed they are. Nevertheless, for illus-
trative purposes, a unit increase in person-level SWB in a nation where liberal de-
velopment is high leads to an increase of β 1j = γ 10 + γ 11 W 1j = .12 + (−.00)(1) =
.12 in the log-odds of confidence in civil service. Thus, the total log-odds that a
“happy person” in a liberally developed nation will be confident in civil service
is equal to ηij = β 0j + β 1j (SWBij – NSWBj ) = −.43 + (.12)(1) = −.31. This
translates into a probability of exp(−.31)/[1 + exp(−.31)] = .42, or 42%.
 428

Table 1. Hierarchical Models Predicting Attitudes from Individual-Level SWB and Nation-Level Peace Factor Scores
Peace Attitude
Confidence in Confidence in Confidence in Army Rule Autocracy
Civil Service Parliament Armed Forces Endorsement Endorsement
Model 1
Intercept, γ 00 −.25 −.44 .54 −2.06 −.85
Liberal development, γ 01 −.18(−.24)† −.12(−.19) −.36(−.41)∗∗ −.61(−.62)∗∗∗ −.21(−.21)
Person-level SWB slope, γ 10 .12(.17)∗∗∗ .11(.17)∗∗∗ .09(.13)∗∗∗ −.02(−.03) −.02(−.03)†
Liberal development, γ 11 −.00(.00) .01(.02) .02(.03)† −.02(−.03) −.03(−.04)∗
Model 2
Intercept, γ 00 −.25 −.44 .54 −2.06 −.85
Violent inequality, γ 01 −.19(−.18) −.20(−.21) −.14(−.14) .45(.58)∗ .41(.45)∗∗
Person-level SWB slope, γ 10 .12(.17)∗∗∗ .11(.17)∗∗∗ .09(.13)∗∗∗ −.02(−.03) −.02(−.03)
Violent inequality, γ 11 −.03(−.05)∗ −.04(−.07)∗∗ −.02(−.04) .02(.03) −.00(−.00)
Model 3
Intercept, γ 00 −.25 −.44 .54 −2.06 −.85
Violent means, γ 01 .14(.18) .10(.11) .48(.53)∗∗∗ .19(.20) .09(.11)
Person-level SWB slope, γ 10 .12(.17)∗∗∗ .11(.17)∗∗∗ .09(.13)∗∗∗ −.02(−.03) −.02(−.03)
Violent means, γ 11 .02(.02) .01(.02) −.01(−.01) .02(.02) .01(.00)
Model 4
Intercept, γ 00 −.25 −.44 .54 −2.06 −.85
Nurturance, γ 01 −.00(−.01) .00(−.01) −.01(−.09) −.24(−.24)† −.18(−.18)†
Person-level SWB slope, γ 10 .12(.17)∗∗∗ .11(.17)∗∗∗ .09(.13)∗∗∗ −.02(−.03) −.02(−.03)
Nurturance, γ 11 −.01(−.02) −.03(.04)∗ −.00(.01) .01(.01) .00(.01)
(Continued)
Diener and Tov
 Table 1. Continued
Peace Attitude
Peace and SWB

Democracy Postmaterial Racially Restrict Fight for

Endorsement Values Intolerant Immigration Country
Model 1
Intercept, γ 00 .01 −1.91 −1.48 −.14 1.26
Liberal development, γ 01 .01(−.02) .43(.45)∗∗∗ −.57(−.60)∗∗∗ .01(.01) −.46(−.49)∗∗
Person-Level SWB Slope, γ 10 .05(.09)∗∗∗ .03(.05)∗∗ −.09(−.13)∗∗∗ −.07(−.09)∗∗∗ .09(.15)∗∗∗
Liberal development, γ 11 .03(.05)∗∗ −.01(.01) −.06(−.10)∗∗∗ .02(.03) .01(−.01)
MODEL 2
Intercept, γ 00 .01 −1.91 −1.48 −.14 1.26
Violent inequality, γ 01 −.33(−.33)∗∗ −.07(.06) −.30(−.25)† −.08(−.06) −.18(−.17)
Person-level SWB slope, γ 10 .05(.09)∗∗ .03(.05)∗∗ −.09(−.13)∗∗∗ −.07(−.09)∗∗∗ .09(.15)∗∗∗
Violent inequality, γ 11 .00(−.01) −.00(.00) .02(.03) .01(.01) .00(−.01)
MODEL 3
Intercept, γ 00 .01 −1.91 −1.48 −.14 1.26
Violent Means, γ 01 −.02(−.01) −.13(−.11) .07(.07) .04(.04) .02(.00)
Person-level SWB slope, γ 10 .05(.09)∗∗ .03(.05)∗∗ −.09(−.13)∗∗∗ −.07(−.09)∗∗∗ .09(.14)∗∗∗
Violent means, γ 11 .01(.02) .00(.01) .02(.03) −.01(−.01) .03(.04)†
MODEL 4
Intercept, γ 00 .01 −1.91 −1.48 −.14 1.26
Nurturance, γ 01 .11(.11) .15(.16)† −.20(−.21) .08(.09) −.06(−.04)
Person-level SWB slope, γ 10 .05(.09)∗∗ .03(.05)∗∗ −.09(−.13)∗∗∗ −.07(−.09)∗∗∗ .09(.15)∗∗∗
Nurturance, γ 11 −.01(−.02) −.01(−.01) −.02(−.03) .01(.02) −.01(−.04)
Note. Coefficients outside parentheses are unstandardized; coefficients inside parentheses are standardized. All intercepts represent cumulative log-odds
except for endorsement of democracy and fight for country.
†
p < .10. ∗ p < .05. ∗∗ p < .01. ∗∗∗ p < .001.
429
430 Diener and Tov

Person-level SWB Effects. The relation of person-level SWB to person-level

peace attitudes are represented by the γ 10 coefficients, which are the average effects
of person-level SWB across the 51 nations. Notice that for each peace attitude, the
estimates of γ 10 remain the same even as we enter different nation-level predictors
(Table 1). This is to be expected due to the large sample size as well as the way
hierarchical modeling partitions the total variance into person- and nation-level
variances (Raudenbush & Bryk, 2002).
On average, person-level SWB was associated with a greater likelihood of
being confident in parliament and civil service, endorsing democracy and post-
materialist values, and a lower likelihood of being racially intolerant or having
a restrictive attitude toward immigration. However, person-level SWB was also
associated with greater confidence in the armed forces and willingness to fight for
one’s country, and was not significantly related to the endorsement of army rule or
autocracy. On balance, happy people appear to have several important peace atti-
tudes, but how peaceful they are depends on which aspect of peace is emphasized.
In terms of valuing freedom and trusting others, person-level SWB is a frequent
correlate and may be an important cause (Tov & Diener, in press). However, with
regards to nonviolence, happy people appear to feel that violence is justified in
certain situations (e.g., defending one’s country). These findings imply that raising
individual well-being does not completely diminish the value that one places on
security. Moreover, the relation between person-level SWB and peace attitudes
is moderated by nation-level variables such as liberal development. Next, we ex-
amine the effects of nation-level predictors on the overall level of peace attitudes
in society (γ 01 ) and on the relation between person-level SWB and person-level
attitudes (γ 11 ).

Model 1: Liberal Development. The significantly negative γ 01 coefficients in

Table 1 suggest that nation-level liberal development is associated with lower so-
cietal levels of confidence in armed forces, endorsement of army rule, and willing-
ness to fight a war. One interpretation of these results is that nonviolent ideologies
are more prevalent in liberal societies. Liberal development was also associated
with lower prevalence of racial intolerance and greater postmaterialist values at
the societal level. These findings are reminiscent of past work on the authoritarian
personality (Adorno, Frenkel-Brunswik, Levinson, & Sanford, 1950). However, it
is important to note that this peace factor was not associated with endorsement of
democracy.
The significant γ 11 coefficients indicate that nation-level liberal development
moderated how strongly person-level SWB correlated with racial intolerance and
endorsement of autocracy and democracy. For endorsement of democracy, both
γ 10 and γ 11 are positive suggesting that a unit-increase in person-level SWB has
a greater positive effect on support for democracy in countries with higher nation-
level scores on liberal development. For racial intolerance and endorsement of
Peace and SWB 431

autocracy, both γ 10 and γ 11 are negative suggesting that a unit-increase in person-

level SWB has a greater negative effect on these attitudes in countries with higher
nation-level scores on liberal development.
Another way to conceptualize these moderator effects (also known as cross
level interactions) is to use the γ coefficients to predict the log-odds of endorsing
these attitudes, and then convert these log-odds into probabilities (Table 3). To
examine the effect of a nation-level variable, we computed the probability of
endorsement in countries that were high or low on that variable. For instance, the
effect of nation-level liberal development was assessed at values that were one
standard deviation above (“high”) and below (“low”) the grand mean. Similarly,
person-level SWB was assessed at values that were one SD above (“happy person”)
and below (“unhappy person”) the mean of person-level SWB (which was zero
because all scores were group-centered).
Table 3 shows that when nation-level liberal development is low, person-level
SWB is not associated with racial intolerance or attitudes toward democracy and
autocracy. However, when liberal development is high, person-level SWB corre-
lates negatively with racial intolerance and endorsement of autocracy, and pos-
itively with endorsement of democracy. To illustrate these effects with concrete
examples, we can compare the correlations within specific countries (after reverse
coding the items to ease interpretation). In China (about −1 SD on liberal devel-
opment), the correlations of person-level SWB with racial intolerance, support for
democracy, and support for autocracy were all nonsignificant (all r’s around .01).
However, in New Zealand (about +1 SD on liberal development), person-level
SWB correlated significantly with racial intolerance and support for democracy
(r’s = −.11 and .10, respectively, p’s < .02), though not significantly with support
for autocracy (r = −.02). Thus, variations among liberally developed nations do
exist. Nevertheless, on average, happy people in these societies tend to be more
supportive of democracy, less supportive of autocracy, and less likely to be racially
intolerant than unhappy people. These findings suggest that full support of democ-
racy may be contingent upon individual well-being in liberally developed nations,
and are consistent with the idea that SWB plays a critical role in legitimizing lib-
eral democratic governments (Inglehart & Klingemann, 2000). Because liberally
developed nations tend to be democratic, unhappy people in these countries may
feel disillusioned. Individuals who are unhappy because they cannot meet even
the most basic needs may feel a discrepancy between their personal experiences
and the ideology of freedom and equality espoused by democratic regimes. Conse-
quently, these individuals may become more critical and less likely to fully endorse
democracy. This issue is quite complex and merits further research.

Model 2: Violent Inequality. The γ 01 coefficients indicate that nation-level vi-

olent inequality was associated with greater societal levels of support for army rule
and autocracy, and lower levels of support for democracy. Thus, in nations marred
432 Diener and Tov

by high rates of homicide and inequality, increasing segments of the population

favor strong leadership or military rule.
Violent inequality also moderated (γ 11 ) how strongly person-level SWB cor-
related with confidence in parliament and civil service. Notice that whereas the
effect of person-level SWB (γ 10 ) on confidence is positive, the moderating effect
of violent inequality (γ 11 ) is negative. This suggests that a unit-increase in person-
level SWB has a less positive effect on confidence in countries with higher nation-
level scores on violent inequality. In other words, person-level SWB correlates
more strongly with confidence in parliament and civil service when nation-level
violent inequality is low than when it is high. As shown in Table 3, happy people
are increasingly more likely than unhappy people to be confident in governance
among societies that are low on violent inequality. As an example, we compare
Belgium (low in violent inequality) with Argentina (high in violent inequality). In
Belgium, person-level SWB correlated .13 and .09 (p’s < .0001) with confidence
in parliament and civil services, respectively. However, in Argentina, the same
correlations were smaller and nonsignificant: .05 and .01.

Model 3: Violent Means. Nation-level violent means was associated with

greater societal levels of confidence in the armed forces. No other significant
effects were observed.

Model 4: Nurturance. Nation-level nurturance was not significantly associ-

ated with societal levels of peace attitudes. Only one significant moderator effect
was observed for confidence in parliament (see Tables 1 and 3). Person-level SWB
correlates more strongly with confidence in parliament when nation-level nurtu-
rance is low than when it is high. As nurturance itself was not significantly related
to a nation’s average log-odds of confidence in parliament, this finding is difficult
to interpret and we refrain from offering possible explanations.

Hierarchical Models Comparing GDP and SWB

At the nation-level, SWB correlated strongly with log GDP, r(51) = .75,
p < .01. Recall that GDP is also a component of the liberal development factor
score (de Rivera, 2004). Indeed, in our sample we found that liberal development
was the only peace factor to correlate with nation-level SWB, r(51) = .66, p <
.01. Therefore, it was of interest to consider GDP separately and in conjunction
with nation-level SWB. In comparing how GDP per capita and nation-level SWB
are associated with peace attitudes, we examined three models. Models 5 and 6
employ GDP and SWB separately as nation-level predictors. Model 7 employs
both GDP and SWB as nation-level predictors. All models included person-level
SWB, enabling us to explore the moderating effects of nation-level SWB and GDP.
Table 2 presents the results of these three models.
 Table 2. Hierarchical Models Predicting Attitudes from Person-Level SWB and Nation-Level SWB and GDP
Peace and SWB

Peace Attitude
Confidence in Confidence in Confidence in Army Rule Autocracy
Civil Service Parliament Armed Forces Endorsement Endorsement
Model 5
Intercept, γ 00 −.25 −.44 .54 −2.06 −.85
GDP per capita, γ 01 −.16(−.14) −.17(−.14) −.47(−.38)∗∗ −.85(−.70)∗∗∗ −.39(−.33)∗∗
Person-level SWB slope, γ 10 .12(.17)∗∗∗ .11(.17)∗∗∗ .10(.13)∗∗∗ −.02(−.03) −.02(−.03)
GDP per capita, γ 11 −.00(−.00) .00(.01) .04(.04)∗∗ −.02(−.02) −.03(−.03)†
Model 6
Intercept, γ 00 −.25 −.44 .54 −2.06 −.85
Nation-level SWB, γ 01 −.00(.00) .02(.02) −.48(−.32)∗∗ −.62(−.42)∗∗ −.45(−.31)∗∗
Person-level SWB slope, γ 10 .12(.17)∗∗∗ .11(.17)∗∗∗ .09(.13)∗∗∗ −.02(−.03) −.02(−.03)
Nation-level SWB, γ 11 −.02(−.02) −.01(−.01) .05(.05)∗∗ −.01(−.01) −.02(−.01)
Model 7
Intercept, γ 00 −.25 −.44 .54 −2.06 −.85
GDP per capita, γ 01 −.36(−.31)† −.43(−.36)† −.40(−.33)† −1.08(−.88)∗∗∗ −.26(−.22)
Nation-level SWB, γ 02 .33(.23) .41(.29) −.11(−.07) .37(.23) −.21(−.14)
Person-level SWB slope, γ 10 .12(.17)∗∗∗ .11(.17)∗∗∗ .09(.13)∗∗∗ −.02(−.03) −.02(−.03)
GDP per capita, γ 11 .02(.03) .02(.02) .01(.01) −.03(−.03) −.03(−.04)
Nation-level SWB, γ 12 −.04(−.04) −.02(−.03) .04(.05)† .02(.02) .00(.02)
(Continued)
433
 434

Table 2. Continued
Peace Attitude
Democray Postmaterial Racially Restrict Fight for
Endorsement Values Intolerant Immigration Country
Model 5
Intercept, γ 00 .01 −1.91 −1.48 −.14 1.26
GDP per capita, γ 01 .09(.08) .61(.51)∗∗∗ −.69(−.58)∗∗∗ −.00(−.01) −.57(−.44)∗∗
Person-level SWB slope, γ 10 .05(.09)∗∗∗ .03(.05)∗∗ −.09(−.13)∗∗∗ −.07(−.09)∗∗∗ .09(.15)∗∗∗
GDP per capita, γ 11 .03(.03)† −.02(−.02) −.09(−.11)∗∗∗ .04(.04)∗ .01(−.01)
Model 6
Intercept, γ 00 .01 −1.91 −1.48 −.14 1.26
Nation-level SWB, γ 01 .25(.17)† .76(.51)∗∗∗ −.70(−.48)∗∗∗ −.02(−.02) −.34(−.18)
Person-level SWB slope, γ 10 .05(.08)∗∗ .03(.05)∗∗ −.09(−.13)∗∗∗ −.06(−.09)∗∗∗ .09(.14)∗∗∗
Nation-level SWB, γ 11 −.00(−.01) −.04(−.04)∗ −.07(−.07)∗ .06(.06)∗∗∗ −.01(−.03)
Model 7
Intercept, γ 00 .01 −1.91 −1.48 −.14 1.26
GDP per capita, γ 01 −.14(−.12) .33(.28)∗∗ −.61(−.51)∗∗ .02(.02) −.75(−.61)∗∗
Nation-level SWB, γ 02 .38(.26)† .45(.31)∗∗ −.14(−.09) −.04(−.04) .30(.25)
Person-level SWB slope, γ 10 .05(.08)∗∗∗ .03(.05)∗∗ −.09(−.13)∗∗∗ −.06(−.09)∗∗∗ .09(.15)∗∗∗
GDP per capita, γ 11 .07(.09)∗∗ .01(.01) −.10(−.12)∗∗ −.01(−.00) .03(.02)
Nation-level SWB, γ 12 −.07(−.08)∗∗ −.05(−.05)† .02(.02) .07(.07)∗∗ −.03(−.04)
Note. Coefficients outside parentheses are unstandardized; coefficients inside parentheses are standardized. All intercepts represent cumulative log-odds
except for endorsement of democracy and fight for country.
†
p < .10. ∗ p < .05. ∗∗ p < .01. ∗∗∗ p < .001.
Diener and Tov
Peace and SWB 435

Table 3. Probability of Having Various Peace Attitudes as a Function of Person-Level SWB and
Nation-Level Moderators
Person-Level SWBb
Unhappy Happy
Criterion Variable Nation-Level Moderator Levela Person Person
Confidence in parliament Violent inequality High 33 38
Low 39 48
Nurturance High 37 42
Low 35 44
Confidence in civil service Violent inequality High 37 43
Low 43 52
Autocracy endorsement Liberal development High 27 24
Low 35 35
Democracy endorsementc Liberal development High 48 53
Low 50 51
GDP (Model 7) High 44 51
Low 54 53
SWB (Model 7) High 57 57
Low 41 47
Racially intolerant Liberal development High 13 9
Low 30 29
GDP (Model 7) High 15 10
Low 28 27
Restrict immigration SWB (Model 7) High 46 45
Low 51 44
Note. Except where noted, values represent the cumulative probability of giving an affirmative
response.
a
High levels represent countries that were 1 SD above the grand mean of the nation-level predictor.
Low levels countries that were 1 SD below the mean.
b
Unhappy persons were 1 SD below the grand mean of person-level SWB. Happy persons were 1 SD
above the grand mean.
c
Values represent the probability of giving the strongest endorsement (1 = very good).

Model 5 (GDP Only) and Model 6 (Nation-Level SWB Only). A comparison

of Models 5 and 6 reveals that GDP and SWB have significant effects on the
intercept (γ 01 ) for five peace attitudes, and significant effects on the person-level
SWB slope (γ 11 ) for three peace attitudes (Table 2). In all eight cases, the effects
are in the same direction, suggesting a considerable overlap in the implications of
GDP and nation-level SWB for peace attitudes. Both nation-level variables were
associated with greater societal levels of postmaterialist values, and lower societal
levels of confidence in armed forces, support for army rule and autocracy, and
racial intolerance. In addition, GDP but not nation-level SWB was associated with
a lower overall willingness in society to fight for one’s country. We will postpone
discussion of moderator effects for Model 7, in which the effects of GDP and
nation-level SWB are controlled for each other.
436 Diener and Tov

Model 7: GDP and Nation-Level SWB Entered Simultaneously. In Model 7,

we interpret γ 01 and γ 11 as the effect of GDP on the intercept and slope, respec-
tively. Because Model 7 contains two nation-level predictors, we introduce γ 02 and
γ 12 as the effect of nation-level SWB on the intercept and slope, respectively. After
controlling for both nation-level predictors, GDP (γ 01 ) but not nation-level SWB
(γ 02 ) was associated with lower societal levels of racial intolerance and support
for army rule. The effects of GDP mirror those obtained for liberal development,
suggesting that sociopolitical norms heavily influence societal attitudes. Wealthy,
democratic nations also tend to be multicultural societies, making the issue of di-
versity and racial tolerance more salient. Interestingly, GDP also moderated (γ 11 )
how strongly person-level SWB correlated with racial intolerance. When GDP
is low, person-level SWB does not correlate with racial intolerance. When GDP
is high, person-level SWB tends to correlate negatively with racial intolerance.
Thus, in wealthy countries, happy people are less likely than unhappy people to be
racially intolerant (Table 3). Recall that liberal development also showed a similar
moderating effect (not surprising given that GDP was one component of this peace
factor).
However, neither GDP nor nation-level SWB predicted greater societal levels
of confidence in parliament or civil service. If anything, the effect of GDP on
confidence in governance is in the negative direction. GDP (γ 01 ) was also not
associated with support for democracy at the nation level. In contrast, nation-level
SWB (γ 02 ) was associated with greater societal levels of support for democracy,
although the effect did not reach significance (p = .07). Our failure to obtain
significance may be due to insufficient power to detect these effects at the nation
level. At the person-level, the sample size is large and SWB remained significantly
associated with confidence in governance and endorsement of democracy.
Nevertheless, GDP (γ 11 ) and nation-level SWB (γ 12 ) moderated how strongly
person-level SWB related to endorsement of democracy. When GDP is low, person-
level SWB did not correlate with endorsement of democracy. When GDP is high,
person-level SWB correlated positively with endorsement. As we observed with
liberal development, this suggests that in wealthy democratic societies, support
for democracy is contingent on individual well-being. In contrast, when nation-
level SWB is high, person-level SWB does not correlate with endorsement of
democracy. For instance, in Denmark (high nation-level SWB), person-level SWB
correlated .04 with support for democracy. Notice however, that overall endorse-
ment of democracy tends to be higher in societies that are high on nation-level SWB
(Table 3). When nation-level SWB is low, person-level SWB correlates positively
with democratic endorsement. Interestingly, the moderating effect of nation-level
SWB remained even after controlling for the variance of person-level SWB in a
country (which tends to be smaller where nation-level SWB is high). It is note-
worthy that of the ten countries with the lowest nation-level SWB in our sample,
seven were former Communist societies (e.g. Albania, Ukraine, Russia, Belarus,
Peace and SWB 437

Romania, Bulgaria, and Lithuania). Of these, the correlation between person-level

SWB and support for democracy was significantly positive in Ukraine, Russia,
Bulgaria, and Lithuania (average r = .17, p’s ≤ .001). Given that democratic
changes in these nations have occurred within the past two decades, these findings
raise further questions about the causal link between democracy and SWB (e.g.,
Inglehart & Klingemann, 2000) that should be pursued in future research.
Interestingly, the level of postmaterialist values in a society was predicted by
both GDP (γ 01 ) and nation-level SWB (γ 02 ). In wealthier countries, people place
more value on freedom of speech and having more say in government. Beyond
the effects of national wealth, greater societal SWB also contributes to increasing
postmaterialist concerns at the nation-level. Notice that nation-level SWB pre-
dicts postmaterialist values even after controlling for person-level SWB. These
“compositional effects” suggest that when levels of SWB are high throughout a
society, both happy and unhappy people are increasingly likely to prioritize civil
and political freedom over economic stability or maintaining social order. These
compositional effects are net of GDP and may reflect emotional norms or emo-
tional climates (Basabe et al., 2002).
Finally, nation-level SWB significantly moderated the association between
person-level SWB and restrictive immigration attitudes. Person-level SWB corre-
lates more strongly with restrictive attitudes when nation-level SWB is low than
when it is high. The results in Table 3 suggest that an unhappy person in an “un-
happy country” is increasingly more likely to endorse restrictions on immigration
than is a happy person in the same country.
In sum, both GDP and nation-level SWB have important implications for
peace attitudes. Given that these two predictors are so strongly correlated with
each other as well as with other third variables (e.g., rights and freedom), we would
caution against arguing that one is more important than the other. Furthermore,
person-level SWB remains significantly related to peace attitudes at the individual
level.

Discussion

Global peace efforts often give strong emphasis to economic and democratic
development within societies. The belief is that freedom, equality, and wealth
will bring lasting peace to the world. Our results provide support for these efforts
in showing that the social, political, and economic structures of a society are
related to peace, as reflected in the attitudes of the people living in that society. In
nations where GDP and liberal development are high, there are greater levels of
opposition to military rule and less willingness to fight a war for one’s country. In
addition, liberal development and GDP were associated with a greater emphasis on
postmaterialist concerns and lower levels of racially intolerant attitudes in society.
Overall, these two nation-level variables have important links to peace. They may
438 Diener and Tov

be involved in fostering or reinforcing ideologies that emphasize individual rights,

democratic participation, and nonviolence.
However, there is no single key for attaining a culture of peace in its various
forms. For instance, GDP and liberal development did not predict other desirable
peace attitudes. Overall levels of confidence in government or full endorsement of
democracy were not necessarily higher among more liberally developed, wealthy
nations. If a peaceful society is one in which participation is not merely optional
but valued as a means of effecting change, then there must be a perception that
the political system is fair and just. Our analyses suggest that increasing national
wealth or civil and political liberties—though important—does not ensure that
people will be confident in their government.
In contrast, individual well-being was related to several peace attitudes. Person-
level SWB was associated with greater confidence in parliament and civil services,
and these effects were not moderated by GDP or liberal development. Person-level
SWB was also associated with endorsement of democracy, greater emphasis on
postmaterialist values, and less intolerance of immigrants and members of differ-
ent racial and ethnic groups—attitudes that are important for a peaceful society
that is open and free to all people. These findings may not be coincidental, given
the role of positive emotions in facilitating trust and cooperation (Lyubomirsky
et al., 2005; Diener & Tov, in press). Therefore, we contend that SWB be thought
of not only as a byproduct of peace, but also as a crucial element in sustaining
peace over time. In that regard, SWB may be a critical base for a culture of peace.
At the same time, important moderator effects were observed. Whether happy
people endorse a peace attitude is influenced by the norms of their society and
whether the conditions for peace exist. For example, although person-level SWB
is positively related to confidence in government, the effect is stronger where
violent inequality is low. This finding counters the belief that happy people are
“Pollyannas” who see everything positively, blind to objective social conditions.
On the contrary, a happy person is increasingly more confident in the government
when homicide rates and economic inequality are low. Similarly, person-level
SWB is associated with less racial intolerance, but not where there is poverty or
little protection of human rights. Thus, we would not want to simply increase SWB
without improving the political and economic situation of the people living in a
society. Even so, it is interesting that in the former Communist societies where
nation-level SWB is low, happy people are still more supportive of democracy and
less supportive of restrictive immigration policies compared with their unhappy
compatriots.
Future research would benefit from a larger sample of nations, as well as a
closer examination of how happy people differ across nations. Happy people in an
“unhappy country” might base their well-being on a different set of considerations
than happy people in a “happy country.” Greater attention should also be focused
on the role that SWB plays in actually bringing about and sustaining a culture
of peace. The present analyses are correlational, limiting our ability to infer the
Peace and SWB 439

direction of causal influence. Multiple methods could yield additional insights.

For instance, experimental economics could be used to simulate the conditions
under which SWB or short-term positive moods influence cooperation and social
perception.
We have seen that wealth, equality, and freedom are each associated with
different peace attitudes. Subjective well-being of the person and the nation are also
related to peace attitudes, and these associations are not fully accounted for by other
socioeconomic and political indicators. Multimethod research is needed to explore
and disentangle these effects. To that end, efforts to establish a culture of peace
have much to gain by incorporating subjective measures into a comprehensive set
of peace indicators.

References

Adorno, T. W., Frenkel-Brunswik, E., Levinson, D. J., & Sanford, R. N. (1950). The authoritarian
personality. New York: Harper & Row.
Basabe, N., Paez, D., Valencia, J. L., Rimé, B., & Diener, E. (2002). Cultural dimensions, socioeconomic
development, climate and emotional hedonic level. Cognition and Emotion, 16, 103–125.
de Rivera, J. (2004). Assessing the basis for a culture of peace in contemporary societies. Journal of
Peace Research, 41, 531–548.
Diener, E. (2000). Subjective well-being: The science of happiness and a proposal for a national index.
American Psychologist, 55, 34–43.
Diener, E., Diener, M., & Diener, C. (1995). Factors predicting the subjective well-being of nations.
Journal of Personality and Social Psychology, 69, 851–864.
Diener, E., & Seligman, M. E. P. (2004). Beyond money: Toward an economy of well-being. Psycho-
logical Science in the Public Interest, 5, 1–31.
Diener, E., & Tov, W. (in press). National accounts of well-being. In K. C. Land (Ed.), Encyclopedia
of social indicators and quality-of-life studies. New York: Springer.
Fredrickson, B. L., Tugade, M. M., Waugh, C. E., & Larkin, G. R. (2003). What good are positive
emotions in crises? A prospective study of resilience and emotions following the terrorist attacks
on the United States on September 11th, 2001. Journal of Personality and Social Psychology,
84, 365–376.
Heston, A., Summers, R., & Aten, B. (2002). Penn World Table Version 6.1. Philadelphia: Center
for International Comparisons at the University of Pennsylvania (CICUP) [producer and dis-
tributor]. Retrieved September 21, 2005 from http://pwt.econ.upenn.edu/Downloads/pwt61-
doc.exe.
Inglehart, R. (2000). Globalization and postmodern values. Washington Quarterly, 23, 215–228.
Inglehart, R., & Klingemann, H.-D. (2000). Genes, culture, democracy, and happiness. In E. Diener
& E. M. Suh (Eds.), Culture and subjective well-being (pp. 185–218). Cambridge, MA: MIT
Press.
Lyubomirsky, S., King, L., & Diener, E. (2005). The benefits of frequent positive affect: Does happiness
lead to success? Psychological Bulletin, 131, 803–855.
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis
methods (2nd ed.). Thousand Oaks, CA: Sage Publications.
Rucker, D. D., Polifroni, M., Tetlock, P. E., & Scott, A. L. (2004). On the assignment of punishment:
The impact of general-societal threat and the moderating role of severity. Personality and Social
Psychology Bulletin, 30, 673–684.
Skitka, L. J., Bauman, C. W., & Mullen, E. (2004). Political tolerance and coming to psychological
closure following the September 11, 2001, terrorist attacks: An integrative approach. Personality
and Social Psychology Bulletin, 30, 743–756.
440 Diener and Tov

Tov, W. & Diener, E. (in press). The well-being of nations: Linking together trust, cooperation, and
democracy. In B. A. Sullivan, M. Snyder, & J. L. Sullivan (Eds.), Cooperation. Malden, MA:
Blackwell.
World Values Survey Organization. (2005). Values Survey official aggregate data file, 1995 & 1999/2000
waves [computer file]. Madrid, Spain: Análisis Sociológicos, Económicos y Polı́ticos/JD Sys-
tems (APES/JDS)[distributor]. Retrieved February 22, 2006 from http://www.worldvaluessurvey.
org/services/index.html

ED DIENER is Alumni Distinguished Professor of Psychology at the University

of Illinois. He received his doctorate at the University of Washington in 1974 and
has been a faculty member at the University of Illinois ever since. His research
is focused on the measurement of well-being, temperament and personality influ-
ences on well-being, theories of well-being, income and well-being, and cultural
influences on well-being. He has edited three recent books: Well-being: The Foun-
dations of Hedonic Psychology (with Daniel Kahneman and Norbert Schwarz),
Advances in Quality of Life Studies (with Don Rahtz), and Culture and Subjective
Well-Being (with Eunkook Suh).

WILLIAM TOV is a fourth-year graduate student at the University of Illinois at

Urbana-Champaign. His research interests include subjective well-being, values,
moral judgements and moral emotions. He has written on culture and well-being
as well as on the interrelationships among happiness, trust, and cooperation.

Appendix

Sample of countries included in analyses: Albania∗ , Algeria, Argentina∗ ,

Australia∗ , Austria∗ , Bangladesh∗ , Belarus∗ , Belgium, Brazil∗ , Bulgaria,
Canada∗ , Chile∗ , China∗ , Croatia∗ , Czech Republic, Denmark, Finland∗ , France∗,
Germany∗ , Greece, India∗ , Indonesia, Ireland, Italy∗ , Jordan, Korea∗ , Lithua-
nia∗ , Luxembourg∗ , Mexico∗ , Morocco∗ , Netherlands, New Zealand∗ , Norway∗ ,
Pakistan, Peru∗ , Philippines∗ , Poland, Portugal, Romania∗ , Russia∗ , Slovenia∗ ,
South Africa∗ , Spain∗ , Sweden, Switzerland∗ , Turkey, United States∗ , Ukraine∗ ,
Uruguay∗ , Venezuela∗ , Zimbabwe∗ .
∗ Countries included in analyses of willingness to fight for country.