Religious Influences on American Public Attitudes Toward Military Action, 2008–2022

Abstract

Analysts of American politics have given only modest attention to the way religious factors shape public attitudes toward foreign policy, including the use of US military force. The Cooperative Election Studies from 2008 to 2022 provide an excellent data source for such analysis. Attitudes toward different uses of the military are well measured and the massive sample permits examination of even small ethnoreligious groups. We find that American religious groups vary greatly on overall willingness to use the military but also respond in distinctive—and predictable—ways to each type of intervention. Although religious influences differ somewhat by racial group and are partly mediated by ideology and partisanship, they often play an independent role, even under stringent statistical controls for other variables commonly found to influence public attitudes.

1. Religious Influences on Uses of Military Force

Scholars have long debated the role that public opinion plays in shaping American foreign policy. After World War II, most political scientists argued that Americans had little interest in foreign policy, were woefully uninformed about international affairs, and lacked any overarching framework for making sense of such matters, an assessment often summarized as the “Almond–Lippmann consensus” (Holsti 2004, pp. 28–40). In this view, public opinion exerted little, if any, influence on key foreign policy actors, including the president, his foreign policy advisors, and congressional leaders.

While Americans remain primarily concerned with domestic policy issues and their store of information about foreign affairs has probably not grown, scholars have discovered that attitudes on many foreign policy issues are shaped by structured and relatively stable beliefs, allowing citizens to be “cognitive misers” (for a review, see Holsti 2004). The most influential approach has been the “Wittkopf–Holsti–Rosenau” typology of militant internationalism and cooperative internationalism. Although not subsuming every foreign policy issue confronting the United States, these orientations have survived substantial changes in the international agenda (Wittkopf 1990; Holsti 2004; Eichenberg 2007). Each has a very different perspective on the role of military force: “militant internationalists” see it as the primary instrument of American foreign policy, while “cooperative internationalists” prefer diplomacy and resort to force only for “altruistic” purposes such as supporting democracy, preventing genocide, or backing UN peacekeeping efforts.

Scholars have identified many social, ideological, and demographic influences on these attitudinal structures, but much variation remains unexplained. Religious factors have been largely neglected, at least until recently, despite the growing literature on their role in domestic politics (Smidt et al. 2009). Indeed, one extensive review of Americans’ attitudes concluded that there was nothing to study: “with the exception of policy toward Israel, religion has little bearing on how they think about international affairs” (Kohut and Stokes 2006, p. 94). That perspective has been challenged by scholars from disciplines other than political science, who have made strong claims both for religion’s influence on Americans’ foreign policy views and the resulting impact on political leaders (for a brief review, see Guth 2013a). And some recent work in the discipline has argued for the influence of domestic religious influences on presidential foreign policy (Buckley 2024; Guth and Nelsen 2024).

A review of this growing body of work reveals two major themes on the role of religion in American foreign policy. A hegemonic theme argues that religion fosters attitudes of unilateralism, moralism, and nationalistic assertiveness, consistent with a militant internationalist interpretation (cf. Buckley 2024). A competing altruistic theme focuses on religious support for human rights and religious freedom, protecting the global environment, expanding international relief and rescue operations, combatting AIDS in Africa, and working for economic development (Wuthnow 2009). The altruistic approach minimizes the utility of military force, reflecting the core approaches of cooperative internationalism.

Unfortunately, even the scholars who examine the influence of religious factors on foreign policy often use simplistic assessments of American religion. First, they sometimes focus on religious elites, rather than those in the pews, despite some enormous opinion gaps between organization leaders and the rank and file (Guth et al. 1997). Second, much work has been preoccupied with white Evangelicals (Barker et al. 2008; Baumgartner et al. 2008; Froese and Mencken 2009; Page and Bouton 2006) and has neglected the other 80 percent of the public. This creates an artificial analytic dualism pitting Evangelicals against “secular” opinion (everyone else), despite evidence that many religious groups have distinctive attitudes toward foreign policy, especially on the use of military force (Hero 1973). This should not be surprising: Catholic bishops, Mainline Protestant officials, and Jewish leaders have long sought to influence both the mass public and policymakers on such issues (Hanson 1987; Wald 1992; Rock 2011; Kurtz and Fulton 2002). Such dualism also neglects the possibility that the growing ranks of religiously unaffiliated Americans—especially agnostics and atheists—may have distinctive preferences (Hansen 2011; Campbell et al. 2021, p. 157). Thus, the existing literature often overstates the distinctiveness of one religious group, ignores the potential influence of others, and generally treats American religion simplistically.

Another deficiency is the failure to include all the significant dimensions of religion, especially religiosity and religious beliefs, which are often more powerful predictors of political variables than mere affiliation. Indeed, a few recent works using sophisticated religious measures have discovered substantial differences among religious groups on foreign policy issues, including the use of military force. Analyses of surveys by the Chicago Council on Global Affairs, the American National Election Studies, and the National Survey on Religion and Politics have demonstrated that religious tradition, religious beliefs, and religious behavior all have a substantial influence on citizens’ positions on the militant and cooperative internationalist dimensions (Guth 2009, 2013a). In another study, Gries (2014, p. 122) not only found substantial differences among religious traditions in the proportions of “hawks and doves” but also discovered that religious beliefs and behaviors influenced such attitudes. Such findings suggest the need for careful measurement of religious variables in analyzing public support for military action. To understand the theoretical context for such analysis, we turn to the dominant theoretical perspectives on the “politics of American religion”.

2. Religious Groups in American Politics

In considering theoretical perspectives from which to understand the role of religion in American politics, we draw on two alternative views, one drawn from the work of American historians and the other from the sociology of religion. The “ethnocultural” historians emphasized the role of competing ethnoreligious groups in the structuring of American politics. They argued that nineteenth-century electoral contests entailed assembling the winning coalitions of contending ethnoreligious groups. Although partisan religious alignments were often complex, the Republican Party (GOP) drew from historically dominant Mainline Protestant churches, such as Episcopalians, Presbyterians, and Methodists, while Democrats spoke for religious minorities: Catholics, Jews, and southern Evangelicals (after the Civil War).

Although the class-based politics of the New Deal added an economic overlay to this pattern, the religious alignments remained largely intact for a time. But by the 1980s, changes were underway: Mainline Protestants had dwindled in numbers, Evangelicals had moved toward the GOP, the ancient Catholic–Democratic alliance had frayed, and Black Protestants had become a critical Democratic bloc (Kellstedt and Guth 2013). Growing religious diversity has added Latino Catholics and Protestants, Muslims, Hindus, Buddhists, and many others to the equation, usually on the Democratic side. But even today, many analysts think in ethnoreligious terms, referring to the “Evangelical”, “Catholic”, “Jewish”, or “Muslim” vote. Although the underlying assumptions were often incompletely articulated, historians usually argued that ethnoreligious groups held differing worldviews, cultural preferences, and negative reference groups—all shaping their views on public policy (Swierenga 2009).

The “ethno” part of ethnoreligious suggests that influences on foreign policy attitudes may well be more the result of racial differences than of religious factors. Indeed, some recent studies have found striking differences among American racial groups on questions of military power and its use. Smeltz et al. (2023) argue that Asian Americans are most supportive of military actions, followed by white Americans, with “Native Americans, Black Americans, and Hispanic Americans” reporting “the least comfort with the use of force” in various scenarios. This finding confirms much earlier findings by Wittkopf (1990, pp. 44–45) and Holsti (2004), who found that by 2000, the racial gap between white and Blacks was “emerging as one of the more powerful sources of foreign policy cleavages” (p. 228). As ethnoreligious perspectives suggest, however, race and religion often intersect in various combinations. Thus, we expect that Latino Catholics differ from Latino Evangelicals, who in turn differ from white Evangelicals. In the same vein, white Catholics will differ from Latino Catholics, and so on. This suggests the need to incorporate controls for race in any analysis using ethnoreligious traditions. If ethnoreligious differences persist after controls for race, that may demonstrate the independent impact of religion.

Historians of American foreign policy have chronicled a number of distinctive influences by ethnoreligious groups and many contemporary controversies also suggest their power. The strong antipathy of Irish-American Catholics toward any American alliance with Great Britain, the isolationism of German Lutherans and Catholics during World War I, and the powerful support of American Jews for Israel since 1948 are just three examples of ethnoreligious influence. American Catholic anticommunism in the 1950s and 1960s was shaped not only by Church condemnation of that “Godless” system but also by ethnoreligious solidarity with Eastern European brethren under Soviet domination. Support by Palestinian Americans, both Muslim and Christian, for their brethren in Gaza, the interest of Black Protestants in policy toward Africa, and the persistent lobbying of public officials by Armenian Americans over events in their homeland are just three examples of distinctive contemporary concerns of a host of American “ethnoreligious fragments” (Uslaner 2007; see also Holsti 2004, pp. 227–31). Indeed, some critics fear that American foreign policy might be unduly influenced by ethnoreligious “diasporas”, especially Latino Americans (Huntington 2004, pp. 285–91). Other observers see the numerical growth of ethnoreligious minorities preparing the way for a more “cooperative” internationalism, as these citizens might bring wider global sensitivities to public opinion.

Sociologists of religion focusing on contemporary changes in American religion have suggested an alternative perspective on religiopolitical alignments: the religious restructuring (Wuthnow 1988) or culture wars theory (Hunter 1991). Both Wuthnow and Hunter saw new religious battles emerging within old traditions, based on theology. Although their terminology differed slightly, they saw “orthodox” or “conservative” believers adhering firmly to traditional doctrines on one side of the battle, with “progressives” or “liberals” who jettisoned old religious tenets in favor of new ones based on personal experience or scientific rationality, on the other (Hunter 1991, p. 44). Although defined primarily by religious belief, the orthodox are usually more observant than progressives, complicating efforts to distinguish the effects of belief and behavior. More recently, progressives have been joined politically by a burgeoning secular population that sees morality in a similar vein (Hansen 2011; Campbell et al. 2021).

Whatever the differences among ethnoreligious traditions, some recent studies have shown the strong influence of restructuring (belief and behavior) measures on attitudes toward foreign policy generally, and military affairs in particular. Orthodox believers and the observant in most religious groups tend to be more supportive of a wider range of military actions than the religiously progressive and less observant (Guth 2009, 2013a, 2013b; Gries 2014). Even “specialized” beliefs, seldom tapped in standard academic surveys, such as Protestant dispensationalism, often have an important impact on key foreign policy and military issues, such as support for Israel or willingness to support UN action (Barker et al. 2008; Ruotsila 2008). Similarly, framing policy choices in “providential” terms may shape religious responses to policy (Glazier 2013). One unresolved question is whether religious influences have direct impact on such attitudes (Taydas et al. 2012) or are almost entirely mediated by education, partisanship, and ideology (Wittkopf 1990; Roy 2016).

3. Data and Methods

Although the influence of religious factors on support for the American military is a fascinating one, empirical analysis has been hindered by the paucity of surveys including both extensive batteries on military action and adequate religious measures. Ideally, a survey should have queries about a variety of military actions, reflecting the reality that citizens differ substantially on what circumstances should activate U.S. forces. It should also include sophisticated measures of religious belonging, behavior and belief (the “Three Bs” of religious analysis). Few academic surveys combine these traits.

Our analysis uses a pooled sample from the Cooperative Congressional Election Study series (now the Cooperative Election Study), drawing on a battery of items found in seven surveys from 2008 to 2016, 2020, and 2022. This massive data set (N = 323,707) has several advantages for our purposes but also some limitations. Militarism was well measured: respondents were asked about seven items which ascertained whether they would support use of American military force: to ensure oil supplies, destroy terrorist camps, end genocide or civil war, assist in spreading democracy, protect American allies, and help the UN uphold international law—or not permit any of these. Items such as attacking terrorists and securing oil supplies might seem to tap Wittkopf’s militant internationalism, but others appear more attuned to cooperative internationalism, such as providing support for the UN or preventing genocide. Unfortunately, however, response categories are dichotomous, “yes” or “no”, without measuring the strength of attitudes, thereby limiting variation on each item. Nevertheless, all the options are obviously relevant to American foreign policy.

Like the military items, CES religious measures have both strengths and weaknesses. First, the full battery of affiliation questions allows precise assignment of respondents to ethnoreligious traditions, the sine qua non of religious measurement (Smidt et al. 2009; Steensland et al. 2000). And the huge sample permits the study of groups that are too small to matter in the typical 800- or 1000-respondent survey. For example, the pooled sample has 5018 Latter-day Saints, 3396 Latino Protestants, 2878 Buddhists, 1744 Eastern Orthodox, 1337 Muslims, 1144 Jehovah’s Witnesses, and 859 Hindus. As we shall see, this produces a much richer picture of the increasingly diverse American religious community. The CES also has measures of religious service attendance, the importance of religion, and the frequency of prayer. Combined, these produce a highly reliable religiosity measure (alpha = 0.81). The useful measures of affiliation and behavior are not, unfortunately, matched by sophisticated belief questions; the survey asks only if the respondent is a “born-again or Evangelical Christian”. We use this as an admittedly limited proxy for theological orthodoxy among sample Christians. In this sense, it substitutes for the Biblical literalism item that Gries (2014) found as a powerful influence. These two items typically correlate at about r = 0.50, depending on the wording.

Although the CES usually lacks questions on broader foreign policy attitudes, like those in the Chicago Council of Global Affairs (CCGA) surveys and, sometimes, the American National Election Studies (ANES), the series has consistent measures of many demographic variables, such as race, gender, income, and education, often found to influence attitudes on military issues (Eichenberg 2007, 2016, 2017; Wittkopf 1990; Holsti 2004; Page and Bouton 2006). The CES also asks about the respondent’s connections to the American military, of obvious interest here (Jost et al. 2022). The survey also taps political interest, identifying the “attentive public”, which often has attitudes distinct from those of less engaged citizens (Almond 1960). Finally, the survey has standard items for partisan and ideological identification, which are increasingly powerful influences on foreign policy attitudes generally (Holsti 2004; Gries 2014). As the surveys have different numbers of respondents, we have weighted them by the average survey size (46,244), and to account for differences by year of administration, we included dummy variables for the earlier and later years in the series, but do not report the results here.

All in all, the CES provides more than adequate raw material. We proceed as follows: First, we provide an overview of how ethnoreligious factors, including religious tradition and race, structure responses to the military items, noting some clear bivariate patterns. Then, we use logistic regression to examine the impact of ethnoreligious tradition, race, demography, and political influences on each item, starting with the most-supported use, defending allies under attack, and moving toward the least approved, spreading democracy abroad. As several of these items have been used elsewhere, especially in Page and Bouton (2006), we also note how our findings comport with, add to, or challenge earlier findings, especially on the influence of religious and racial variables, much better measured in this data set. We find some distinctive religious and racial responses to these items. And although religious and racial influences are mediated in part by partisanship and ideology, they often retain an independent role, even under stringent statistical controls.

4. Results and Discussion

How do Americans regard various uses of the military? As the first line of

Table 1. Religion, race and uses of the American military.

	Mean Uses	Some Use	Assist allies	Destroy terrorists	Help UN	Stop genocide	Secure oil	Assist democracy
Total	3.53	0.89	0.73	0.64	0.45	0.41	0.22	0.18
Religious Tradition
Latter-day Saints	3.82	0.93	0.81	0.73	0.40	0.46	0.24	0.23
Jews (non-Orthodox)	3.80	0.93	0.76	0.69	0.52	0.52	0.21	0.16
White Evangelical	3.77	0.93	0.80	0.74	0.35	0.41	0.30	0.22
White Mainline	3.73	0.93	0.78	0.71	0.47	0.40	0.25	0.18
Orthodox Jews	3.71	0.91	0.76	0.75	0.30	0.45	0.33	0.21
White Catholic	3.64	0.91	0.74	0.70	0.43	0.39	0.27	0.19
Agnostic/Atheist	3.53	0.90	0.73	0.55	0.62	0.50	0.11	0.11
All Others	3.42	0.88	0.70	0.60	0.44	0.42	0.20	0.18
Black Protestant	3.35	0.87	0.66	0.52	0.47	0.40	0.20	0.23
Latino Evangelical	3.30	0.86	0.67	0.59	0.35	0.41	0.22	0.21
Eastern Orthodox	3.30	0.87	0.63	0.62	0.38	0.37	0.24	0.19
Buddhist	3.26	0.86	0.63	0.47	0.57	0.47	0.12	0.14
Latino Catholic	3.16	0.86	0.61	0.53	0.41	0.38	0.18	0.20
Hindu	3.16	0.85	0.51	0.52	0.47	0.35	0.21	0.24
Nothing in Particular	3.13	0.83	0.64	0.53	0.45	0.37	0.17	0.14
Muslim	2.84	0.82	0.45	0.36	0.43	0.42	0.15	0.21
Jehovah’s Witnesses	1.77	0.48	0.34	0.31	0.21	0.23	0.12	0.08
Born-Again Status
Born Again	3.64	0.91	0.76	0.69	0.36	0.40	0.29	0.24
Not Born Again	3.48	0.89	0.72	0.62	0.48	0.42	0.20	0.16
Religiosity
Very High	3.70	0.88	0.77	0.71	0.36	0.43	0.27	0.24
High	3.61	0.88	0.74	0.67	0.41	0.40	0.26	0.22
Moderate	3.51	0.89	0.72	0.65	0.43	0.39	0.24	0.19
Low	3.43	0.91	0.70	0.62	0.47	0.40	0.21	0.15
Very Low	3.41	0.91	0.71	0.56	0.57	0.45	0.14	0.12
Race
White	3.62	0.91	0.75	0.67	0.45	0.42	0.24	0.18
Asian	3.26	0.85	0.63	0.52	0.49	0.40	0.17	0.20
Black	3.19	0.84	0.62	0.49	0.46	0.38	0.18	0.21
Latino	3.13	0.84	0.61	0.52	0.41	0.39	0.17	0.19

Source: Cooperative Congressional Election Studies, 2008–2016, 2020; Cooperative Election Study, 2022. N = 323,707.

suggests, Americans give quite varied responses to hypothetical uses of U.S. armed forces. Almost 9 in 10 are in principle willing to use the military in at least some circumstance, rejecting a thoroughgoing pacifist stance. Seventy-three percent would support allies under attack, almost two-thirds back action against terrorist camps, but slightly less than half were willing to enforce UN actions, and only four-tenths would use the military to stop genocide. Just over one in five would countenance using American forces to protect oil supplies and less than one-fifth would use the military to help build democracy abroad. Of course, real-world events might alter these findings in any particular survey, but the responses are fairly stable over the period from 2008 to 2022. Most responses fluctuated somewhat from survey to survey, although willingness to hit terrorist camps declined more or less monotonically from 2008 to 2022. Even considering the temporal variation, the limited range of responses certainly suggests fairly stable levels of “permissive consensus” for these military actions (cf. Page and Bouton 2006, p. 46).

Whatever the overall levels of public support, how do members of America’s diverse ethnoreligious and racial communities respond to these “uses” of U.S. armed forces? Table 1 reports our findings for ethnoreligious groups—ordered from those most willing to use force to those least willing to do so. We also note results for different levels of religiosity and “born-again” status. Finally, we show data for the four largest racial groups, which overlap the religious categories in complex ways.

The findings provide some fascinating insights into both strong general tendencies and interesting peculiarities. First, Latter-day Saints, non-Orthodox Jews, and white Evangelical Protestants are most sympathetic to military action overall, followed closely by white Mainline Protestants and Orthodox Jews. White Catholics trail at some distance, with agnostics and atheists just above the sample average. Virtually all the ethnoreligious minorities fall below the sample average, as do respondents claiming “Nothing in Particular” (NIPS). Jehovah’s Witnesses are the least “militaristic” of all, followed by American Muslims. Religiosity also shows a clear pattern: the most observant are the strongest supporters of a range of military actions, while the least observant are the most resistant. There is a similar gap between “born-again” Christians and those not taking that label, with the former approving more military actions. And as some scholars have suggested, the racial gap is fairly substantial: white Americans are substantially more like to approve military action than are Asian Americans, Blacks, or Latinos, who differ only modestly from each other. We note that this order differs slightly from that in Smeltz et al. (2023): Asians in that study were more militaristic than whites; here, they are similar to Blacks and Latinos.

Although reactions to the individual items naturally tend to fall in the same pattern as the cumulative score, there are some clear variations. Among the ethnoreligious groups, Jehovah’s Witnesses invariably appear as the least willing to engage in military action, living up to their pacifist roots. Muslims react in predictable ways, being less likely to support American forces supporting allies (perhaps Israel?), attacking terrorist camps, or seizing oil resources. As Page and Bouton (2006) discovered, white Evangelicals tend to be quite positive about a range of military actions, but are least supportive of action supporting the UN, exceeded in their distrust only by Orthodox Jews. The greatest support for UN assistance comes from agnostics and atheists, who with the two Jewish groups are also the most likely to favor action in cases of genocide or civil war. White Evangelicals and Orthodox Jews also far exceed the sample average in willingness to seize oil supplies by force. Born-again Christians are usually more willing to use force—except in support of the UN or to stop genocide—with a similar pattern appearing among the most observant. And although whites are much more inclined to support allies and destroy terrorist camps than other racial groups, differences on other items are relatively modest. And the minority racial groups do not usually differ much from each other in their relative hesitancy to use military force.

To explore the ethnoreligious and racial sources of attitudes, we now turn to a multivariate examination of each of the six occasions for military action. We use logistic regression to parse the impact of ethnoreligious and racial factors, applying controls for a variety of demographic and political variables shown to influence public attitudes on military matters. Although the CES does not tap the general foreign policy orientations used by Page and Bouton (2006) to explain those attitudes, we do have partisanship and self-identified ideology, thought by many scholars to have an increasingly strong influence on contemporary foreign policy orientations (Holsti 2004; Gries 2014).

We report four models in each table: the first incorporating only religious variables, the second adding race, the third adding demographic traits, and the final model, two political variables. Each equation also controls for year of administration. For ethnoreligious groups, the omitted reference category is “All Other Religions”, a group close to the mean on most of the individual items (and theoretically of less interest). For race, the omitted reference category is “All Other Races or More than One”, also a substantial group collectively close to the mean on most items. To facilitate a focus on ethnoreligious influences, we order the groups in each table from the one with the largest positive coefficients in the first model to that with the largest negative tendency (always the Jehovah’s Witnesses).

4.1. Occasion One: Defending Allies Under Attack

As we have seen, the scenario eliciting the widest approval for military action is to “protect American allies under attack by foreign nations”. Despite this broad consensus, there are still clear religious differences here. As the first column in

Table 2. Religious, demographic and political influences on military support for allies.

	Religion	+ Race	+ Demographics	+ Politics
Religious Tradition
Latter-day Saints	0.564 ***	0.455 ***	0.427 ***	0.360 ***
White Evangelical	0.512 ***	0.363 ***	0.325 ***	0.266 ***
White Mainline	0.441 ***	0.300 ***	0.211 ***	0.194 ***
Jews (non-Orthodox)	0.355 ***	0.241 ***	0.155 ***	0.180 ***
Agnostic/Atheist	0.279 ***	0.219 ***	0.179 ***	0.141 ***
Orthodox Jews	0.240 **	0.145	0.133	0.113
White Catholic	0.233 ***	0.091 ***	0.049 *	0.031
Black Protestant	−0.179 ***	0.220 ***	0.212 ***	0.267 ***
Nothing in Particular	−0.180 ***	−0.196 ***	−0.095 ***	−0.051 **
Latino Evangelical	−0.193 ***	0.147 ***	0.120 **	0.120 *
Buddhist	−0.225 ***	−0.232 ***	−0.275 ***	−0.250
Eastern Orthodox	−0.274 ***	−0.372 ***	−0.419 ***	−0.449 ***
Latino Catholic	−0.384 ***	−0.035	−0.039	0.018
Hindu	−0.752 ***	−0.665 ***	−0.704 ***	−0.663 ***
Muslim	−1.026 ***	−0.932 ***	−0.863 ***	−0.840 ***
Jehovah’s Witnesses	−1.528 ***	−1.450 ***	−1.345 ***	−1.181 ***
Other Religious Factors
“Born Again”	0.004	0.002	0.033 **	0.017
Religiosity	0.068 ***	0.085 ***	0.086 ***	0.035 ***
Race
White		0.031	0.106 ***	0.127 ***
Asian		−0.237 ***	−0.072 *	0.030
Latino		−0.461 ***	−0.290 ***	−0.216 ***
Black		−0.519 ***	−0.385 ***	−0.246 ***
Demographics
Male			0.472 ***	0.366 ***
Higher Education			0.103 ***	0.068 ***
Higher Income			−0.001	−0.001 ***
Age			0.006 ***	0.001 ***
South			0.073 ***	0.070 ***
Married			0.067 ***	0.024 **
Active Military Duty			−0.277 ***	−0.358 ***
Family			0.171 ***	0.119 ***
Veteran/Family			0.293 ***	0.240 ***
Politics
Political Interest				0.253 ***
Conservative/GOP				0.159 ***
Constant	0.568 ***	0.679 ***	−0.486 ***	−0.790 ***
Nagelkerke R square w/o controls	0.032	0.037	0.075	0.084

Source: Cooperative Congressional Election Studies, 2008–2016, 2020; Cooperative Election Study, 2022. N = 323,707. Controls for the year of administration included in each model. The omitted religious reference category is All Other Religions; for race, it is All Other Races or More Than One. * p < 0.05; ** p < 0.01; *** p < 0.001. + indicates the addition of a new set of variables to the equation.

shows, the strongest support comes from Latter-day Saints, Evangelicals, and Mainline Protestants, with both Jewish groups, agnostics/atheists, and white Catholics trailing. Most ethnoreligious minorities are distinctly less supportive, especially Hindus and Muslims. Religiosity provides some modest additional support for allies, but born-again status drops out completely.

Adding race to the equation nudges up the R squared from 0.032 to 0.037. Whites do not differ from the omitted reference group, but Asians, Latinos, and Blacks are less likely to support allies. The controls for race, however, have an interesting impact on the ethnoreligious coefficients, reducing but not eliminating those of most “white” religious groups. But the impact of religious affiliation among some racial minorities is revealing. When race is controlled, Black Protestants are actually more supportive of allies than other African Americans (bivariate data show 66 percent of Black Protestants approving, versus 56 percent of other Blacks). A similar effect occurs among Latino Evangelicals, whose coefficient becomes positive once race is controlled. And the negative coefficient for Latino Catholics disappears when race is controlled. Thus, among these Christian ethnoreligious minorities, membership in a religious community produces more support for allies than is present among their non-religious brethren.

In the third equation, we introduce demographic variables found previously to be influential on military policy: gender (Page and Bouton 2006, p. 122), education, income, age, region (Holsti 2004, pp. 224–29), marital status (Shafer and Spady 2014), and military connections (Jost et al. 2022). Page and Bouton argue that women are consistently less militaristic, but “most other demographic factors do not” matter (2006, p. 122). To the contrary, we find that men are indeed more likely to support allies, but higher education also contributes to support, as does age, southern residence, and being a veteran or member of a veteran’s family. Interestingly, active-duty personnel have the more skeptical attitude toward military action discovered in some surveys (Jost et al. 2022), although their families lean a little the other way. Contrary to Page and Bouton’s findings, the introduction of demographic variables adds substantially to the variance explained, but leaves coefficients for religious variables virtually unchanged, if modifying those for racial groups a bit. This finding also differs from Wittkopf (1990, p. 45), who found that religion and race drop out of the equation once education and other demographic factors are controlled.

To this point, we have seen that religious, racial, and other demographic traits influence public attitudes toward military action. We also expect that “political mobilization”, measured here by political interest (Inglehart 1970), should have an impact, tapping the effects of being in the “attentive public”, long the focus of foreign policy opinion studies (Almond 1960). We also suspect, of course, that ideational factors will have a major influence, being closer in the “causal order” (Miller and Shanks 1996, p. 190ff) to military policy options. In recent years, scholars have increasingly stressed the role of ideology and partisanship in shaping foreign policy attitudes. Some analysts stress ideology (Gries 2014), others emphasize partisanship (Holsti 2004), and yet others argue that there is little difference between these factors in contemporary American politics, as parties are largely sorted by ideological preferences. In our sample, partisanship and ideology are so highly correlated (r = 0.71) as to warrant combining them in the analysis. We anticipate that their influence may run in different directions, depending on the military use. Republicans and conservatives should favor “militant” uses, while Democrats and liberal should prefer “cooperative” ones.

As Table 2’s last column shows, adding political factors boosts the explanatory power of our analysis. High political interest increases the likelihood of supporting action in defense of allies, as does being strongly “conservative/GOP.” Not surprisingly, inclusion of political variables reduces somewhat the coefficients for the ethnoreligious traditions, but most remain solid and in the same direction as the bivariate tendencies as modified by race. Thus, even though Latter-day Saints and white Evangelicals are massively conservative and Republican in identification, their religious location still adds to their willingness to support allies under attack. For white Catholics, however, religious membership adds nothing to that propensity, once politics are taken into account. On the other side, many ethnoreligious groups that lean heavily to liberal/Democratic do not move much from Models 2 and 3 to Model 4, suggesting sustained religious influence on the “non-military” side. Thus, even on this widely approved use of the American military, the residual ethnoreligious impact still tends to reinforce GOP willingness to support allies under attack—and pull Democrats and liberals in the other direction.

4.2. Occasion: Destroying Terrorist Camps

As we noted above, military action “to destroy terrorist camps” is the second most “popular” use of American forces, although that support declined steadily after the first survey in 2008.

Table 3. Religious, demographic and political influences on military anti-terror actions.

	Religion	+ Race	+ Demographics	+ Politics
Religious Tradition
White Evangelical	0.601 ***	0.523 ***	0.465 ***	0.296 ***
Orthodox Jews	0.651 ***	0.606 ***	0.608 ***	0.512 ***
Latter-day Saints	0.566 ***	0.513 ***	0.484 ***	0.270 ***
Jews (non-Orthodox)	0.472 ***	0.415 ***	0.341 ***	0.421 ***
White Mainline	0.462 ***	0.389 ***	0.275 ***	0.219 ***
White Catholic	0.457 ***	0.383 ***	0.316 ***	0.238 ***
Eastern Orthodox	0.128 *	0.084	0.050	−0.038
Agnostic/Atheist	−0.013	−0.028	−0.062 **	−0.051 *
Latino Evangelical	−0.110 **	0.222 ***	0.188 ***	0.131 **
Nothing in Particular	−0.183 ***	−0.154 ***	−0.042 *	−0.034
Latino Catholic	−0.248 ***	0.090 **	0.068 *	0.119 ***
Hindu	−0.267 ***	−0.177 ***	−0.196 **	−0.041
Black Protestant	−0.354 ***	0.120 ***	0.122 ***	0.239 ***
Buddhist	−0.422 ***	−0.400 ***	−0.451 ***	−0.352 **
Muslim	−0.923 ***	−0.808 ***	−0.732 ***	−0.642 ***
Jehovah’s Witnesses	−1.234 ***	−1.112 ***	−1.052 ***	−0.922 ***
Other Religious Factors
“Born Again”	0.072 ***	0.067 ***	0.083 ***	−0.001
Religiosity	0.063 ***	0.076 ***	0.087 ***	−0.017 **
Race
White		−0.119 ***	−0.047 *	0.015
Asian		−0.319 ***	−0.088 **	0.020
Latino		−0.531 ***	−0.329 ***	−0.188 ***
Black		−0.673 ***	−0.519 ***	−0.240 ***
Demographics
Male			0.707 ***	0.555 ***
Higher Education			0.054 ***	0.040 ***
Higher Income			0.000	−0.001 ***
Age			0.011 ***	0.006 ***
South			0.089 ***	0.069 ***
Married			0.137 ***	0.057 ***
Active Military Duty			−0.130 ***	−0.194 ***
Family			0.180 ***	0.101 ***
Veteran/Family			0.268 ***	0.203 ***
Politics
Political Interest				0.254 ***
Conservative/GOP				0.424 ***
Constant	0.243 ***	0.437 ***	−0.936 ***	−1.245 ***
Nagelkerke R square w/o controls	0.050	0.055	0.120	0.155

See Table 2 notes for details. * p < 0.05; ** p < 0.01; *** p < 0.001.

reports an analysis of responses to that item. With more variation in the dependent variable, we naturally expect to be able to explain more. Indeed, the patterns are primarily somewhat “sharper” replicas of those in the previous analysis. Evangelicals, Orthodox Jews, Latter-day Saints, other Jews, and White Mainline Protestants and Catholics are all above the mean on support for such use, while ethnoreligious minorities line up on the other side, with Muslims and Jehovah’s Witnesses most opposed. Once more, religiosity bolsters the approval of military use, here with an added push from “born-again” status. Ethnoreligious variables alone explain 5 percent of the variance.

When race is added, the variance explained rises a little, with all the large racial groups more skeptical than the “miscellaneous” omitted reference group. Once again, controls for race flip the signs for Black Protestants and both Latino groups, suggesting that religious membership has a “militarizing” effect even in these racial groups. With race in the equation, religiosity’s impact actually rises a little, while the “born-again” coefficient is somewhat reduced.

The demographic and political variables also largely replicate the patterns in Table 2: men, the highly educated, older citizens, southerners, the married, and veterans or their families tend to support such action, with active-duty personnel again moving in the other direction (but not their families). Strong political interest supports action against terrorists, as does conservative/GOP identification. And although the political variables essentially absorb the influence of religiosity and born-again status, the coefficients for most of the ethnoreligious groups remain in the same direction as in the earlier models, though occasionally reduced in magnitude. With everything in the mix, white and Asian Americans are not distinctive, although Latinos and Black Americans are still somewhat more skeptical of military action against terrorists. All variables together account for 16 percent of the variance, a very respectable result. And a comparison of Table 2 and Table 3 suggests that the basic structure of opinion on these two issues is quite similar.

4.3. Occasion: Supporting the UN

As Page and Bouton note, “‘Dovish’ uses of military force for multilateral peacekeeping and other humanitarian purposes often evoke patterns of demographic support that are muted or reversed compared with those for ‘hawkish’ uses” (2006, p. 129). Certainly, support for helping the UN “uphold international law” falls within this rubric. And their suspicion about the unique sources of support for such action is warranted.

As

Table 4. Religious, demographic and political influences on military support for the UN.

	Religion	+ Race	+ Demographics	+ Politics
Religious Tradition
Agnostic/Atheist	0.432 ***	0.291 ***	0.248 ***	0.122 ***
Buddhist	0.370 ***	0.256 ***	0.212 ***	0.018
Black Protestant	0.326 ***	0.274 ***	0.251 ***	0.082 **
Jews (non-Orthodox)	0.195 ***	0.025	−0.055 *	−0.196 ***
White Mainline	0.136 ***	−0.067 ***	−0.073 ***	0.043 **
Hindu	0.072	0.000	−0.039	−0.202 ***
Muslim	−0.061	−0.109	−0.085	−0.233 ***
Latter-day Saints	−0.097 **	−0.268 ***	−0.263 ***	0.119 ***
White Catholic	−0.099 ***	−0.301 ***	−0.282 ***	−0.126 ***
Nothing in Particular	−0.143 ***	−0.275 ***	−0.242 ***	−0.158 ***
Latino Evangelical	−0.155 ***	−0.114 *	−0.108 *	0.061
White Evangelical	−0.164 ***	−0.375 ***	−0.342 ***	−0.065 ***
Latino Catholic	−0.178 ***	−0.126 ***	−0.108 ***	−0.091 *
Eastern Orthodox	−0.305 ***	−0.460 ***	−0.501 ***	−0.361 ***
Orthodox Jews	−0.611 ***	−0.759 ***	−0.784 ***	−0.580 ***
Jehovah’s Witnesses	−1.046 ***	−1.142 ***	−1.046 ***	−0.912 ***
Other Religious Factors
Religiosity	−0.127 ***	−0.116 ***	−0.126 ***	−0.011
“Born Again”	−0.304 ***	−0.298 ***	−0.253 ***	−0.098 ***
Race
White		0.450 ***	0.452 ***	0.377 ***
Asian		0.342 ***	0.282 ***	0.342 ***
Latino		0.196 ***	0.200 ***	0.070 *
Black		0.291 ***	0.283 ***	−0.028
Demographics
Male			−0.099 ***	0.015
Higher Education			0.122 ***	0.062 ***
Higher Income			−0.004 ***	−0.003 ***
Age			0.001 ***	0.000
South			−0.050 ***	0.003
Married			−0.124 ***	−0.037 ***
Active Military Duty			0.011	0.009
Family			−0.040 **	0.018
Veteran/Family			0.012	0.046 ***
Politics
Political Interest				0.098 ***
Conservative/GOP				−0.642 ***
Constant	−0.300 ***	−0.549 ***	−0.891 ***	−0.970 ***
Nagelkerke R square w/o controls	0.040	0.043	0.055	0.141

See Table 2 notes for details. * p < 0.05; ** p < 0.01; *** p < 0.001.

demonstrates, support for UN action comes from a very different ethnoreligious, racial, demographic, and political coalition than our previous cases of support for allies and attacking terrorists. To begin with, support for the UN comes from different ethnoreligious groups, with agnostics/atheists, Buddhists, and Black Protestants leading the way, followed at some distance by non-Orthodox Jews and Mainline Protestants. On the other side, most ethnoreligious groups that favored support for allies and attacks on terrorists are less supportive of this purpose, with Orthodox Jews most critical (aside from the Jehovah’s Witnesses, of course), followed by the Eastern Orthodox, both Latino groups, white Evangelicals, and “NIPS”. More significantly, both high religiosity and born-again identity work powerfully against support, suggesting that religious traditionalism still produces opposition to the UN, based no doubt on some combination of historic theological objections by conservative Protestants and contemporary resistance by religious traditionalists to a panoply of liberal social policies long supported by UN agencies (cf. Ruotsila 2008; Page and Bouton 2006, p. 128; Gries 2014, pp. 241–249).

When race is incorporated in the equation, we add only modestly to the variance explained (0.040 to 0.043) but see some significant changes in many religious group coefficients: those for most of the “supporting” ethnoreligious groups are reduced substantially, but those for some in the “opposing” coalition are actually enhanced. At the same time, religiosity and born-again status retain solid negative coefficients. All this confirms a role for religious traditionalism in limiting support for the UN across most racial groups.

The demographic variables also reveal a somewhat different pattern from earlier analyses. Men are less likely to support US military action supporting the UN, as are southerners, married people, and wealthier citizens. Military status has much less impact on this option, but higher education still produces more support for military action. Finally, political interest has a mildly supportive effect and conservative/GOP identity has a massive negative one, as might be expected. The incorporation of the political variables mediates much of the influence of the “supporting” religious traditions, with the coefficients for agnostics/atheists, Buddhists, and Black Protestants substantially reduced—but still positive—and that for non-Orthodox Jews “flipping signs”. On the other side, notably, the political variables substantially reduce the strong negative coefficient for white Evangelicals and for religiosity, although born-again identity retains some independent influence. And inclusion of the political variables almost triples the variance explained from Model 3. On the whole, the configuration of the supporting and opposing coalitions on this usage looks much different from those on the two more widely supported uses.

4.4. Occasion: Preventing Genocide or Civil War

Another “dovish” use of the American military might be to “intervene in a region where there is genocide or a civil war”. Like support for the UN, this item almost divides the American public in half, with only 41 percent of the pooled sample approving this use. As

Table 5. Religious, demographic and political influences on military action to prevent genocide.

	Religion	+ Race	+ Demographics	+ Politics
Religious Tradition
Jews (non-Orthodox)	0.436 ***	0.390 ***	0.336 ***	0.293 ***
Agnostic/Atheist	0.406 ***	0.375 ***	0.295 ***	0.193 ***
Buddhist	0.248 ***	0.242 ***	0.192 ***	0.116 **
Latter-day Saints	0.155 ***	0.109 ***	0.024	0.168 ***
Orthodox Jews	0.120	0.081	−0.020	0.081 ***
Muslim	0.005	0.038	−0.043	−0.091
White Evangelical	0.003	−0.053 **	−0.006	0.093 ***
Latino Evangelical	−0.014	0.002	0.015	0.089
Black Protestant	−0.044 *	0.078 **	0.105 ***	0.055
White Mainline	−0.050 **	−0.103 ***	−0.076 ***	−0.029
White Catholic	−0.108 ***	−0.162 ***	−0.134 ***	−0.063 ***
Nothing in Particular	−0.166 ***	−0.185 ***	−0.149 ***	−0.068 ***
Latino Catholic	−0.177 ***	−0.158 ***	−0.129 ***	−0.084 *
Eastern Orthodox	−0.198 ***	−0.237 ***	−0.330 ***	−0.266 ***
Hindu	−0.248 ***	−0.207 ***	−0.323 ***	−0.362 ***
Jehovah’s Witnesses	−0.877 ***	−0.869 ***	−0.746 ***	−0.540 ***
Other Religious Factors
Religiosity	0.044 ***	0.049 ***	0.063 ***	0.087 ***
“Born Again”	−0.072 ***	−0.072 ***	−0.037 ***	0.032 **
Race
White		0.047 *	0.103 ***	0.069 ***
Asian		−0.065 *	−0.171 ***	−0.065 *
Latino		−0.025	−0.028	−0.029
Black		−0.131 ***	−0.102 ***	−0.149 ***
Demographics
Male			0.160 ***	0.124 ***
Higher Education			0.151 ***	0.096 ***
Higher Income			−0.002 ***	−0.002 ***
Age			−0.009 ***	−0.014 ***
South			−0.065 ***	−0.048 ***
Married			−0.017 *	−0.005
Active Military Duty			−0.118 *	−0.183 ***
Family			0.072 ***	0.064 ***
Veteran/Family			0.090 ***	0.072 ***
Politics
Political Interest				0.301 ***
Conservative/GOP				−0.203 ***
Constant	−0.519 ***	−0.513 ***	−1.234 ***	−1.234 ***
Nagelkerke R square w/o controls	0.010	0.011	0.036	0.064

See Table 2 notes for details. * p < 0.05; ** p < 0.01; *** p < 0.001.

shows, the response to this usage falls along somewhat similar lines as that for UN operations but is less well explained by religious variables—or for that matter, by all the usual predictors, even in combination. The modest explanatory power of the analysis below probably reflects the “double-barreled” nature of the question, conflating the very different cases of “civil war” and “genocide”. Questions referencing “genocide” alone elicit much higher levels of support (Page and Bouton 2006, p. 276). Here, it seems likely that various religious groups may focus on different sides of the question.

However the question is interpreted, the main supporting constituency for such action are, understandably, non-Orthodox Jews, agnostics/atheists, Buddhists, and Latter-day Saints—groups with historical sensitivities, perhaps, to religious or racial oppression. Here, Orthodox Jews, Muslims, and White and Latino Evangelicals all fall in the middle of the spectrum, with no significant leaning either way. Most other ethnoreligious groups have mild inclinations to oppose such intervention, with Jehovah’s Witnesses once again at the far end. Here, religiosity and born-again identity work in different directions, with the former being positive and the latter being negative. But all the religious variables in combination explain very little of the variation in attitudes, nor does the addition of racial identity have much explanatory power, with white Americans slightly more inclined to intervene and other racial groups leaning the other way. The addition of demographic and political variables provides a little more explanatory power. Men, the highly educated, the less well-off, young people, non-southerners, and veterans are more likely to support action, but once again the active-duty personnel are inclined the other way. High political interest once again produces more support for action, but the partisanship measure reveals greater liberal/Democratic support, but not at a level characterizing support for the UN. Once again, the incorporation of demographic and political variables modifies the strength of the direct religious influences but seldom changes their direction. The only notable “sign flip” is for white Evangelicals, who become modestly pro-action when everything is in the equation.

4.5. Occasion: Assuring Oil Supplies

Another item taps more “self-interested” motives for the use of the American military: “to ensure the supply of oil”. Although surveys in 2002 showed that almost two-thirds of Americans supported such use, this had declined to just over half by 2004 (Page and Bouton 2006, pp. 106, 278). By the 2008 CES survey, this had declined to around one-quarter, dropping slightly to around one-fifth in 2014 and remaining fairly constant thereafter.

As

Table 6. Religious, demographic and political influences on military action to secure oil supplies.

	Religion	+ Race	+ Demographics	+ Politics
Religious Tradition
Orthodox Jews	0.702 ***	0.733 ***	0.679 ***	0.534 ***
White Catholic	0.444 ***	0.478 ***	0.435 ***	0.317 ***
White Evangelical	0.410 ***	0.446 ***	0.386 ***	0.165 ***
Eastern Orthodox	0.312 ***	0.346 ***	0.326 ***	0.206 ***
White Mainline	0.312 ***	0.347 ***	0.313 ***	0.218 ***
Latter-day Saints	0.239 ***	0.276 ***	0.207 ***	−0.087 *
Jews (non-Orthodox)	0.221 ***	0.254 ***	0.297 ***	0.429 ***
Hindu	0.159	0.144	0.121	0.309 ***
Latino Evangelical	−0.025	0.227 ***	0.209 ***	0.073
Nothing in Particular	−0.027	0.028	0.047 *	0.017
Latino Catholic	−0.049	0.202 ***	0.179 ***	0.180 ***
Black Protestant	−0.114 ***	0.063	0.118 ***	0.272 ***
Muslim	−0.254 ***	−0.211 **	−0.264 ***	−0.108 ***
Agnostic/Atheist	−0.421 ***	−0.380 ***	−0.389 ***	−0.296 ***
Buddhist	−0.513 ***	−0.489 ***	−0.485 ***	−0.311 ***
Jehovah’s Witnesses	−0.604 ***	−0.510 ***	−0.567 ***	−0.502 ***
Other Religious Factors
Religiosity	0.077 ***	0.078 ***	0.112 ***	−0.006
“Born Again”	0.250 ***	0.247 ***	0.198 ***	0.078 ***
Race
White		−0.150 ***	−0.095 ***	−0.009
Asian		−0.092 *	0.035	0.112 **
Latino		−0.367 ***	−0.306 ***	−0.126 **
Black		−0.292 ***	−0.241 ***	0.139 ***
Demographics
Male			0.602 ***	0.484 ***
Higher Education			−0.060 ***	−0.043 ***
Higher Income			0.000	−0.001 ***
Age			0.000	−0.003 ***
South			0.151 ***	0.117 ***
Married			0.129 ***	0.037 ***
Active Military Duty			0.201 ***	0.166 **
Family			0.075 ***	0.012
Veteran/Family			0.069 ***	0.019 *
Politics
Political Interest				0.060 ***
Conservative/GOP				0.555 ***
Constant	−1.631 ***	−1.515 ***	−1.771 ***	−1.814 ***
Nagelkerke R square w/o controls	0.037	0.038	0.066	0.118

See Table 2 notes for details. * p < 0.05; ** p < 0.01; *** p < 0.001.

shows, the “militaristic” coalition in this case looks much more like that in the first two occasions (supporting allies and attacking terrorists) than that on supporting the UN and preventing genocide. Orthodox Jews, white Catholics, white Evangelicals, the Eastern Orthodox, Mainline Protestants, and non-Orthodox Jews form the support base for this use, with most ethnoreligious minorities and agnostics/atheists on the other side. Higher religiosity also contributes to support for such action, as does born-again identity. The addition of race adds little to the explanation and has only modest and inconsistent effects on the religious group coefficients, sometimes reducing them a little, but just as often boosting them, an effect most notable among Latino Evangelicals and Catholics, as well as Black Protestants. Among both Latinos and Black Americans, religious affiliation produces more militaristic opinions.

As usual, the demographic and political variables boost the power of our equations. Men are much more likely to approve this use, as are those with more modest educations, southerners, married citizens, and in a change from earlier usages—active-duty military personnel and their families. Interestingly, political interest provides only a modest push for military use, but conservatives and Republicans are much more willing to undertake such action. Still, even the inclusion of demographic and political variables in the equation leaves most of the supporting and opposing ethnoreligious group coefficients operating in the same direction, sometimes reduced a little, but often actually increased. All this suggests a strong underlying influence of religious factors.

4.6. Occasion: Spreading Democracy

The final item in the CES battery is a little difficult to characterize in Page and Bouton’s hawkish vs. dovish dichotomy or in Wittkopf’s cooperative vs. militant internationalism scheme. The use of American troops to “assist the spread of democracy” abroad might be thought of in liberal internationalist terms or, alternatively, as a part of the second Bush administration’s “war on terror”. Whatever the case, it musters the least support of all the options in the CES battery over the decade and a half of surveys: 25 percent supported this purpose in 2008, dropping gradually to 17 percent in 2022.

The ambiguity of the item (like that of the genocide/civil war item) is perhaps reflected in the unusual pattern of support and opposition, both on the ethnoreligious variables and the other predictors. As

Table 7. Religious, demographic and political influences on military action to promote democracy.

	Religion	+ Race	+ Demographics	+ Politics
Religious Tradition
Hindu	0.538 ***	0.306 ***	0.310 ***	0.369 ***
Latter-day Saints	0.266 ***	0.382 ***	0.353 ***	0.263 ***
Latino Catholic	0.256 ***	0.240 ***	0.238 ***	0.258 ***
Orthodox Jews	0.249 **	0.350 ***	0.333 ***	0.259 **
Muslim	0.243 ***	0.169 *	0.121	0.169 *
Black Protestant	0.197 ***	0.061	0.092 **	0.155 ***
White Catholic	0.188 ***	0.324 ***	0.319 ***	0.284 ***
Eastern Orthodox	0.155 *	0.257 ***	0.252 ***	0.219 ***
Jews (non-Orthodox)	0.133 ***	0.251 ***	0.306 ***	0.339 ***
White Evangelical	0.101 ***	0.241 ***	0.226 ***	0.152 ***
White Mainline	0.090 ***	0.226 ***	0.234 ***	0.209 ***
Latino Evangelical	−0.004	−0.014	−0.013	−0.053
Nothing in Particular	−0.029	0.032	0.024	0.037
Buddhist	−0.050	−0.078	−0.074	−0.022
Agnostic/Atheist	−0.123 ***	−0.039	−0.036	−0.031
Jehovah’s Witnesses	−0.937 ***	−0.922 ***	−0.956 ***	−0.851 ***
Other Religious Factors
“Born Again”	0.264 ***	0.262 ***	0.230 ***	0.197 ***
Religiosity	0.180 ***	0.173 ***	0.193 ***	0.149 ***
Race
White		−0.113 ***	−0.094 ***	−0.062 *
Asian		0.339 ***	0.380 ***	0.444 ***
Latino		0.040	0.033	0.115 **
Black		0.165 ***	0.149 ***	0.279 ***
Demographics
Male			0.195 ***	0.118 ***
Higher Education			−0.037 ***	−0.047 ***
Higher Income			−0.002 ***	−0.002 ***
Age			−0.003 ***	−0.005 ***
South			0.063 ***	0.052 ***
Married			−0.018	−0.053 ***
Active Military Duty			0.436 ***	0.410 ***
Family			0.108 ***	0.076 ***
Veteran/Family			0.046 ***	0.017
Politics
Political Interest				0.124 ***
Conservative/GOP				0.173 ***
Constant	−1.916 ***	−1.940 ***	−1.797	−1.972 ***
Nagelkerke R square w/o controls	0.025	0.026	0.032	0.040

See Table 2 notes for details. * p < 0.05; ** p < 0.01; *** p < 0.001.

reveals, ethnoreligious support here includes Hindus, Latter-day Saints, Latino Catholics, Orthodox Jews, and Muslims. Other groups provide more modest backing, including Black Protestants, Eastern Orthodox, non-Orthodox Jews, and both white Protestant groups. The strongest opponents are agnostics and atheists. Both higher religiosity and born-again identity provide a boost for support. Adding race does little to boost the explanation: white Americans are somewhat less likely on balance to favor this use, but Asian and Black Americans are more approving. Including race in the equation also modifies the ethnoreligious coefficients, with the religiously affiliated often showing stronger support than their non-affiliated racial brethren.

Demographics also play a role. Men are somewhat more likely to approve this use, as are citizens with less education and a lower income. Young people and southerners are also more likely to back such action, as are active-duty personnel. Political interest and conservative/GOP identification also provide a very modest push in favor of support for democracy. The incorporation of political variables eliminates neither the independent effect of religious group membership nor the pro-military influence of religiosity and born-again status.

4.7. Conclusions and Implications

What can we learn about the connection between religion and attitudes toward militarism from this analysis? Although the CES studies have some limitations—especially the absence of religious belief items—we can still discover a good bit about the way religious factors influence the American public’s attitude toward the use of force, especially in discovering the varied ethnoreligious coalitions supporting different purposes, in analyzing the impact of religiosity, and identifying the impact of “born-again” status.

First, we see a general tendency for America’s historically dominant ethnoreligious groups to be more supportive of military action than most ethnoreligious minorities. On an overall score, Latter-day Saints, Jews, white Evangelicals, Mainline Protestants, and Catholics approve a broader range of military actions, while most ethnoreligious or “new” religious minorities are much more selective in their willingness to use American force. This pattern is also broadly evident on the most “popular” uses of armed activity, supporting allies and fighting terrorism, as well as on the less-favored seizing of oil supplies, but almost completely reverses on support for the UN. Finally, the two opposing coalitions are seemingly “scrambled” in reaction to cases of genocide/civil war and on military action designed to promote democracy.

The positioning of the non-religious “Nothing in Particular” and the explicitly “religion-rejecting” group of agnostics and atheists is also of interest, given their rapid growth in numbers in recent years and their political importance, especially in the Democratic Party. Campbell et al. (2021, p. 157) find these two groups—especially the latter—on the liberal side of almost all issues, including defense spending. Although they hug the national mean on overall willingness to use the military, they vary significantly on specific uses: the “NIPS” are fairly consistent in their preference to avoid US military action, whatever the rationale. Agnostics and atheists, on the other hand, do take consistently “liberal” approaches: supporting allies, strongly backing the UN, hoping to prevent genocide, and opposing both the seizure of oil and the promotion of democracy. These findings show the importance of sorting out various contingents of the “non-religious”.

Second, we see a fairly consistent pattern of influence for religiosity to promote support for military action, except in the case of helping the UN, where it has a solid negative impact. The effects of religiosity usually survive the addition of demographic and political variables to the analysis. The influence of “born-again” identity is a little more varied, having little influence on support for allies, but a solid positive effect on fighting terrorism, and a major impact on seizing oil and promoting democracy. On the other hand, born-again status works against fighting genocide and, especially, supporting the UN, two “dovish” purposes, according to some analysts. This suggests that better measures of theological traditionalism might have even more powerful effects.

Third, we see that many of these religious effects survive the addition of demographic and political variables to the analysis. The racial, demographic, and political variables, for the most part, behave as predicted on the basis of previous studies, but they do not eliminate the direct effect of religion. Although coefficients for the ethnoreligious group, religiosity, and born-again variables are sometimes reduced by the addition of race, other demographics, and political factors, they usually retain significant direct effects. Even contemporary partisanship, strongly structured by religious factors, does not entirely mediate the impact of religion.

All this suggests that scholars studying attitudes toward the military and, more broadly, Americans’ foreign policy orientations ought to take religious factors more seriously. This is especially true for religious beliefs, usually unmeasured in most public opinion surveys on foreign policy. The Chicago Council on Global Affairs’ popular surveys of public attitudes have actually reduced the religious measures available to a very crude measure of religious affiliation, with no religiosity measure or any kind of belief item, despite studies showing considerable (if varied) religious impact on foreign policy questions (Page and Bouton 2006; cf. Gries 2014).

In this analysis, we have demonstrated the value of adding religious measures to the study of opinion on military questions, confirming earlier findings that religious factors influence not only the overarching attitude structures that Americans use to assess foreign policy but also opinions on specific foreign policy issues. Here, we find that even in the presence of other variables that are often thought to be much more powerful influences, religious factors make a substantial contribution, sometimes indirectly through political orientations, but often as direct forces in their own right. In some instances, such distinctive attitudes can be understood as a function of membership in a particular ethnoreligious tradition (e.g., Jewish support for ally Israel and for action to stop genocide, Jehovah’s Witnesses’ negative reaction to all warfare). In other instances, one suspects that attitudes reflect the influence of specific beliefs not often measured in political surveys (e.g., Evangelical eschatology leading to opposition to any action supporting the UN). Or, as Peter Gries (2014) has argued, religious influences may reflect deeper psychological or moral orientations (see also Kertzer et al. 2014). In any event, future studies should take care to incorporate the various dimensions of religious faith and practice in their surveys. Indeed, such inclusion is especially warranted as the policies of the Trump administration appear likely to scramble both the political and religious sources of foreign policy attitudes in the American public (Guth and Nelsen 2024).