By Donald Rothchild and Nikolas Emmanuel; University of California, Davis

Draft Copy

Work in Progress

The primacy of the United States in today’s world entails inescapable responsibilities. A great power cannot always stand aloof and ignore the widespread destruction of civil war, genocide, and the spread of disease and hopelessness without some loss of its own security and well-being in its region and beyond. Life in a dangerous global neighborhood has damaging effects on every country. Consequently, the United States must at times assume the mantle of a “protector.” In circumstances where the African state fails to manage tensions within its domain responsibly and effectively, the international community has an obligation to act. As Gareth Evans and Mohamed Sahnoun state, “[t]he responsibility to protect implies a duty to react to situations in which there is compelling need for human protection.” [1]

U.S. leaders sometimes find themselves facing a difficult choice: whether to refrain from intervening in the internal affairs of African states and to respect their sovereign jurisdiction or to intercede forcefully, alone or in coalition with other states, in an effort to prevent starvation and repression – even the mass murder of vulnerable peoples. In addition, policymakers must select among a variety of means. Public officials deemed diplomatic pressures and incentives to be appropriate to deal with intra-state conflicts and humanitarian crises in Angola, Mozambique, South Africa, Namibia, Kenya, Sudan, Burundi, and the Democratic Republic of the Congo (DRC). Moreover, an extensive military intervention was regarded as appropriate and necessary to deal with the challenges of delivering food and restoring political order in Somalia. Only when the costs of acting as a protector became apparent in Somalia did the United States and United Nations reverse policies and decide upon a course of disengagement. And this decision to remain on the sidelines had terrible ramifications in terms of protecting other vulnerable peoples in Rwanda and Sudan. [2]

Although foreign policymakers have wide discretion to make decisions on a variety of conflicts lacking mass media attention, they generally need Congressional and public support, tacit if not explicit, for interventions of greater concern to domestic interests. Clearly, decision-making is a complex process, influenced in large part by public concerns and state capabilities. Commercial organizations, missionary groups, and ethnic lobbies all exert an influence on foreign policy far out of proportion to their numbers. [3] It would also be shortsighted to overlook the contribution that a society’s norms and values contribute to this process. Societal values become evident in the demands that elite and civil society interests make upon the state. They have an impact on the formulation of public goals across regimes, and especially in democratic countries that require public support to achieve its purposes. As Joseph Nye comments: The national interest, which is broader than vital strategic interests, “can include values such as human rights and democracy, particularly if the American public feels that those values are so important to our identity or sense of who we are that people are willing to pay a price to promote them. … If the American people think that our long-term shared interests include certain values and their promotion abroad, then they become part of the national interest.” [4]

The American public can play an import role in foreign policy formulation, then, by pressing its objectives or constraining the actions of the policy elite. This paper seeks to illustrate this by examining the impact of Congressional and public attitudes on executive decision-making in a single issue-area: e.g., the preparedness of the United States to act as a protector of vulnerable peoples in Africa at times of acute distress. To this end, the paper asks four questions:

1. Do U.S. Congressmen and the general public exhibit a general preference for the use of options that involve non-military force rather than military force in Africa?

2. Do Congressmen and the U.S. public prefer multilateral to unilateral interventions when dealing with Africa’s internal discord?

3. Do key demographic characteristics have a significant impact on the level of U.S. public support for intervention in intra-state conflicts?

4. Is the public a constraint on U.S. interventions to protect Africa’s vulnerable peoples?

To address these questions, we have based our analysis on 149 public opinion polls concerning American preferences about intervention in seven ethnic conflicts: Burundi, Bosnia-Herzegovina, Kosovo/Yugoslavia, Liberia, Rwanda, Somalia, and South Africa. This material was gathered from the Lexis-Nexis Universe “Polls and Surveys” search engine, which is maintained by the Roper Center for Public Opinion Research and includes opinion polls from sources such as Gallup, Harris, Roper, ABC, CBS, CNN, NBC, The Los Angeles Times, The New York Times, USA Today, and The Wall Street Journal. [5]

II. The U.S. Role as Protector

With a few exceptions – Angola, South Africa, Somalia, and, currently, Sudan – the U.S. government has been extremely cautious about becoming diplomatically or militarily caught up in Africa’s internal disputes. Despite a Wilsonian tradition that calls for self-determination, multilateralism, and an end to arbitrary power, America’s liberal internationalist commitment comes into conflict with other deeply-held principles on unilateral action and economic expansionism. As William Schneider notes, the American public “believes in internationalist principles, … [b]ut when it comes to policy, the isolationist impulse begins to intrude.” [6] Accordingly, the Chicago Council on Foreign Relations data for 1990 indicated that 61% of the people surveyed sought cut backs in economic assistance to other countries, and 73% called for reductions in military aid. [7] As publicist Robert D. Kaplan expressed the non-interventionist preferences held by many members of the general public: the United States should only intervene where “moral, economic, and strategic interests intersect.” [8] Giving substance to these sentiments, the continuous reductions in taxes and the budgetary restrictions put into place during and after the Reagan period have had the effect of limiting federal government capacity to protect Africa’s vulnerable peoples.

U.S. officials have, despite these substantial constraints on international action, engaged occasionally in far-reaching diplomatic and humanitarian operations in Africa, alone or in cooperation with the United Nations. In certain selected cases, the U.S. role as protector of Africa’s vulnerable has been evident in various ways during the phases of the conflict process – in particular, during the Gestation or Politicization, Triggering/Escalation, and Post-Conflict Phases of conflict.

Preventive action can, at times, contribute importantly to the avoidance of highly destructive encounters. As Michael Doyle and Nicholas Sambanis note, it is important for third-party protectors to intervene early “before the parties have done extensive killings.” [9] Ideally, it is most useful for U.S. policymakers to address deep-rooted socio-economic, environmental, and structural causes of disagreements during the Potential Conflict Phase, when popular discontent is latent and scattered and when it remains at relatively low levels of intensity. [10] The U.S. role in preventing conflict has not received either the governmental or public attention it deserves in the early phases. Public opinion polls tend to take notice of Africa’s ethnic conflict issues during the Escalation Phase and less so during the periods when outside influence can prove most effective – e.g., the Gestation or Post-Conflict Phases. Thus, 101 of the 159 questions regarding public support for U.S. intervention in Africa’s ethnic conflicts were posed after the conflict had flared up and had become intense, and only 17 were asked during the Gestation or Politicization Phase.

The Politicization Phase clearly represents a critical juncture in the conflict process. Unless the United States and other third parties act decisively to influence local leaders to negotiate their differences, the fleeting opportunity to bridge inter-group differences may be lost. The United States and other actors, as a consequence of their economic and political power, often have considerable leverage at their disposal ; they can, for example, freeze economic assistance, back counter-elites, include or exclude target regimes from international organizations, close embassies, and place sanctions on trade or arms. In worst cases, such as Rwanda, they might have prevented the cascades of violence from occurring in 1994 had they used their influence to enlarge the peacekeeping force on the scene and been more determined in implementing the peace agreement. [11] As Figures 1 and 2 indicate, the U.S. public, while less than enthusiastic about military intervention, gave policymakers considerable freedom to act had they been so inclined to do so.

It is often difficult for external protectors to generate the kind of political will necessary to deal with the root causes of conflict in Africa. U.S. policymakers can no doubt recognize that glaring economic disparities, population pressures, or the denial of human rights may exacerbate tensions in Zimbabwe or Liberia, but they are often reluctant to intercede and at times find themselves frustrated when it comes to convincing Congress to allocate the resources needed to address these problem meaningfully.

As a consequence, what starts as elite manipulation of ethnic consciousness or a failure of communications has been allowed at times to lead to expanded enmities. In this Politicization Phase of conflict, elite ambition or communal sentiment can gather momentum and coalesce to the point where it cannot be ignored by potential third-party protectors. [12] Although the possibility of group violence has increased, a determined third-party intervention can still hold out the possibility of preventing deadly conflict from occurring. In this regard, U.S. and other intercessions in the early phases of conflict did help to keep the conflict in South Africa at manageable levels.

In the 1980s and early 1990s, the U.S. government, with extensive public support, did contribute, somewhat reluctantly, to a relatively peaceful transition in South Africa from an apartheid regime to a non-racial democracy. The U.S. Congress, which passed the 1986 Comprehensive Anti-Apartheid Act over the opposition of the Reagan administration, employed a variety of economic pressures to induce South African authorities to engage in political and economic reform and to negotiate with the African-led opposition. Later in 1992, it combined soft diplomatic incentives to support both the internal negotiating process and the United Nations’ initiative in dispatching an observer team. [13] Virtually all the questions in our sample asked during the Politicization Phase were raised about South Africa, and they were generally supportive of a non-military approach.

However, when, as in Rwanda, Sudan, Liberia, Sierra Leone, and the DRC, conflict increased noticeably in intensity and the efforts of external facilitators proved unavailing, the conflict entered the Triggering Phase. The triggering of mass violence represented a dangerous expansion of tensions, for polarization became rigid and norms of inter-group reciprocity and political exchange weakened. As the level of tension rose, negotiation was complicated by a combination of elite manipulation of ethnic loyalties and by the zero-sum perceptions of group members, as well as by the lack of reliable information regarding the intentions of adversary leaders and the difficulty of making credible commitments that will survive the implementation process. Clearly, the United States and other third parties have limited scope for initiative once the fighting erupts; even so, an awareness of the constraints under which they operate is not an excuse for inaction. Thus the bitter conflicts that were triggered in such countries as Angola, Sudan, the DRC, Sierra Leone, East Timor, Macedonia, Cyprus and elsewhere have responded over time to mediation efforts by diplomats from the United States, the United Nations, and others.

It is necessary to engage in such diplomacy early on in order to prevent a cycle of violence and possible warfare from taking place. At times, as in Macedonia and the DRC, the United States has contributed to efforts to contain the spirals of violence and to promote peace negotiations under turbulent conditions. In other cases, such as Sudan, Western Sahara, and Israel-Palestine, the U.S. has attempted to facilitate an end to ongoing wars that appear to have escalated out of control, and the short-term Triggering Phase has evolved into a long-term Escalation Phase marked by recurring rounds of warfare. Facilitation is the form of intervention preferred by U.S. government decision-makers and seems relatively independent of public opinion. Nevertheless, it is intresting that during the Escalation Phase public opinion polls have been most likely to raise questions about the preferences of U.S. residents on intervention in Africa’s conflicts. The results have tended to show cautious public support for Operations Other Than War (OOTW) and somewhat less enthusiastic backing for direct military intervention.

In time, when military victory appears to be beyond the reach of the adversaries and when peace appears to be less risky and costly than continued warfare, mediation and negotiation may succeed in shifting conflict relations back toward regularized encounters among groups. This Post-Conflict Phase is marked by decided risks of its own, for the hostility and distrust of civil wars does not dissipate easily and can, as in Sudan, Angola and Rwanda, lead back to a resumption of the fighting soon afterwards. [14] To avoid this destructive scenario, external third parties such as the United States can play an important role in helping to monitor and validate the short-term military-related aspects of the implementation process (supporting the maintenance of the cease-fire, helping to separate the contending forces and quartering the troops at specified assembly points, overseeing the disarmament and demobilization of forces, re-integrating the rival armies, and encouraging and assisting in the reform of the police). [15] Even though the U.S. public is inclined to be cautiously supportive of these initiatives during the Post-Conflict Phase, the United States government has been extremely reluctant to play an important role in helping post-conflict societies achieve their confidence- and security-building guidelines. Something of an exception here is Angola. Following the signing of the peace agreement in 2002, U.S. diplomats repeatedly assured Angolan officials that they would provide generous assistance for the peace implementation process. This, however, represents a special transitional situation in which the United States is extricating itself from a Cold War involvement.

III. The U.S. Public’s and Congress’s Response

At every phase in the conflict cycle, the U.S., as the world’s primary state actor, made or could have made a critical difference in intra-state conflicts in Africa. But did Congress and the public support U.S. government intervention, and if so, what kind of intervention did they prefer? Certainly, various administrations require considerable room to maneuver on many issues that lie outside the public’s radar screen. On other more visible issues, however, public sentiment may serve as a useful guide as to what is moral and practicable. [16] If the President acts contrary to public sentiments, as Clinton did when he sent troops to Bosnia in 1995, this can involve considerable risks -- in this case in his relations with a reluctant House of Representatives. [17] Moreover, without public support, the President lacks “leverage.” [18] As John E. Rielly warns, “policies which over time are contrary to public sentiment will almost certainly fail – along with the leaders responsible for them.” [19]

It is time to look more closely at public attitudes to see whether the members of the public in fact place constraints on U.S. governmental action in intervening in ethnic-related conflicts in Africa. Does the American public have clear preferences concerning U.S. intervention in such conflicts? And do members of the public differentiate in terms of the different tools to be used, in what situation, and when they should be employed? In an attempt to answer these questions, we examine three basic methods for dealing with ethnic conflicts (i.e., non-military force interventions, operations other than war, and military force) and whether the public has stable preferences about applying them.

Non-military force interventions refer to actions taken by third parties designed to alter the perceptions and behavior of a target state or movement without recourse to the use of military action. [20] The third party seeks to alter local behavior by means of diplomatic pressure and packages of incentives. [21] Examples of the incentives used include monetary assistance, diplomatic pressure, security guarantees, recognition, investment restrictions, trade embargoes, and economic sanctions. These actions are best typified by the American policies towards South Africa in the 1980s. In addition, the cases in this study illustrate other types of incentives that have been used as well.

Operations Other Than War (OOTW) consist of a wide range of activities in which the armed forces of the third party or parties are deployed in non-combat missions. The United States has had a long tradition of deploying its forces in such actions. The classic example is the Berlin Airlift of 1948-49. This operation used military aircraft to fly food and fuel to the beleaguered city to break the Soviet blockade. While this type of operation has been going on for decades, their frequency picked up dramatically after the Cold War. [22] The U.S. Army defines operations other than war as "military activities during peacetime … that do not necessarily involve armed clashes between two organized forces". [23] These operations include humanitarian assistance, disaster relief, nation-building assistance, arms control, support to US civil authorities, and peacekeeping. [24] In the seven cases examined in this study, the United States undertook OOTW in four of them: a noncombatant evacuation and brief humanitarian effort in Rwanda, more extensive humanitarian relief operations in Somalia, and peacekeeping efforts in Bosnia-Herzegovina, and Kosovo.

The surveys used in this study refer primarily to two types of OOTW: humanitarian-military interventions and peacekeeping. Humanitarian interventions include any action, possibly involving troops in non-combat missions, that are destined to respond in combined non-military and military ways to humanitarian problems. Under this category, the opinion polls we refer to include such examples as: airlifting supplies like food and medicine to refugees, assuring the distribution of humanitarian aid, providing medical assistance for refugees, protecting civilians in or fleeing from a conflict zone, and restoring survival support systems, (i.e., water lines, sewers, food storage facilities, power supply stations, and so forth). The UN agency specializing in humanitarian affairs describes such operations as “… emergency relief and longer-term assistance … on behalf of people struck by emergencies.” [25] For such activities, N.A.T.O. has adopted the following definition: “A humanitarian intervention is an armed intervention in another state, without the agreement of that state, to address (the threat of) a humanitarian disaster, in particular caused by grave and large-scale violations of fundamental human rights.” [26] This definition is largely in agreement with those put forward in the scholarly literature on this subject. In the words of Bhikhu Parekh, humanitarian intervention is "an act of intervention in the internal affairs of another country with a view to ending the physical suffering caused by the disintegration or gross misuse of authority of the state, and helping create conditions in which a viable structure of civil authority can emerge." [27]

Peacekeeping interventions include a wide range of activities. According to the United Nations Department of Peacekeeping Operations, peacekeeping interventions use combined military, police, and civilian forces to “help implement peace agreements, monitor ceasefires, create buffer zones, or support complex military and civilian functions essential to maintain peace and begin reconstruction and institution-building in societies devastated by war.” [28] When opinion surveys prompt questions on the subject of peacekeeping, they frequently use the term “peacekeeping forces” to clarify what is intended. However, such questions can also refer to peace enforcement, the maintenance of peace agreements, and so forth.

Military force refers to a third party’s use of armed forces in explicit combat roles in an effort to overcome resistance on the part of the target state or movement. Direct military force or combat actions are implemented by third parties with the goal of decisively altering the balance of forces on the ground. [29] For example, combat troops have been considered an option to stop aggressors in a civil war (e.g., Bosnia) or in the case of genocide (e.g., Rwanda). Third parties usually resort to military interventions when other means are thought to be insufficient. In reference to military force, public opinion polls prompt respondents by referring to “the use of U.S troops...to end the civil war”, “sending troops … to fight and repel the aggressor”, “using armed forces … to try to stop the violence”, or “sending in ground troops to stop the killing.”

IV. Analysis of the Data

What tool does the American public prefer when dealing with ethnic conflicts? Are these preferences generalizable across cases, or are they region specific? To initiate a discussion of these issues, this study examines 159 survey questions from ethnic conflict in seven countries, five in Africa and two in the Balkans. The questions have been broken down into the three categories based on descriptions of the tools of interventions noted above.

The three figures entitled “Level of U.S. Public Opinion Support by Type of Intervention” (Figures 1-2) illustrate the rank order relationship between the three intervention tools that are central to a discussion of public preferences. In both the African and Balkan cases the public appears to prefer non-military intervention, then OOTW, and lastly military force.

Furthermore, in the chronological country-case figures (Figures 3-6), the hierarchy of public preferences is even more obvious, for non-military means receive the highest scores, closely followed by support for operations other than war, like humanitarian interventions and peacekeeping. These forms of action are both clearly preferred over military force.

Descriptive statistics in Table 1 further demonstrate this rank order of preferences with the relative frequencies of support for each category of intervention. Clearly, the mean value of support for military measures is significantly lower than the others when it comes to confronting ethnic wars in both groups of counties. In contrast to the Balkans, OOTW is virtually evenly supported with non-military measures in the African cases. Furthermore, American public opinion also supported sending US troops to Liberia in the summer of 2003 for peacekeeping and combat operations. The mixed results for the African cases appear to be driven by the relatively large number of questions on non-military intervention in South Africa (21 of 25 polls), which appears to be driving the overall frequency distribution for the African cases in this category. Bosnia and Kosovo are indicative of the rank order preferences generally. Hence, our overriding conclusion is that the public is largely opposed to sending U.S. troops to fight to bring an end to ethnic civil war as compared to other less violent options; at the same time, the data seem to indicate that the level of public support for intervention is high relative to the level of actual government intervention.

Table 1: Mean Level of Support by Type of Intervention and Country.

(A = Africa, B = Balkans)

We also conducted a series of regression analyses to test the relative influence of the three types of interventions on the level of public support, while controlling for four other factors that are prominent in the literature (i.e., multilateralism, presidential cues, congressional opposition, and risk – Table 2). An Ordinary Least Squares (OLS) regression is used to analyze the impact of our independent variables on the levels of support on the African cases (made up of 82 opinion polls), as well as on the comparative cases of Bosnia and Kosovo (77 polls).

Table 2:Multivariate Analysis of Public Support for Intervention in Ethnic Conflict

(ordinary least squares regressions)

It is important to indicate that we have operationalized the three types of interventions as being mutually exclusive. However, because of the problem of perfect multicolinearity (that is, two of the three intervention types predict the third), one of these dummy variables had to be dropped from the regressions analysis. The variable for non-military intervention was therefore left out of the model, but its estimated value can be easily identified from the Constant in Table 2. This allows the level of support for non-military interventions to be the benchmark by which the other tools and controls can be measured. That is to say, their coefficients indicate the estimated deviation from the level of support for non-military force.

Interestingly, the coefficients of the constant for the African and Balkan cases in Table 2 are very close. They both indicate a strong preference for such types of action. The expected score on an intervention using non-military force is 70 percent of public support in the African cases and 63 percent for Bosnia and Kosovo. From that baseline, the coefficients for the other two types of intervention can be interpreted. Unfortunately, the estimations derived from the surveys concerning OOTW are not as statistically significant for the African cases. However, the negative direction in its coefficient makes some sense, suggesting that there is a slightly lower expected value for public support of OOTW than non-military force. The Balkan cases, on the other hand, are more revealing, perhaps because they were primarily thought of as OOTW. Their coefficient on the OOTW variable is highly significant and estimates that a set of responses in the sample on such a question would result in a score 15 points lower than that of a non-military force survey.

Finally, the coefficients on military force in both groups of countries are highly statistically significant and negative. One would expect that a survey asking if the public supports sending U.S. troops to fight the aggressors in the civil war in a given country, it would lead to a level of support 21 points (or about 51 percent) less than measures involving non-military force in an African scenario, and 31 point lower (about 32 percent) in Bosnia or Kosovo. Based on the data from our sample of polls, this reiterates the overall public preferences against military interventions, and in favor for non-military force measures and OOTW in dealing with ethnic conflict. An explanation of this could be that Bosnia and Kosovo were fundamentally peacekeeping missions (within the OOTW set), while none of the African cases evolved into this type of operation. By far the most powerful variable in any of the models is the military force variable, which is always negative and statistically significant. This indicator plays a significant role in driving the results of this model.

Multilateralism vs. Unilateralism: Does the American public have a preference for multilateral over unilateral intervention? Clearly, multilateral interventions in uncertain situations provide two clear advantages over unilateral action. First, such operations allow the Untied States to share the burden or risks involved with a potentially costly and/or dangerous operation. Second, they provide an added sense of legitimacy to any action when a large number of other countries feel that they have a shared interest in the initiative. This question on public preferences has been a recurrent theme in the public opinion literature. As Kull and Destler (1999) recently argue, "Americans are highly sensitive about operations being either multilateral of unilateral." [30] Beyond a general sensitivity on this issue, a number of studies have demonstrated that the U.S. public has exhibited a clear preference for multilateral interventions in a wide range of post-Cold War conflicts. [31] For Hinckley, "Most [of the American people] …take a cooperative or multilateral stance, favoring modification of U.S. interests to consider those of other countries." [32]

In this study of public attitudes regarding U.S. intervention, there are 159 public opinion polls when all the African and non-African cases are considered together, and 82 polls when the African cases are analyzed separately. The data indicate that a preference for multilateral intervention does not exist for every type of intervention, but applies only to certain types of intervention (e.g., mainly those dealing with highly coercive combat operations where the risks and costs are high.) We coded each question on the opinion polls we surveyed using a dichotomous variable to test for a public preference on this issue. We coded a “1” when a question contained direct references to a multilateral operation (e.g., U.N., N.A.T.O., allies, international community, and so forth), and coded a “0” when the question referred to the U.S. acting alone. The following two tables summarize the breakdown of the percentages and numbers of polls associated with each county-case (Table 3) and each intervention type in the African sample alone (e.g., Burundi, Liberia, Rwanda, Somalia, and South Africa) (Table 4). The results are set out below.

TABLE 3: Preference for Multilateral Intervention by Country-Case:

(A = Africa, B = Balkans)

TABLE 4: Preference for Multilateral Intervention by Type (African Cases):

Our data appear to run counter to those compiled by Reilly (1999), for he argues that, in general, Americans prefer multilateral approaches to unilateral ones in addressing international crises. Reilly states: “Seventy-two percent [72 percent to 21 percent] of the public … think the United States should not take action alone in responding to international crises if it does not have the support of allies.” [33] There is no significant public preference for multilateralism (an overall average of 55 percent of respondents gave support to questions using unilateral language and 56 percent supported questions mentioning multilateral action). In the African cases (Table 4), 63 percent favored multilateralism and 59 percent preferred unilateralism. [34] However, when the type of intervention is taken into account, interesting differences do emerge in our data. [35] The public strongly prefers a multilateral approach in situations where combat operations are envisioned (Tables 4). This data seems to follow the logic suggested by Kull (1999), who indicated that "…the majority is very hesitant about using military force unilaterally, but relatively comfortable with the idea of being part of a multilateral operation." [36]

A similar preference for multilateralism (although less significant in magnitude) is seen in regard to the use of non-military force. Even though "operations other than war" (including humanitarian and peacekeeping missions) are considered to be primarily multilateral operations, there is no clear preference for them in our data set. The correlation between a preference for multilateralism and the level of public support is higher when the five African cases in our study are considered without Bosnia and Kosovo (all cases 0.056, African cases = 0.1335 ).

In both groups of cases, there is positive support for multilateral intervention. Questions indicating a preference for multilateral intervention have higher than average levels of support in the opinion polls as compared with the polls referring to unilateral action. Although multilateralism has an effect on the support for intervention in ethnic-related conflicts, it is by no means the most important factor. In the seven-case model, the multilateralism variable was regressed with five other independent variables regarding the level of support from the public opinion polls.

In both the seven -case and African-case analyzes the coefficient for multilateralism was positive and statistically significant (Table 2). That is to say that there was a relationship between higher than average levels of support and the possibility of a multilateral intervention, especially with the African sample. This analysis of our data runs counter to Jentleson and Britton’s conclusion, for they argue that, "multilateralism … is never statistically significant. Its coefficients are all small, sometimes positive, sometimes even negative…;” suggesting that this variable is not a viable indicator of public preferences. [37] Rather our data tends to agree with Kull’s (1995) findings that the public is more inclined to support intervention when the action involves a multilateral force.

Furthermore, the U.S. public does not seem to run parallel with Congressional resistance to multilateralism over the past decade. Since the 103rd Congress, members of the House and Senate have been highly critical of any multilateral involvement, such as United Nations peacekeeping missions. This pressure from Congress even caused President Clinton to back away from his earlier policy of “assertive multilateralism” at the end of his first term. Reflecting this change of priorities, his speech before the U.N. General Assembly on September 27, 1993 not surprisingly emphasized the limits of U.S. involvement in U.N. multilateral missions. Clinton stated that, “The United Nations simply cannot become engaged in every one of the world’s conflicts … If the American people are to say yes to U.N. peacekeeping, the United Nations must know when to say no.” [38] Only a week later, after the deaths of 19 U.S. soldiers in Somalia, this feeling among American leaders became more pronounced. Senator Mitch McConnell (R-KY) typified this growing sentiment, contending that “[c]reeping multilateralism died on the streets of Mogadishu … This is not just about Somalia. This is about how we should operate in the post-cold war world.” [39]

In sharp contrast to these opinions, the American public, in polls conducted by both the Chicago Council on Foreign Relations (CCFR) and the Program on International Policy Attitudes (PIPA), actually favored multilateralism over both unilateralism or isolationism at times when the U.S. confronted international crises. [40] The 1999 CCFR report concluded that the U.S. public preferred multilateral approaches by 72 percent to 21 percent when confronting international crises. [41] Leaders were more evenly split, with 48 percent favoring multilateral action and 44 percent supporting unilateralism. [42] Our survey data point to a similar conclusion, but with some caveats. Thus the public has a clear preference for multilateralism when U.S. troops are to be used to intervene and actively end a civil war. However, such a choice is not as apparent when the other foreign policy tools are discussed.

Risk: defined here in terms of potential or actual loss of American life, has been identified in the recent literature on public opinion as an important variable that requires serious attention. It must be analyzed in relation to other factors, particularly the type of intervention that is undertaken. [43] As Larson argues, “it is … the prevalence of a particular class of operation that explains the apparent recent low tolerance for casualties in U.S. military interventions.” [44] Jentleson and Britton reach a similar conclusion: “the public has limited tolerance for casualties, the limits vary somewhat with the PPO [principal policy objective] being pursued.” [45] Our survey data also finds risks to be related to the type of intervention.

We coded a dichotomous variable linking the exposure of U.S. to public attitudes on intervention. That is to say, we coded a “1” when the sending of U.S. troops was strongly considered and fighting was going on in the theater of deployment; we coded a “0” once the situation had become relatively settled (i.e., no regular combat) and a ceasefire was being respected. Rwanda during the height of the genocide and the RPF advance (April to mid-July, 1994) would be an example of the first, while Bosnia after February 1996, regardless of a few minor incidents, would fit into the second category.

Based on this coding, the data from our multivariate OLS model indicated that risk was negative and highly statistically significant for the seven African cases, but not for Bosnia or Kosovo. . The American public seems to be hesitant about intervening in any African conflict where hostilities are on-going. It is interesting to note, however, that the regression coefficient in regard to the African cases was negative, but American public responses were positive with respect to both Bosnia and Kosovo. This could mean that the American public was more risk averse in the African cases than with regard to the Balkans.

Influence of Elites: What preferences do American leaders hold concerning intervention in African ethnic conflicts? Do significant gaps emerge between elites and the public on conflict management issues? Based on the polls set out in Table 8, two interesting points stand out. First, both the American public and opinion leaders share a preference for non-military interventions and operations other than war, while being extremely cautious about the use of military force. Second, the public and foreign policy leaders sometimes react differently to specific situations in different countries. The largest gap emerges in regards to the 1994 Rwanda genocide. From our survey data in Figure 6, one year after the genocide, some 70 percent of the American public thought that it would have been better to “have gone in with a large military force to occupy [Rwanda] and stop the killings.” On the other hand, only 18 percent of foreign policy elite would have used U.S. military forces conflict reemerged in Rwanda and threatened further acts of genocide. This 50 percent disparity is somewhat of an anomaly in terms of the similarities of opinion that elites and the public held on dealing with other confrontations in Africa.

TABLE 5: Assessments of American Opinion Leaders on Ethnic War in Africa. [46]

Presidential Cues: Because of the primacy of the executive in foreign affairs, it is understandable that the public could be influenced by arguments expressed by the President on a given foreign policy issue. A number of studies have examined the potential impact of presidential leadership on public opinion and reached similar conclusions regarding the President’s cue-making capacity. [47] Illustrating this point with respect to Lebanon, Russett provides an interesting example of the direct influence of President Reagan on public opinion:

"The president's speeches can be a powerful political tool. … A sharp and dramatic demonstration of this comes from a Washington Post/ABC three-day poll which happened to straddle the night of President Reagan's speech on Lebanon of October 17, 1983. On the day before the speech 41 percent of the sample said they approved of his policy on Lebanon; on the following day the percentage was 52 (and within two months it had dropped back again)…" [48]

It seems logical that most of the time, as in the example above, there would be a positive relation between a presidential cue and public opinion. However, such a cue can also serve to undermine public opinion, complicating any attempt to measure its influence. In the case of South Africa in the mid-1980’s, we find an activist Congress seeking to impose sanctions on apartheid South Africa and being countered by strong resistance from the Reagan administration. While public opinion in favor of more restrictive sanctions averaged 74 percent during the two months prior to Reagan’s nationally televised speech on July 22, 1986, his address sent support on this issue plummeting to an average of 46 percent in the following months. . Reagan's negative impact on public support for sanctions lasted until well into 1987, when support began to return to the levels prior to the President's speech. Here the President used his ability to cue the public to undermine a policy that he opposed. The effect was to complicate simplistic expectation that presidential cues will have a positive impact on public opinion.

In order to control for any potential effect of presidential cueing, our study tested the relationship between cueing and public opinion by coding any presidential cue as a “1” when the President has publicly taken a position of support for action at the time of the survey in one of the country-cases, and an “0” when there was no public intervention by the President around the time of the survey. The source for this information were Presidential Public Papers, available online from the Reagan, Bush and Clinton Presidential Libraries. [49] In this case, however, presidential cues did not appear to be statistically significant in either the African or non-African cases.

Congressional Opposition: How does a disagreement between the executive and the legislative branches over the means of dealing with an ethnic conflict alter U.S. public opinion? More specifically, can significant congressional opposition to the president reduce the level of support for a given intervention, as is frequently argued in the case of Somalia.

A number of researchers have looked into the question of how the emergence of deep divisions between the U.S. Congress and the President can polarize public opinion. [50] They frequently contend that when leaders agree that an operation is worthwhile, this increases public support for that action. When leaders are divided, however, the public tends to be similarly divided. [51] This argument seems logical and was used in our analysis as an additional control for the impact of elites on public opinion. Using information available from the Congressional Record, we coded a “1” if there was significant opposition in Congress to the President’s position, and a “0” if there was not. [52] Based on current research findings, we would expect congressional opposition to the president to lower public support for any type of intervention in an ethnic conflict.

Our data indicate that there is a negative correlation between congressional opposition and public support overall. This negative correlation is much stronger and statistically significant only concerning the African cases. This could be attributable to the influence of events in specific counties in our sample. First, South Africa, as was already briefly described, provides an example of when Congress was an advocate of a foreign policy position, in this case anti-apartheid sanctions opposed by the executive branch. Their opposition to President Reagan appears to have had a positive effect on the level of public support for sanctions against the apartheid regime. This case runs counter to the above-mentioned expectation regarding the impact of a disagreement between the two branches of government.

Second, in the case of Somalia a strong dissensus between the U.S. Congress and the President did not appear to have any real impact on public opinion. On October 15, 1993, about two weeks after the failed operation in Mogadishu left 19 U.S. soldiers dead, both houses of Congress passed a non-binding resolution to push President Clinton to withdraw all U.S. troops from Somalia, and threatened a cut off of funding for the operation if he failed to do so. [53] The well-informed and highly respected Democratic Senator Sam Nunn resisted calls for continued American military support to UN mission declaring that “[o]ur role is too important in areas of the world that are significant to United States military interests to allow our military effectiveness to be dissipated in places where we have no economic and no security interests.” [54] This shift in Congressional support for the operation in Somalia was not reflected by a parallel downturn in public approval for the Somali intervention. Several polls taken shortly after the October 3 debacle indicate that just over 60 percent of the respondents still supported the humanitarian aspect of the U.S. initiative in Somalia. [55] These findings were in line with levels of support for the operation since its beginning. It seems that Congress’s position on withdrawal was not paralleled by an equally strong demand on the part of the American public.

Impact of Demographics: The impact of demographic characteristics on attitudes toward U.S. foreign policy has been largely neglected in the public opinion literature. When analyzing the data from our sample, several interesting trends (or non-trends) emerge with regard to six of these categories -- those on sex, party affiliation, education, income, age, and race.

First, contrary to what one might anticipate, the data indicate that gender provides little analytic leverage in explaining public support for intervention into ethnic conflicts. Although a slight difference in preferences does exist between males and females, they are regular and relatively unchanging across cases, as well as across intervention types (Tables 6 and 7). [56] On average women only provide 5-6 percentage points less support than males for U.S. involvement in ethnic wars.

TABLE 6: Average Level of Support by Country: Impact of Sex and Race

TABLE 7: Average Level of Support by Intervention Type: Impact of Sex and Race

In light of these finding, sex cannot explain significant differences in levels of support on these issues.

Second, party affiliation in the aggregate, plays little role in forming public preferences on dealing with ethnic conflicts. This goes counter to Ole Holsti’s argument that as a general rule, party cleavages run deep on a wide range of foreign policy issues. [57] Tables 8 and 9 further demonstrate that when our surveys are broken down by type of intervention, as well as along the multilateralism-unilateralism dimension, no clear relationship between party and support exists.

TABLE 8: Party Differences and Type of Intervention

TABLE 9: Party Differences and Multilateralism

However, an interesting pattern does become apparent when specific interventions are taken into consideration. For example, Republicans were more inclined to back the effort in Somalia because it was initiated by the Bush Administration. Logically, it can be argued that Democrats favored action in Bosnia and Kosovo more than Republicans because it was a Clinton proposal. Partisanship may have had an impact during times of divided government. Thus, Republicans appear to have supported President Bush’s intervention in Somalia, but opposed Clinton’s actions in support of opinions voiced by their party leadership in Congress. This could partially explain the Republican tendency to offer less support to the efforts in the Balkans.

TABLE 10: Party Differences and Intervention by Country-Case

This appears to run contrary to the conclusions drawn from our aggregate data indicated concerning the significance of the impact of presidential cues and congressional opposition on public preferences. In this limited sample, elite influences on the public appear to be stronger when we control for both partisanship and targeted country.

Third, both income and education have a significant positive relationship with the public’s support for intervention in the ethnic conflicts in our sample. As income and education increase (both of which correlate highly together), so does the desire to get involved. On the other hand, age has a more complicated relation with support for intervention. Those ages 18 to 30 show less support for intervention than those aged 30 to 49. The widest gap is between those in the 30 to 59 age-range and the respondents over 70. When the use of military force is considered, those in the 30 to 59 year old group support military action about 44 percent of the time; however, the over 70 age group is more cautious about intervention and prefers using military force only 28 percent of the time. This seems to contradict Holsti’s recent findings, for his research findings suggest that “[a]ge based differences … were negligible”. [58]

Finally, race does appear to be a salient factor in specific cases, but not in others. Some writers detect a link between race and the level of support for intervention in certain countries or regions. Thus, Miroslav and Donna Nincic note that, “Black Americans may be somewhat more apt to approve of intervention in support of causes specifically involving countries with black populations suggesting that social identity is a broad concept with a powerful impact.” [59] On the surface, this makes sense. However, the information from our sample indicates that differences in support of using military force among blacks and whites does not appear to be generalizable.

Regarding the ethnic conflicts in Somalia and Rwanda, support for intervention among black Americans, vis-à-vis whites, reveals two interesting but clearly distinguishable patterns. With respect to the U.S. intervention in Somalia (Dec. 1992- March 1994), support among black Americans was inconsistent, and at times lower than whites (Figure 7).

At the outset of the mission, blacks (compared to whites) preferred U.S. military participation in the humanitarian relief effort by about 10 percent. This gap disappeared shortly after, and then reemerged in October 1993 when the 19 U.S. soldiers were killed in Mogadishu trying to arrest the clan leader Mohammed Farah Aideed. Far from being an overwhelming difference, black Americans on average supported action in Somalia by only 4 percent more than whites over the whole period of the military intervention. This points to a finding that the members of a U.S. racial community might not always be more supportive of intervening in countries with populations of a similar demographic background than are the members of other racial communities.

By contrast, the case of Rwanda shows a wide gap between black and white Americans. During the Rwanda genocide (roughly April 6, 1994, to July 16, 1994) U.S. African-Americans overwhelmingly supported an armed intervention to stop the mass killings (Figure 8).

In this 100-day period, 52% of blacks supported using U.S. combat troops to bring an end to the genocide, compared with only 36% of their white compatriots. This gap disappeared after the genocide, when white support for intervention rose dramatically and paralleled that of black Americans. However, polls in our sample from the other countries in Africa, as well as Bosnia and Kosovo, did not display such a striking cleavage. Unlike the gap of support that exists between men and women and between Democrats and Republicans which are minimal on military force in ethnic conflicts, the gap between whites and blacks is at times highly salient.

Published: Friday, April 30, 2004