WPS4714 Policy ReseaRch WoRking PaPeR 4714 Individual and Country-Level Factors Affecting Support for Foreign Aid Pamela Paxton Stephen Knack The World Bank Development Research Group Human Development and Public Services Team September 2008 Policy ReseaRch WoRking PaPeR 4714 Abstract In recent years donor countries have committed to attitudes. Two datasets are used to measure attitudes in dramatic increases in the supply of foreign aid to donor countries: (1) the 1995 World Values Survey has developing countries. Meeting and sustaining such information from approximately 6,000 individuals in commitments will require sufficient support among nine countries and asks a rich battery of questions at donor country voters and taxpayers. The determinants the individual-level, and (2) the 2002 Gallup "Voice of of public opinion in donor countries on foreign aid have the People" survey asks fewer questions of individuals received little attention. This paper examines attitudes to but includes 17 donor countries. Using both surveys foreign aid with a large, multi-level, cross-national study. combines their distinct strengths and allows tests of It outlines a theoretical rationale for support for foreign individual and national-level theories across disparate aid, discussing the importance of both individual factors samples. The results generally support the predictions and economic and social structures. The theory is tested that attitudes toward aid are influenced by religiosity, with multi-level models, including both individual- beliefs about the causes of poverty, awareness of level and country-level variables to predict positive international affairs, and trust in people and institutions. This paper--a product of the Human Development and Public Services Team, Development Research Group--is part of a larger effort in the department to make aid more effective in promoting economic development. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at sknack@ worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team INDIVIDUAL AND COUNTRY-LEVEL FACTORS AFFECTING SUPPORT FOR FOREIGN AID* Pamela Paxton Ohio State University Stephen Knack The World Bank *Address all correspondence to Stephen Knack, Lead Economist; The World Bank; MSN MC3-313; 1818 H St. NW; Washington DC 20433. Funding for this project was provided by the World Bank's Knowledge for Change Program. Foreign aid is an important issue for both developing and developed countries. Aid in the form of money, goods or technical assistance can develop infrastructure, strengthen institutions, or address humanitarian crises in recipient countries. Foreign aid can exceed 10% of a recipient country's national income in many instances (World Bank 2005). There is growing evidence that aid can produce positive outcomes, from democracy (Finkel, Pérez-Liñán, and Seligson 2007) to economic growth (Burnside and Dollar 2000; but see Roodman 2007). Foreign aid can help donor countries as well. In donor countries, foreign aid may be viewed as a strategic foreign policy tool (USAID 2008). When foreign aid is used by recipient countries to finance public goods, it can lead to welfare improvement in both the recipient country and in the donor country (Hatzipanayotou and Michael 1995). The supply of foreign aid to recipients is influenced by public opinion in the donor countries. Public opinion influences public policies of many types (Shapiro and Jacobs 2000; Jencks 1985; see Burstein 1998 for an excellent review). In the area of foreign aid, Mosley (1985) demonstrates that states are influenced by citizens in determining the quantity (and quality) of distributed foreign aid. Other theorists view foreign aid as a public good with a market (Dudley and Montmarquette 1976; Mosley 1985). The level of public support for aid affects not only the quality of aid but also the effectiveness of each dollar spent on aid. Paul Collier in his book, The Bottom Billion (2007: 183), asserts: "The key obstacle to reforming aid is public opinion" in donor countries. Where there is relatively little support for aid, aid agencies are overly risk averse, according to Collier, and are constrained to deliver aid in sub-optimal ways ­ e.g. on projects that provide photo opportunities for politicians. Aid in general may be more effective in reducing poverty when channeled through multilateral institutions such as the World Bank or the regional development banks, because national donor agencies such as USAID often must pursue diplomatic, commercial or other objectives that sometimes conflict with development objectives. However, where public opinion is less supportive of aid, governments may be less willing to yield control over how it is spent to multilateral institutions. Figure 1 depicts a strong relationship between support for increased aid in donor countries and the share of aid that countries funnel through multilateral agencies.1 For these reasons, it is important to understand public support for foreign aid. One existing study (Chong and Gradstein 2006) begins to explore the determinants of support for foreign aid. But two important issues remain: (1) research remains at the individual level without considering the larger social and institutional structures in which public opinion is embedded, and (2) the range of factors, even at the individual-level, that could affect support for foreign aid remains underspecified. In this paper, we argue that individual-level factors such as attitudes toward poverty, as well as country-level factors such as a country's existing level of aid, could affect public opinion in support of foreign aid. We examine support for foreign aid with a multi-level, cross-national study. We begin by outlining a theoretical rationale for the factors that should affect support for foreign aid, discussing the importance of both individual-level factors and country-level economic and social structures. We introduce two separate datasets to examine attitudes to foreign aid in donor countries: (1) the 1995-1997 wave of the World Values Survey has information from approximately 6,000 individuals in nine countries and asks a rich battery of questions at the individual-level, (2) the 2002 Gallup International "Voice of the People" survey, asks fewer questions of individuals but includes seventeen donor countries2. We test our theory with multi-level models, including both individual-level and country-level variables to predict support for foreign aid. INDIVIDUAL AND CONTEXTUAL INFLUENCES ON SUPPORT FOR FOREIGN AID Citizens of donor countries have opinions on the provision of foreign aid, defined as the transfer of capital or goods from one nation to another. Individuals can support foreign aid disbursement by their country in general. Individuals may also feel, marginally, that their country should provide more or less aid than current commitments. 1Causation is not necessarily from public opinion to how it is allocated. However, very few survey respondents are likely to know much about the share of aid channeled through multilaterals. The most plausible interpretation is that support for aid gives policymakers more latitude to spend it in ways that further development goals rather than the donor country's commercial, security or other goals. 2See www.worldvaluessurvey.org and www.voice-of-the-people.net. 2 As argued by others (e.g., Mosley 1985), we assume that the electorate is capable of forming clear opinions about their demand for foreign aid. Previous research has shown that public opinion about world affairs is structured and consistent (Chanley 1999; Shapiro and Page 1988; Page and Shapiro 1992; Hurwitz and Peffley 1987). Indeed, previous work has demonstrated that individuals hold "stable, internally consistent...generalized beliefs" about global issues such as the trustworthiness of other nations (Brewer et al 2004). Support for foreign aid can be seen as part of a general orientation to foreign policy labeled "cooperative internationalism" as opposed to "militant internationalism" (Witkopf 1990; Chanley 1999). Support for Foreign Aid: Individual Characteristics Characteristics of individuals are likely to impact support for foreign aid. For example, the religiosity of respondents ­ their attendance at religious services and the importance of religion to them ­ is likely to increase support for aid. Most religious teachings discuss relief for the poor. Jewish and Christian Biblical teachings, for example, urge individuals and leaders to support the poor, the sick, widows, travelers, and orphans. Proverbs 22:9 states: "Those who are generous are blessed, for they share their bread with the poor." Certainly, religion has been found to impact altruism of other types such as philanthropy (Wuthnow and Hodgkinson 1990) and trust (Delhey and Newton 2005; La Porta et al. 1997; Fishman and Khanna 1999), as well as a variety of economic attitudes (Guiso et al. 2003). And surveys show that demand for aid is partly altruistic ­ individuals cite moral or humanitarian obligations to help others (Mosley 1985, citing Bowles 1978). In short, there are a variety of reasons to expect that those who are more religious will express greater support for foreign aid. In a similar manner, an individual's position on poverty should influence their support for aid. Individuals likely extrapolate from their beliefs about the poor, or their general ideological position, in making decisions about supporting the poor in other countries. If individuals feel that the poor are lazy, or that they can easily escape poverty, then they are unlikely to support aid, either domestically or abroad. The theoretical argument we make here is related to low-information rationality ­ that citizens use information shortcuts to allow opinions on politics even when they lack expert 3 knowledge (e.g., Fiske and Taylor 1984). Individuals "cope with an extraordinarily confusing world...by structuring view about specific foreign policies according to their more general and abstract beliefs (Hurwitz and Peffley 1987:114). For example, economic individualism has been linked to attitudes about welfare policy (Feldman 1983). Overall, it is reasonable to assume that an individual's position on the left-right political spectrum, or their attitudes about the poor in general, can help predict their attitudes about the provision of foreign aid.3 Because foreign aid is exactly that ­ foreign ­ an individual's attention to international affairs may also affect support for foreign aid. Individuals with greater awareness of foreign and international affairs should be more likely to support foreign aid. Not only would individuals with greater exposure to events outside their own country's borders understand the issues facing other countries, they are more likely to be aware of the inequality that exists between (and within) countries. Some existing evidence suggests that individual's breadth of perspective matters for economic attitudes. For example, Mayda and Rodrik (2005) find that individuals who express stronger attachment to their neighborhood, county, region or nation are less favorable toward free trade than are respondents who express stronger attachment to their continent. Other factors should also demonstrate breadth of perspective. Certainly education and media exposure influences the views that individuals have of other nations (Page and Shapiro 1992). Certain occupations, such as being a teacher, or a member of the armed services, are more likely than others to expose individuals to international affairs. A fourth theoretical perspective suggests that certain psychological characteristics, such as a sense of agency or trust in others, are also likely to affect support for foreign aid. Personality traits are understood to develop early in the life course, influence both the perception and interpretation of events, and are generally stable, consistent, and enduring. Political psychologists therefore argue that "the influence of basic traits such as anxiety, hostility, or low self-esteem should be both pervasive and distinctive, exerting a consistent and unifying pressure upon a wide range of responses" (Sniderman and Citrin 1971:402). Indeed, Sniderman and Citrin (1971) 3Individuals on the Left may also be more likely to view foreign investment and trade as exploitative, with foreign aid the only remaining tool for developed economies to help developing countries. 4 find that self-esteem is related to isolationism. Further, previous work links trust to public opinion about world affairs (Brewer and Steenbergen 2002; Popkin and Dimock 2000: Uslaner 2002). For example, Uslaner (2002:196) argues that those who trust have "greater comfort" with individuals unlike themselves, including those from other nations. In discussing trust, it is worth noting that individuals can trust organizations and institutions as well as generalized others (Barber 1983; Giddens 1990; Paxton 1999; Brewer et al. 2004). Trust in the government, for example, has been defined as a "general orientation toward the government predicated upon people's normative expectations of government operation." (Hetherington and Globetti 2002:254; Miller 1974). Apart from trust in individuals, trust in the government or international institutions is therefore another factor that could be important in an individual's decision to support foreign aid. When an individual trusts government or other institutions, they perceive those institutions to be working well and evaluate them positively. Those who distrust government are likely to advocate restricting its activities while those who trust the government should support its activities. Institutional trust will matter for the support or rejection of any government activity, including foreign policy. However, Hetherington and Globetti (2002) argue that trust in the government is particularly important when the benefits of the government's activity accrue to others (as it would with foreign aid). Further, citizens who are confident in their government's abilities are more likely to believe it can successfully intervene in other nations (Popkin and Dimock 2000). And previous research has related pride in one's country with more positive attitudes toward trade (Mayda and Rodrik 2005). Besides trust in the government, trust in international institutions, such as the World Bank, UN or international NGOs may also impact whether an individual feels that foreign aid funds will be used wisely or wasted (see also Brewer et al. 2004). The issue is particularly relevant for citizens of donor countries who channel substantial amounts of aid through these multilateral institutions. The United States channels only about one- tenth of its aid through multilaterals. In contrast, in most years Italy channels over one- half of its aid through multilaterals. Overall, nearly a quarter of Development Assistance 5 Committee (DAC) members' foreign aid is channeled through these institutions, making public trust in them particularly relevant. Finally, demographic and life-course factors should matter for support for foreign aid. Gender is an obvious first candidate for consideration. Women and men differ in their opinions across a range of policy preferences (see Kaufman 2006 for a review). For example, women tend to be less militaristic and more opposed to war than men (Conover and Sapiro 1993). Chong and Gradstein (2006) find no differences between the genders in looking across multiple donor countries. However, there are significant gender differences within many countries, with men more supportive of aid in some countries and women more supportive in others. There is no obvious explanation for these conflicting findings. Income may also affect an individual's support for foreign aid. Those with higher incomes have the flexibility to take risks and this is likely to encourage trusting and altruistic behaviors of all kinds. Generally, some research suggests that individuals structure attitudes on economic issues along the rich/poor class schema (Hamill, Lodge, and Blake 1985). In testing foreign aid directly, Chong and Gradstein (2006) find that support for foreign aid increases with income. Another demographic factor that could change an individual's attitude about foreign aid is the significant life-course event of having a child. The direction of effect is unclear, however. On the one hand, children may act to broaden an individual's perspective, helping them be more sympathetic to others and thereby increase support for foreign aid. On the other hand, parents may feel the need to focus resources on their own children and therefore favor more domestic policies over international ones. Support for Foreign Aid: Contextual Effects To this point, we have focused on individual-level influences on support for foreign aid. But the aggregate social and institutional features of a nation may also encourage or discourage support. We must distinguish theory at the individual level (e.g., higher income increases support for foreign aid) from theory at the aggregate level (e.g., countries with higher GDP per capita display more support for foreign aid). 6 Countries vary systematically on political, economic, and cultural characteristics, all of which could impact individual levels of support. To begin, at the national level, a country's existing level of foreign aid could impact citizen opinion on foreign aid. On the one hand, citizens in countries with high existing foreign aid allocations may express more support for foreign aid in general. On the other, when considering marginal aid (the question of whether to increase foreign aid, as asked by Gallup International) citizens in countries with high levels of existing aid may express less support than citizens from other countries. Relatedly, military spending may affect attitudes toward aid in donor countries. Definitions of foreign aid often couple military and economic spending.4 Where military spending is high citizens may feel that their country is fulfilling its global citizenship duties through enhancing security. While this logic is very likely to apply in the United States, it is also likely to matter in France, the U.K. and even Canada, which specializes in international peacekeeping. Country population may matter as a country-level variant of the global awareness variables. In smaller nations, a larger share of those one can interact with are foreigners, and the greater one's awareness of foreign affairs. In contrast, in countries with larger populations, citizens are able to assist "far-off" disadvantaged people without resources ever crossing national boundaries. As an example, even if Americans know as much about the way people live 500 miles away as an Austrian or Spaniard does, the people 500 miles away are still Americans, not Hungarians or Moroccans or Portuguese. The size of a country ­ measured by population and area ­ almost inevitably makes larger countries more provincial. Similarly, citizens in donor countries with experience as a colonial power may have greater awareness about development issues in aid recipients. A history as a colonial power indicates sustained contact with a developing country. Citizens of colonial powers are likely to be better informed about events in the developing world, understand development issues facing other countries, and be aware of inequality between colonizers and the colonized. These citizens may also feel a greater sense of responsibility for the welfare of people in ex-colonies and the rest of the developing world. Citizens of colonial powers are therefore hypothesized to be more likely to 4The standard data source on aid (the OECD DAC) excludes military aid. 7 support foreign aid. Finally, in an aggregate version of the income hypothesis discussed earlier, we might expect countries with higher aggregate incomes to demonstrate more support for foreign aid. DATA AND METHODS Data come two independent sources. First, the World Values Survey (European Values Study Foundation and World Values Survey Association, 2006) contains individual-level demographic information as well as a wide range of questions capturing religiosity, an international focus, left/right position, and individual psychological characteristics. The World Values Survey (WVS) includes information for 5,693 individuals in 9 donor countries.5 To measure the dependent variable, SUPPORT FOR FOREIGN AID, individuals answered the following question: "Some people favor, and others are against, having this country provide economic aid to poorer countries." It is coded with four response categories indicating increasing favor. The second survey is Gallup International's Voice of the People 2002 survey. This survey contains many fewer individual-level variables than the WVS. It contains demographic information, as well as questions about political interest and trust of individuals and various organizations. The strength of the Gallup International (GI) survey is that it includes information for 6,194 individuals in 17 of the 22 possible donor countries.6 To measure the dependent variable, GIVE MORE FOREIGN AID, individuals answered the following question: "Do you think that the wealthier nations should give more financial help to the poorer nations or are they giving enough now?" It is coded 1/0; or "should give more" / "giving enough." Individual-Level Independent Variables In the World Values Survey analyses, the individual-level independent variables correspond to the four theories outlined above. Two variables in the WVS measure 5The countries included in the WVS analysis are Australia, Finland, Germany, Japan, New Zealand, Norway, Spain, Sweden, and the United States. 6The countries included in the Gallup analysis are Austria, Canada, Denmark, Finland, Germany, Ireland, Italy, Japan, Luxembourg, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom, and the United States. 8 religiosity: IMPORTANCE OF RELIGION ranges in four categories from "very important" to "not at all important." How often a respondent ATTENDS SERVICES is coded as days per year. Seven WVS variables capture a respondent's attention to international affairs. Individuals who respond that they are a MEMBER OF THE WORLD are distinguished (1) from those who chose towns, regions, or countries (0). NATIONAL PRIDE is measured in four categories from "very proud" to "not at all proud." INTEREST IN POLITICS is coded in four categories from the question, "How interested would you say you are in politics?" TELEVISION CONSUMPTION ranges from 0 hours of television viewing to over 3 hours per day. Two occupations with an international focus are included: members of the ARMED FORCES and professional workers including LAWYERS AND TEACHERS. Finally, respondents who were BORN IN COUNTRY are distinguished from all others. The WVS analyses also include four variables to measure Left/Right Position. First respondents were asked to place themselves on a 1 to 10 scale indicating their LEFT/RIGHT POSITION, where higher values indicate positions closer to the right. Next, three variables capture traditional left/right views on poverty. First, individuals who feel that others "are poor because of laziness and lack of will power" (POOR ARE LAZY = 1) are contrasted to those who believe others are poor because society treats them unfairly (0). Individuals who believe the POOR CAN ESCAPE POVERTY (1) are distinguished from those who believe they have very little chance (0). Finally, an individual's assessment of whether there are MORE POOR today than ten years ago is coded in three response categories. Lastly, the WVS analyses include three psychological characteristics. Respondents indicated their SATISFACTION WITH THEIR FINANCIAL SITUATION on a 1 to 10 scale with higher values indicating greater satisfaction. AGENCY is also measured as a 1 to 10 scale, with higher values indicating a feeling of "free choice and control." TRUST is assessed with the question, "generally, speaking, would you say that most people can be trusted or that you can't be too careful in dealing with people?" The WVS measures INCOME as quintiles within each country's income structure. Three other demographic variables are included in the WVS analysis: a respondents AGE, whether the respondent is FEMALE, and whether a respondent has ANY CHILDREN. 9 The Gallup International data contain fewer individual-level variables. Demographic variables record a respondent's AGE, whether the respondent is FEMALE, whether the respondent is EMPLOYED, and an individual's level of EDUCATION in three categories. POLITICAL INTEREST in the GI survey is measured with a question asking whether the respondent had contacted any public officials to provide an opinion on a public issue (1) or not (0). More questions about individual trust are available in the GI survey. TRUST is measured with the same question as in the WVS survey -- "generally, speaking, would you say that most people can be trusted or that you can't be too careful in dealing with people?" But the GI survey also asks whether individuals feel "this country is run by a few big interests looking out for themselves, or that it is run for the benefit of all the people" (RUN FOR BIG INTERESTS). Further, GI asks individuals about their trust in a variety of specific institutions: TRUST IN GOVERNMENT, TRUST IN THE UNITED NATIONS AND THE WORLD BANK, and TRUST IN NGOS (non-governmental organizations such as environmental or social advocacy groups). The trust in institutions variables are coded along a four category scale, from "a lot of trust" to "no trust at all." Contextual-Level Independent Variables Analysis is restricted to the 22 donor countries that are members of the Development Assistance Committee (DAC). Of these, 19 appear either in our analysis of WVS data (2), Gallup data (10), or both (7). Belgium, France, and Greece are missing from both surveys. Country-level variables are all taken from the World Bank's World Development Indicators dataset. Data correspond to 2002 in analysis of the GI data, and to 1995 for the World Values Survey analysis. ODA is measured as Official Development Assistance (ODA) as a share of GNP, or alternatively on the basis of aid in dollars per capita, using aid data collected by the OECD-DAC (Development Assistance Commission). Income per capita is measured using the purchasing power parity adjustments. MILITARY SPENDING is measured as a share of GNP. In adding up (and logging) the number of former COLONIES, we consider only the most recent colonizing power (e.g. Philippines counts as an American, not Spanish colony). 10 Table 1 lists the country-level variables for the larger 17-nation GI sample, the percentage of survey respondents in each country supporting increased aid, and the percentage of survey respondents who favor aid only to non-corrupt countries. Interestingly, the percentage of respondents who favor aid only to non-corrupt countries is always less than the % who favor aid in general, except in the United States. Table 1 also provides level-2 summary statistics for the GI sample. These rows demonstrate considerable variability across countries in all of these variables, including support for foreign aid. Table 2 provides bivariate correlations among support for aid and the country- level independent variables. Two of the significant correlations in this table show that survey respondents in larger countries tend to be less supportive of increasing aid, even though larger countries already tend to provide less aid. Larger countries tend to spend a larger percentage of national income on the military, consistent with collective action theories of Olson (1965) and Olson and Zeckhauser (1966). Richer countries tend to provide more aid. The Multi-Level Model To test the theories outlined above, we use multilevel models that simultaneously estimate individual- and country-level effects. The data in both surveys are hierarchically organized with individuals nested within countries, and information at the both the individual level and the country level is used to determine support for foreign aid. We use the statistical package HLM 6.04 for the analyses (Raudenbush, Bryk, Cheong, and Congdon 2001). For the WVS analysis, the full specification of level one of the multi-level model is: yij = 0j + 1j FEMALE + 2j AGE + 3j ANY CHILDREN+ 4j INCOME+ 5j IMPORTANCE OF RELIGION + 6j ATTENDS RELIGIOUS SERVICES+ 7jMEMBER OF THEWORLD + 8j NATIONALPRIDE+ 9j INTEREST IN POLITICS+ 10j TV CONSUMPTION+ 11j BORN IN COUNTRY+ 12j MILITARY+ 13j LAWYERS AND TEACHERS + 14j LEFT-RIGHT POSITION + 15j POOR ARE LAZY + 16j POOR CAN 11 ESCAPE + 17j POOR GROWING + 18j AGENCY+ 19j SATISFACTION WITH FINANCES + 20j TRUST where yij is respondent i in country j's attitude to foreign aid. Preliminary random- coefficients models (not shown) indicated that female, age, any children, income, member of the world, political interest, TV consumption, born in country, military, lawyers and teachers, and satisfaction with finances did not vary and were therefore fixed. All other coefficients were estimated as random (allowed to vary across countries). In the GI analysis, support for foreign aid is a 0/1 variable so we use a Bernoulli distribution with a logit link function for estimation. The full specification of level one of the multi-level model is: Log[pij/(1-pij)] = 0j +1j FEMALE + 2j AGE + 3j EMPLOYED+ 4j EDUCATION+ 5j POLITICAL INTEREST+ 6j TRUST + 7j RUN FOR BIG INTERESTS + 8j TRUST NGOS where pij is the probability that respondent i in country j supports giving more foreign aid (GIVE MORE FOREIGN AID =1). Preliminary random-coefficients models indicated that employed and political interest did not vary and were therefore fixed. All other coefficients were estimated as random. In analyses of both surveys, the specification of the first country-level model is: 0j = 00 + 01 ODA + u0j With only 9 countries in the WVS analysis, the level-two variables are included sequentially. Subsequent level-two models include GDP, MILITARY SPENDING POPULATION, AND LOG COLONIES in place of ODA. In GI, the same level-two variables are assessed. In both the WVS analyses and the GI analyses, all non-dummy individual-level independent variables are grand-mean centered, creating a variable with a mean of zero across all the cases. Grand-mean centering holds compositional differences in individual characteristics constant and is appropriate when aggregate versions of the variables are not included in the model. Weights are included to compensate for sampling issues in the GI analyses. HLM 6.04 uses a weighting technique developed for hierarchical data (Pfefferman et al. 1998). 12 RESULTS AND DISCUSSION Calculating the ICC demonstrates that approximately ten percent of the total variance in support for foreign aid can be attributed to between-country differences (7% in the WVS and 13% in GI).7 Thus, a nontrivial amount of the variance in support occurs across countries and it is important to explain this variation as well as the variation within countries. Indeed, a simple ANOVA indicates that there is statistically significant country-level variance in support for foreign aid in both analyses. That is, support for foreign aid varies a significant amount across countries. We begin our discussion of the results with the WVS analyses. Table 4 tests individual-level theories of support for foreign aid. These five models consider only individual-level effects but account for the clustering of individual respondents within countries with a random coefficients regression model (Bryk and Raudenbush 1992:20). A preliminary look across the models in Table 4 suggests that all four theories of individual differences in support for foreign aid find some support. First, although the IMPORTANCE OF RELIGION to an individual does not have a significant effect on support for foreign aid, ATTENDING SERVICES does have a positive and significant effect. Using the coefficient from model 2, compared to those who never attend religious services, individuals who attend at least weekly are expected to increase their support for foreign aid by .21, which across the dependent variable's four response categories is approximately ¼ of a standard deviation.8 Variables to assess an individual's attention outside their own country find mixed support in Table 2. Individuals who respond that they are a MEMBER OF THE WORLD are significantly more likely to support foreign aid than those who state that they are a member of a town, region, or country. INTEREST IN POLITICS is coded in four categories from the question, "How interested would you say you are in politics?" and has the largest effect size among this set of variables. If an individual were to increase their interest in politics from the lowest level (not at all interested) to the highest level (very interested) their support for foreign aid would increase by .36, or approximately ½ a 7For Gallup, an ICC approximation for binomial models is used (Snijders and Bosker 1999:224). 8Auxiliary models revealed no differences across denominations in support for foreign aid. 13 standard deviation. TELEVISION CONSUMPTION has a significant but small effect on support for foreign aid, at least in the range considered in the WVS. Individuals who increase their television by an hour per day increase their support for foreign aid by .04. Finally, professional workers including LAWYERS AND TEACHERS are more likely to support foreign aid than individuals in other occupations while those who are BORN IN COUNTRY are less likely to support foreign aid than those born outside the country. The other variables intended to measure attention outside an individual's country, NATIONAL PRIDE and ARMED FORCES do not significantly impact attitudes toward aid. LEFT/RIGHT POSITION has a significant effect on support for foreign aid. A one standard deviation change toward the right (approximately 2 units on the 1 to 10 scale) decreases individual's support for foreign aid by approximately .08. The three other variables assessing left/right position have very similar effect sizes although only POOR ARE LAZY and POOR CAN ESCAPE Poverty have statistically significant effects. For example, if an individual agrees with the statement that others "are poor because of laziness and lack of will power" their support for foreign aid diminishes by .12. Of the three psychological characteristics assessed in the final model, TRUST has the largest effect size. Individuals who trust support foreign aid more than individuals who do not trust others. The effect size is approximately ¼ of a standard deviation on the dependent variable. SATISFACTION WITH FINANCIAL SITUATION also has a statistically significant effect, but a smaller effect size. It would take movement from the bottom of the scale of satisfaction (1) to the top of the scale (10) to reach an equivalently sized effect. Finally, the demographic variables indicate differences in support for foreign aid by gender and age. Women support foreign aid more strongly than men, although the effect size (.06) is small. Increasing age decreases support for foreign aid, but again the effect size is quite small ­ a ten year increase in age would only lead to a .02 decrease in support. INCOME also predicts support in four of five models. Once psychological characteristics are included in the model, however, the effect size of income is approximately halved and it loses statistical significance. Having children is unrelated to support for aid. 14 The more limited individual-level Gallup International results are generally consistent with the level-1 findings from the WVS. For example, Table 5 reveals that women are more supportive of wealthier nations giving more financial help to poorer nations. Remembering that the dependent variable in the GI analysis is a dichotomous variable ("should give more" / "giving enough"), interpretation of effect sizes requires transformation of the estimated coefficient. For example, in models 5, compared to men, being a woman is expected to increase the odds of favoring more aid by 35% ((EXP[.30×1]-1)×100). An alternative interpretation focuses on the change in the predicted probability of favoring more aid. An unemployed male, who has never contacted a public official, does not trust NGOs, and has average education, age, etc. has a predicted probability of favoring aid of .72 (calculated by: 1/(1+EXP-[.95])). If that individual were instead female, her predicted probability would be .78, leading to a change in the predicted probability of .06 ­ a fairly strong effect. Although the GI survey has fewer individual-level variables in general, it does have an expanded set of questions on trust. As in the WVS analysis, TRUST has a positive and significant effect on support for more foreign aid. Declaring trust in others increases the odds of favoring aid by approximately 50%. Models 3-5 show substantial differences in the effect of trusting particular institutions -- GOVERNMENT, THE UNITED NATIONS AND THE WORLD BANK, and NGOS. Trust in one's own government does not significantly increase an individual's support for aid. Contrast the lack of effect of trust in one's own government to the significant, but small, effect of trust in international donor institutions (the UN and World Bank) and the significant and large coefficient for trust in NGOs. Each one-unit increase in expressed trust in international donors leads to a 7% increase in the odds of favoring more aid to poor countries. Each one-unit increase in trust in NGOs, however, leads to a 57% increase in the odds of favoring more aid to poor countries. Table 6 introduces country-level effects on support for foreign aid as well as the individual-level effects. Theory suggests that existing levels of aid from donor countries may have differing effects in the WVS and GI analyses, i.e. the ODA coefficients may have opposite signs. The WVS question asks respondents whether they favor provision of economic aid to poor countries, while the GI question asks whether wealthy nations 15 should increase such aid. To the extent policies reflect public preferences, existing ODA levels should be positively associated with support for aid in the WVS analysis, with causation going mostly from preferences (i.e. support for aid) to policy (i.e. levels of ODA). In the GI analysis, in contrast, high levels of existing aid may reduce support for further increases in it, producing a negative coefficient on ODA.9 Coefficients on ODA in Table 6 are consistent with these arguments. Although they are not significant at conventional levels, the ODA coefficient is positive in the WVS analysis and negative in the GI analysis. In the GI analyses it becomes highly significant if two outlying cases, the U.S. and Japan, are dropped (new coefficient = - 1.41). Interpreting this effect size suggests that if a country were to increase from an average level of foreign aid, about 0.5% of GNP, to 0.75% of GNP, the odds of a citizen's support for foreign aid would decrease by 30%. So there is some evidence for the hypothesis that citizens in countries with high levels of existing aid express less support for increasing aid than citizens from other countries. Across both surveys, the GDP of a society has a significantly negative effect on support for foreign aid. A $1000 increase in income per capita is expected to decrease the odds of individuals in that country supporting foreign aid by 4%. A one standard deviation change in income per capita ($6,500 in GI) leads to a 23% decrease in the odds. Interestingly, therefore, in contrast to the generally positive effect of socio-economic status on support for foreign aid at the individual-level, in the aggregate, richer countries do not exhibit greater support for foreign aid. The results for military spending differ across the WVS and GI analyses. Military Spending negatively effects support for foreign aid in the smaller WVS sample, but not in the larger GI sample. Even using the WVS, however, this finding is highly sensitive to inclusion of the United States. The coefficient on Military Spending is substantially reduced and becomes non-significant if the U.S. is removed from the analysis. Although U.S. citizens may view their country as fulfilling its global citizenship duties through its 9If preferences over aid levels were a sufficiently important determinant of actual aid levels, variations in aid levels among donor countries could merely reflect variations in the preferences of their respective median voters. Under this strong assumption, we should observe, in each donor country in the GI survey, a roughly equal number of respondents favoring increases and decreases in aid. In most countries, however, far more respondents support increasing than decreasing aid. 16 high military spending, this hypothesis does not generalize to other countries. In support of a "global awareness" perspective, citizens in countries with larger populations show reduced support for foreign aid. The WVS and GI samples both show a significant negative effect of population size on support for foreign aid. That is, in countries with larger populations citizens appear less willing to assist disadvantaged people across national boundaries. But the WVS result is again heavily influenced by the United States ­ the coefficient is approximately halved and loses significance if the U.S. is removed from the analysis. The GI analyses are also influenced by the U.S., but to a lesser extent. There, the coefficient remains largely stable but loses significance, reaching only p<.10. Finally, we see the more expansive colonial powers exhibit greater support for foreign aid. A 10% increase in number of former colonies is expected to increase the odds of individuals in that country supporting foreign aid by 2%. While not an extremely strong effect size, it can make a substantial difference in countries that have colonized many countries, namely the U.K., France, and Spain. Increasing colonies by 50% is expected to increase the odds of individuals in that country supporting foreign aid by 12%.10 Overall, we see that both individual-level and contextual-level variables have effects on support for foreign aid. At the individual level, we find some support for all five theoretical perspectives: demographic factors matter, as does religion, attention outside one's country, left/right position and attitudes toward poverty, and psychological characteristics. At the aggregate level, national aid disbursement impacts public opinion, as does GDP and links to former colonies. 10Auxiliary analyses considered other country-level variables, including income growth rates, income inequality, foreign-born, exports, unemployment, tax revenue, and estimates of private charitable giving. In some cases the variable was not significant in either survey, e.g., exports measured as a share of GDP. Other variables were significant in one survey and not another. For example, the percent foreign born (including refugees) was negative and significant in GI and non-significant in the WVS. In some cases, results for these variables were significant but with differing signs in the two surveys. For example, the average growth rate for the preceding 5 years has a positive effect in the GI sample, but negative in the WVS sample. Similarly, income inequality (measured by the Gini coefficient) was positive for the GI countries, but negative for the WVS sample. Chong and Gradstein (2006) obtained a similar negative effect of inequality using the WVS, but it is contradicted by findings from the larger GI sample. 17 CONCLUSION In this paper, we investigated public attitudes toward foreign aid across 19 donor countries. We argued that both individual- and country-level factors contribute to support for foreign aid. Using data from two sources, the World Values Survey, and the Gallup International Voice of the People Survey, we tested theories using multilevel models. Using two surveys allowed us to combine their distinct strengths and provide tests of our individual- and national-level theories across disparate samples. Like the sole previous study on this topic (Chong and Gradstein 2006), our paper demonstrates that individual-level factors matter to support for foreign aid. But this paper goes beyond a focus on demographic variables to test hypotheses related to a respondent's religiosity, attitudes about poverty, attention to international affairs, and psychology. Models provide support for all four proposed theories. For example, in the WVS, individuals who attend religious services, watch TV, believe the poor can escape poverty, and trust others are more likely to support foreign aid. Of note in the GI survey, individual trust in the United Nations and the World Bank appears more important to support for foreign aid than trust in one's own government. Apart from these individual-level effects, the paper highlights the contextual embeddedness of attitudes about foreign aid. The country-level results suggest that wealth and existing development support matter for individual support of aid. For example, citizens in countries with high levels of existing aid express less support for increasing aid. Citizens from countries with a history of colonization also express more support for foreign aid on average. In many ways the United States appears exceptional in these analyses. It is also the most important donor to understand, as the largest economy but with a relatively low level of aid as a share of national income. The United States spent about 0.16 percent of gross national income (GNI) in 2007, compared to the average of 0.45% among the 22 DAC members.11. If U.S. aid as a share of GNI had been equal to that of France - the median value among the other 21 DAC members ­ total DAC aid in 2007 would have been increased by nearly one-third, from $103.7 billion to about $136 billion. Increasing U.S. aid is challenging however in the face of low public support. Only 55.5% of 11See DAC data on ODA for 2007 at http://www.oecd.org/dataoecd/27/55/40381862.pdf. 18 American respondents in the WVS favored providing aid to poorer countries. Support in the other 8 donor countries in the WVS ranged from 69.3% (New Zealand) to 88.5% (Japan). Despite its relatively low levels of aid-giving, support for increasing aid, as measured in the GI survey, was also lower in the U.S. (tied with Japan at 45%) than in any of the 17 donor countries included in that survey. Our results can partially explain low support for aid in the U.S., but it remains somewhat of a mystery. The United States is an influential case for some of our level-2 variables; it is responsible for the negative coefficients for military spending and population size in the World Values Survey analysis. Military spending as a share of national income for the U.S. is about three times as high as for the other DAC donors collectively. Many Americans may view their international policeman's role as a substitute for development aid. The large size of the United States, both in population and in area, may reduce its citizens' awareness of events and conditions in less-developed countries, and a reduced sense of responsibility for the welfare of poor people abroad. Another clue to U.S. exceptionalism in aid-giving may lie in the second question asked about aid in the GI survey. The United States reports the highest percent of respondents who think no aid should be provided to countries with corruption problems. Indeed, the U.S. population is unique among the donor countries in its attention to corruption, being the only country where a higher percentage of respondents favor a more restrictive policy of giving only to non-corrupt countries. Concerns about government corruption, along with stronger traditions of (and tax incentives for) private philanthropy, likely explain Americans' relatively high levels of private giving to charitable organizations providing non-official aid.12 Numerous respondent-level attitudes and behaviors measured in the WVS were found to be associated with support for aid. On several of those variables, Americans on the whole tend toward the extremes among the 9 donor countries represented in the survey. However, some of these differences should increase support for aid among 12Americans' aid giving remains low by DAC standards even when private giving is included. Americans give about 2% of their incomes to charities, and only about 2% of that amount goes to international charities, or .04% of income. Adding this amount to U.S. ODA would not close most of the gap between U.S. and other donors' aid/GNI levels, even if private giving in other donors is ignored. Moreover, while private charitable giving is lower in other donor countries, a higher percentage of their charitable giving goes to international charities (Roodman and Standley, 2006). 19 Americans while others should reduce it, with little net effect. Low trust in people, right- of-center ideology, and beliefs that the poor are lazy and can escape poverty are all significantly associated with lower support for aid. Only 36% of Americans agree people can be trusted most of the time, lower than in 7 of the other 8 countries. The U.S. ranks first in agreement that poor people are lazy and lack willpower; it ranks third in agreement that poor people can escape poverty, and in being right-of-center ideologically. However, it also ranks at the top ­ by a wide margin ­ in church attendance, and above average on political interest and TV watching. Those three variables are all associated with stronger support for aid giving. Satisfaction with one's financial situation in the WVS analysis, and trust in NGOs and international institutions in the GI analysis, are also associated with greater support for aid. The U.S. ranks in the middle among donor countries on these variables, so they cannot account for Americans' low support for aid. It is possible that the implications of some of these attitudinal and behavioral variables differ across countries. For example, interest in politics and TV consumption may increase Americans' knowledge of foreign affairs and development issues less than it does in smaller countries. Exploring these nuances is beyond the scope of this paper, however. Foreign aid helps recipient countries develop infrastructure, strengthen institutions, and address humanitarian crises while providing an important foreign policy tool for donor nations. Paul Collier, in his recent book on foreign aid, argues that improving the effectiveness of aid requires that donors be more willing to accept risks and hence a higher rate of failure. But, "at present, the powerful force of public opinion is driving agencies in precisely the opposite direction. They cannot afford failure....aid agencies are not run by fools; they are full of intelligent people severely constrained by what public opinion permits" (2007: 184). Where support for aid is shallow and citizens are ill-informed about foreign affairs and development issues, "vociferous lobbies" may influence opinion and distort aid policies. Better understanding the factors that influence public support for foreign aid can therefore only advance the quality of distributed foreign aid in the future. 20 21 REFERENCES Barber, Bernard. 1983. The Logic and Limits of Trust. New Brunswick, N.J.: Rutgers University Press. Bowles, T. S. 1978. Survey of Attitudes Towards Overseas Development. London: H.M. Stationery Office. Brewer, Paul R. and Marco R. Steenbergen. 2002. "All Against All: How Beliefs about Human Nature Shape Foreign Policy Opinions." Political Psychology 23:39-58. Brewer, Paul R., Kimberly Gross, Sean Aday, and Lars Willnat. 2004. "International Trust and Public Opinion about World Affairs." American Journal of Political Science 48: 93-109. Bryk, Anthony, and Stephen W. Raudenbush. 1992. Hierarchical Linear Models for Social and Behavioral Research: Applications and Data Analysis Methods. Newbury Park: Sage. Burnside, Craig and David Dollar. 2000. "Aid, Policies, and Growth." American Economic Review 90:847-868. Burstein, Paul. 1998. "Bringing the Public Back in: Should Sociologists Consider the Impact of Public Opinion on Public Policy?" Social Forces 77:27-62. Chanley, Virginia A. 1999. "US Public Views of International Involvement from 1964 to 1993: Time-Series Analyses of General and Militant Internationalism." Journal of Conflict Resolution 43:23-44. Chong, Alberto and Mark Gradstein. 2006. "Who's Afraid of Foreign Aid? The Donor's Perspective." CESifo Working Paper No. 1833. www.CEDifo-group.de Collier, Paul. 2007. The Bottom Billion. New York: Oxford University Press. Conover, Pamela J. and Virginia Sapiro. 1993. "Gender, Feminist Consciousness, and War." American Journal of Political Science 37:1079-1099. Delhey, Jan and Kenneth Newton. 2005. "Predicting Cross-National Levels of Social Trust: Global Pattern or Nordic Exceptionalism?." European Sociological Review 21:311-327. Dudley, Leonard and Claude Montmarquette. 1976. "A Model of the Supply of Bilateral Foreign Aid." American Economic Review 66: 132-142. European Values Study Foundation and World Values Survey Association. 2006. European and World Values Surveys Four-Wave Integrated Data File, 1981- 22 2004. Aggregate File Producers: ASEP/JDS, Madrid, Spain/Tilburg University, Tilburg, the Netherlands. Aggregate File Distributors: ASEP/JDS and ZA, Cologne, Germany. Feldman, Stanley. 1983. "Economic Individualism and Mass Belief Systems." American Politics Quarterly 11:3-29. Fishman, Raymond and Tarum Khanna. 1999. "Is Trust a Historical Residue? Information Flows and Trust Levels." Journal of Economic Behavior and Organization 38:79-92. Finkel, Steven E., Anibal Pérez-Liñán, and Mitchell A. Seligson. 2007. "The Effects of U.S. Foreign Assistance on Democracy Building, 1990-2003." World Politics 59(3): 404-39. Fiske, S.T. and S.E. Taylor. 1984. Social Cognition. New York: Random House. Giddens, Anthony. 1990. The Consequences of Modernity. Stanford: Stanford University Press. Guiso, Luigi, Paola Sapienza, and Luigi Zingales. 2003. "People's Opium? Religion and Economic Attitudes." Journal of Monetary Economics 50:225-282. Hamill, Ruth, Milton Lodge, and Frederick Blake. 1985. "The Breadth, Depth, and Utility of Class, Partisan, and Ideological Schemata." American Journal of Political Science 29:850-870. Hatzipanayotou, Panos, and Michael, Michael S. 1995. "Foreign aid and public goods." Journal of Development Economics 47:455-467. Hetherington, Marc J. and Suzanne Globetti. 2002. "Political Trust and Racial Policy Preferences" American Journal of Political Science 46: 253-275. Hurwitz, Jon and Mark Peffley. 1987. "How are Foreign Policy Attitudes Structured? A Hierarchical Model." American Political Science Review 81:1099-1120. Jencks, Christopher. 1985. "Methodological Problems in Studying "Military Keynesianism." American Journal of Sociology 91:373-79. Kaufmann KM. 2006. "The gender gap." P.S. 39(3):447-53. La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert W. Vishny. 1997. "Trust in Large Organizations." American Economic Review, Papers and Proceedings of the 104th Meeting of the American Economic Association 87:333-38. Miller, Arthur H. 1974. "Political Issues and Trust in Government: 1964-1970." American Political Science Review 68:951-972. 23 Mosley, Paul. 1985. "The Political Economy of Foreign Aid: A Model of the Market for a Public Good." Economic Development and Cultural Change 33: 373-94. Mayda, Anna Maria and Dani Rodrik. 2005. "Why Are Some People (and Countries) More Protectionist Than Others?" European Economic Review 49: 1393-1430. Olson, Mancur. 1965. The Logic of Collective Action. Cambridge, Mass.: Harvard University Press. Olson, Mancur and Richard Zeckhauser. 1966. "An Economic Theory of Alliances." Review of Economics and Statistics 47(Aug.): 266-79. Page, Benjamin I. and Robert Y. Shapiro. 1992. The Rational Public: Fifty Years of Trends in Americans' Policy Preferences. Chicago: University of Chicago Press. Paxton, Pamela. 1999. "Is Social Capital Declining in the United States? A Multiple Indicator Assessment." American Journal of Sociology 105(1): 88-127. Pfeffermann, D., Skinner, C. J., Holmes D. J,and H. Goldstein, Rasbash, J. 1998. "Weighting for Unequal Selection Probabilities in Multilevel Models." Journal of the Royal Statistical Society, Series B 60:123-40. Raudenbush, Stephen W., Anthony S. Bryk, Yuk F. Cheong, and Richard Congdon. 2001. HLM 5: Hierarchical Linear and Nonlinear Modeling. 2nd edition. Scientific Software International. Roodman, David. 2007. "The Anarchy of Numbers: Aid, Development and Cross- Country Empirics." World Bank Economic Review 21(2):255-277. Roodman, David and Scott Standley. 2006. "Tax Policies to Promote Private Charitable Giving in DAC Countries." Center for Global Development Working Paper 82. Shapiro, Robert Y. and Lawrence R. Jacobs. 2000. "Who Leads and Who Follows? U.S. Presidents, Public Opinion, and Foreign Policy" pp 223-246 in Decisionmaking in a Glass House: Mass Media, Public Opinion, and American and European Foreign Policy in the 21st Century, edited by Brigitte L. Nacos, Robert Y. Shapiro, and Pierangelo Isernia. Lanham, MD:Rowman & Littlefield Publishers. Shapiro, Robert Y. and Benjamin I. Page. 1988. "Foreign Policy and the Rational Public." Journal of Conflict Resolution 32:211-47. Sniderman, Paul M. and Jack Citrin. 1971. "Psychological Sources of Political Belief: Self-Esteem and Isolationist Attitudes" American Political Science Review 65:401-417. 24 Snijders, T. A. B. and R. J. Bosker. 1999. Multilevel Analysis. An Introduction to Basic and Advanced Multilevel Modeling. Thousand Oaks: Sage. Popkin, Samuel L. and Michael A. Dimock. 2000. "Knowledge, Trust, and International Reasoning" pp214-38 in Elements of Reason: Cognition, Choice, and the bounds of Rationality, edited by Arthur Lupia, Matthew D. McCubbins, and Samuel L. Popkin. Cambridge: Cambridge University Press. Uslaner, Eric. 2002. The Moral Foundations of Trust. Cambridge: Cambridge University Press. U.S. Agency for International Development. 2008. "USAID History." http://www.usaid.gov/about_usaid/usaidhist.html Witkopf, Eugene R. 1990. Faces of Internationalism: Public Opinion and American Foreign Policy. Durham: Duke University Press. World Bank. 2007. Global Monitoring Report 2007. Washington DC: World Bank. World Bank. 2005. World Development Report 2005: A Better Investment Climate for Everyone. Washington, DC: World Bank. Wuthnow, Robert and Virginia A. Hodgkinson, eds. 1990. Faith and Philanthropy in America: Exploring the Role of Religion in America's Voluntary Sector. San Francisco: Jossey-Bass. 25 Figure 1. Support for Aid and Multilateral Aid Share 0.6 ) %( slaretalit 0.5 0.4 ul M 0.3 ough hrt R2 = 0.40 del 0.2 Funne 0.1 diA 0 0.4 0.5 0.6 0.7 0.8 0.9 1 Support for Increased Aid (%) Multilateral aid share is reported in World Bank (2007). Support for increased aid is from the Gallup International "Voice of the People" survey for 2002. 26 Table 1. Gallup International "Voice of the People" 2002 Sample % who favor aid Military % who favor only for non- GNP per Population spending Number of Donor Country more aid corrupt countries Aid/GNP capita (millions) % of GNP ex-colonies Australia 63% 45% 0.25 26766 19.6 1.8 0 Canada 64% 48% 0.34 28307 31.4 1.2 0 Denmark 53% 42% 0.81 28814 5.4 1.5 0 Finland 61% 31% 0.46 26629 5.2 1.2 0 Germany 68% 30% 0.36 25579 82.5 1.5 0 Ireland 87% 40% 0.42 31478 3.9 0.6 0 Italy 83% 46% 0.29 25750 57.2 2 3 Japan 45% 18% 0.28 26106 127.4 1 3 Luxembourg 71% 26% 0.82 49861 0.4 0.8 0 Netherlands 50% 29% 0.82 28604 16.1 1.5 2 Norway 63% 40% 0.94 35219 4.5 2.1 0 Portugal 89% 43% 0.21 18398 10.4 2 7 Spain 91% 29% 0.27 23119 41.3 1.2 20 Sweden 69% 33% 0.94 27005 8.9 1.8 0 Switzerland 71% 42% 0.44 31204 7.3 1 0 United Kingdom 69% 43% 0.47 27897 59.3 2.5 59 United States 45% 53% 0.22 34669 288.4 3.4 4 Mean 67% 38% 0.49 29141 45.3 1.6 5.8 Standard Deviation 14% 9% 0.26 6665 71.5 0.7 14.6 Minimum 45% 18% 0.21 18398 0.4 0.6 0 Maximum 91% 53% 0.94 49861 288.4 3.4 59 27 Table 2. Correlations among country-level variables, Gallup International sample (N = 17) Military Number ex- Aid/GNI GNP p.c. Population spending colonies Favor more aid -0.23 -0.23 -.45* -0.26 0.18 Aid/GNP 1 .48* -.45* -0.12 -0.15 GNP p.c. 1 0.03 -0.12 -0.17 Population 1 .62** 0.1 Military spending 1 0.34 * or ** indicates significance at .10 or .01 level respectively for 2-tailed tests. 28 Table 3. World Values Survey Individual-level Models Model 1 Model 2 Model 3 Model 4 Model 5 intercept 2.914 *** 2.948 *** 2.955 *** 2.944 *** 2.863 *** (.075) (.08) (.073) (.069) (.079) Demographic female 0.071 ** 0.060 * 0.071 * 0.065 * 0.055 * (.021) (.024) (.025) (.026) (.025) age -0.001 -0.002 t -0.002 t -0.001 -0.002 t (.001) (.001) (.001) (.001) (.001) any children -0.053 t -0.049 -0.030 -0.027 -0.008 (.028) (.028) (.027) (.028) (.028) income 0.055 *** 0.057 ** 0.032 * 0.040 ** 0.022 (.014) (.014) (.011) (.013) (.013) Religion Importance of Religion 0.032 0.028 0.036 0.037 (.021) (.021) (.021) (.021) Attends Services 0.004 * 0.003 t 0.004 * 0.003 * (.002) (.002) (.002) (.001) Attention outside country Member of the World 0.170 ** 0.150 ** 0.146 ** (.047) (.045) (.048) National Pride -0.023 -0.012 -0.018 (.024) (.023) (.021) Interest in Politics 0.102 *** 0.096 *** 0.090 *** (.015) (.015) (.015) TV Consumption 0.052 ** 0.053 ** 0.044 * (.013) (.015) (.014) Born in Country -0.081 -0.093 t -0.094 t (.055) (.049) (.049) Armed Forces -0.093 -0.064 -0.029 (.159) (.148) (.15) Lawyers and Teachers 0.164 *** 0.153 ** 0.126 * (.037) (.041) (.044) Left/Right Position Poor are Lazy -0.117 t -0.119 * (.054) (.051) Poor Can Escape Poverty 0.134 ** 0.109 ** (.033) (.032) Left / Right Position -0.044 *** -0.043 *** (.009) (.008) More Poor -0.045 -0.039 (.03) (.029) Psychological Characteristics Agency 0.007 (.01) Satisfaction with Financial Situation 0.021 ** (.006) Trust 0.190 ** (.047) t .05 one-tailed; * .05 two-tailed; ** .01 two-tailed; *** .001 two-tailed Ni = 5,693 Nj = 9 29 Table 4. Gallup Survey Individual-level Models Model 1 Model 2 Model 3 Model 4 Model 5 intercept 0.86 *** 0.86 *** 0.93 *** 0.93 *** 0.95 *** 0.16 0.17 0.18 0.18 0.17 Demographic education 0.23 * 0.23 * 0.12 0.12 0.06 0.09 0.08 0.09 0.09 0.08 female 0.27 ** 0.27 ** 0.30 ** 0.30 *** 0.30 ** 0.07 0.07 0.08 0.08 0.08 age -0.01 * -0.01 * -0.01 * -0.01 * -0.01 0.00 0.00 0.00 0.00 0.00 employed -0.24 ** -0.24 ** -0.28 ** -0.28 ** -0.28 ** 0.07 0.07 0.08 0.08 0.08 Political Interest contacted public officials 0.00 0.02 0.02 -0.04 0.08 0.09 0.09 0.10 Trust trust 0.42 *** 0.42 *** 0.39 *** 0.08 0.08 0.09 run by big interests -0.16 t -0.17 * -0.16 t 0.08 0.08 0.08 trust government 0.10 0.06 trust UN and WB 0.07 ** 0.02 trust NGOs 0.45 *** 0.06 t .05 one-tailed; * .05 two-tailed; ** .01 two-tailed; *** .001 two-tailed Ni = 6,194 Nj = 17 30 Table 5. Country-level Results, World Values Survey and Gallup Survey WVS Gallup WVS Gallup WVS Gallup WVS Gallup WVS Gallup ODA 0.073 -0.661 a 0.177 0.396 GDP -4E-05 ** -4E-05 * 0.000 0.000 Military Spending -0.150 **b -0.090 0.047 0.138 Population -0.002 ***c -0.003 ***d 0.00 0.001 Log Colonies 0.024 0.218 * 0.045 0.100 t .05 one-tailed; * .05 two-tailed; ** .01 two-tailed; *** .001 two-tailed a Coefficient becomes significantly negative (-1.41, p<.001) if two outlying cases, the U.S. and Japan, are removed from the analysis. b Coefficient is reduced and becomes nonsignificant if an outlying case, the U.S., is removed from the analysis. c Coefficient is reduced and becomes nonsignificant if an outlying case, the U.S., is removed from the analysis. d Coefficient is unchanged but reduces significance (p<.1) if an outlying case, the U.S., is removed from the analysis. Note: All WVS models include the 20 variables at level 1. All Gallup models include the 8 variables at level 1. 31