Too Poor to Care? The Salience of AIDS in Africa

Abstract

Sub-Saharan Africa is the part of the world that is most severely affected by HIV/AIDS. Yet, surveys of attitudes to AIDS across African countries show that voters do not attach great importance to the issue. This paper argues that the low salience of AIDS is in large part a result of the widespread poverty that exists on the continent, which means that AIDS is crowded out by other issues. Using multilevel regressions, results from four waves of survey data support the hypothesis that poverty significantly affects the salience of AIDS as a political issue at the individual level.

Keywords

AIDS poverty policy salience public opinion Africa

In the last couple of decades, HIV/AIDS has evolved into a global pandemic with disastrous human and economic consequences. Recent statistics show that as of 2010, approximately 35 million people were infected with HIV worldwide (Joint United Nations Programme on HIV/AIDS [UNAIDS] 2013). In terms of mortality, the total death toll from the disease amounts to about 36 million people since the first outbreak was discovered thirty years ago.¹ In 2012 alone, AIDS caused around 1.6 million deaths (UNAIDS 2013, 4). While the disease affects many developing countries, Africa is—by far—the most severely hit region in the world (Barnett and Whiteside 2006; Iqbal and Zorn 2010; Justesen 2012; Patterson 2006).² With only 10 percent of the world’s population, Africa is home to two-thirds of all recorded people living with HIV—amounting to more than 22 million people (UNAIDS 2013).³ In comparison, Latin America accounts for only 5 percent of HIV infections worldwide, while in Asia the corresponding number is 20 percent (UNAIDS 2013). To make matters worse, in the most severely affected countries—Botswana, Lesotho, and Swaziland—approximately a fourth of the adult population is infected with HIV (UNAIDS 2013). The social and economic implications of the HIV/AIDS crisis are profound. Apart from the obvious human suffering caused by the disease, life expectancy has declined rapidly, mortality rates have increased, and economic growth suffers, too (Barnett and Whiteside 2006, 297–99).

The existing evidence leaves little doubt that the HIV/AIDS pandemic has a major impact on the lives of millions of Africans. It contributes to keep many African countries trapped in poverty and disease, and threatens to transform whole societies. As emphasized by Hyden (2006, 90), “The AIDS epidemic competes with globalization as the main cause of social change in Africa today . . . because it hits more directly than economic forces at the very core of the continent’s social structure.” Indeed, HIV/AIDS not only affects the people carrying the disease, but also their families, friends, and workplaces, for example, through increased absenteeism (Barnett and Whiteside 2006, 264–65).

In spite of this, HIV/AIDS does not figure prominently on the public agenda in Africa (Bratton, Mattes, and Gyimah-Boadi 2005, 102; de Waal 2006, 42–45; Dionne 2012; Patterson 2006, 63; Swidler and Watkins 2009). It appears paradoxical that such a life-and-death issue as AIDS is not, in general, given a higher priority by Africans, and that, compared with other issues, the salience of AIDS is low even in many highly affected countries. In other words, the question is, “Why is the salience of AIDS so low in Africa?”

While there are obviously many factors—such as national HIV prevalence and personal loss due to the disease—that affect AIDS salience, the explanation emphasized in this paper is that the low salience of the AIDS issue is in large part a consequence of the widespread poverty that exists throughout the African continent. The main worry for poor people is often coping with the lack of basic material necessities on a short-term basis. As it is possible to live with HIV for years without significant symptoms (Dionne 2011, 59), there is less reason to worry about the threat of dying from AIDS years into the future for people facing acute material constraints. In terms of the salience people attach to political issues, this means that poor people are likely to worry more about policies that address immediate concerns such as hunger and food shortage than policies, such as prevention and treatment of HIV/AIDS, which may yield tangible benefits in a relatively distant future.

Consequently, this paper provides a novel explanation of why the salience of AIDS is so low in Africa, despite the fact that it is the most severely affected region in the world. In doing so, the paper makes two novel contributions to the literature. First, while previous work by Whiteside et al. (2002) and Patterson (2006) suggest that economic deprivation may be linked to AIDS salience, this paper develops a theoretical argument emphasizing the specific mechanisms linking individual-level wealth and poverty to the importance people attach to AIDS. Second, the paper provides what appears to be the first systematic, quantitative analysis of the hypothesis linking poverty to AIDS salience. It does so using survey data from four waves of the Afrobarometer, covering a time period from 1999 to 2009. As the surveys contain individual-level data from up to 20 African countries, the empirical tests are conducted using multilevel regression models that examine the simultaneous impact of individual- and country-level variables on the salience of AIDS. To examine the generic implications of the theoretical argument, the paper also tests the link between poverty on the one hand and the salience of unemployment, poverty, and food shortage on the other hand.

The paper therefore contributes to the small but growing literature trying to explain the status of the AIDS issue on the agenda of the public and their political representatives in Africa (de Waal 2006; Dionne 2011, 2012; Lieberman 2007; Patterson 2006; Whiteside et al. 2002; Youde 2012). In a broader perspective, the paper is also related to the literature on policy salience and government responsiveness (Hobolt and Klemmensen 2008; Soroka and Wlezien 2010). Unlike this literature, however, the paper is not directly concerned with the link between issue salience and government responsiveness, but tries to explain why the salience of AIDS varies across individuals and countries.

The remainder of the paper is organized as follows. The next section “Linking Poverty to Policy Salience” develops the theoretical argument and outlines the key hypotheses. The section “Data and Method” describes the data and methods used in the empirical part. The section “Results” after that presents the empirical results, and the final section “Conclusion” concludes on the main findings.

Linking Poverty to Policy Salience

While the salience of AIDS in Africa is generally low, it varies over time and across countries. Based on four waves of surveys conducted by the Afrobarometer in up to twenty African countries from 1999 to 2009, Figure 1 shows the percentage of people that—over time—considers AIDS one of the three most important problems their government should address.⁴

Figure 1.

Most important problem: Four waves of the Afrobarometer.

For comparison, Figure 1 also shows three issues—unemployment, poverty, and food shortage—that are frequently ranked as the most salient by this standard. While the salience of AIDS varies somewhat over time, it is clear from Figure 1 that African citizens do not generally rate AIDS among the most important political issues requiring government action. Indeed, AIDS is consistently ranked much lower than unemployment, poverty, and food shortage in the countries covered by the Afrobarometer.

The salience of AIDS also varies between countries. Table 1 shows statistics for the salience of AIDS across the twenty countries in the fourth wave of the Afrobarometer.

Table 1.

HIV/AIDS Salience in Africa, 2008/2009.

		Most important problems
	HIV prevalence	HIV/AIDS	Unemployment	Poverty	Food shortage
All countries	8.2	4.2	36.1	27.0	23.3
Botswana	24.5	10.9	64.3	40.2	10.5
Lesotho	22.7	4.1	50.2	28.4	29.1
South Africa	17.5	20.5	57.3	23.3	8.7
Zimbabwe	15.7	0.4	27.2	12.8	26.7
Namibia	14.0	9.6	59.6	32.3	9.0
Zambia	13.3	1.2	29.9	28.3	18.3
Malawi	11.9	2.1	8.8	25.1	60.0
Mozambique	11.5	4.8	37.0	26.0	19.8
Uganda	6.7	4.7	26.4	42.5	20.3
Kenya	6.3	2.1	32.4	23.6	18.6
Tanzania	5.7	3.4	5.1	18.3	15.8
Nigeria	3.5	2.0	47.7	34.2	25.0
Ghana	1.7	0.8	47.2	20.5	3.2
Mali	1.1	0.7	11.0	27.7	53.1
Liberia	1.4	0.3	32.7	13.3	26.9
Benin	1.3	0.7	23.2	21.9	21.7
Burkina Faso	1.1	0.5	17.1	26.6	37.0
Madagascar	0.6	0.4	18.6	27.3	23.6
Senegal	0.6	0.3	28.1	11.6	46.4
Cape Verde	0.3	1.0	76.9	36.3	9.1

HIV prevalence is the percentage of the adult population with HIV in 2008 (matching the year of the fourth wave of the Afrobarometer). MIP data are from the fourth wave (2008/2009) of the Afrobarometer. Countries in bold have generalized HIV epidemics, defined as prevalence rates > 5 percent (Barnett and Whiteside 2006, 101). Numbers in rows for “most important problem” are respondents’ answers (in percentages) to the question of what constitutes the most important problems the government should address. Numbers show the proportion of first, second, and third responses to this question. Numbers do not sum to 100 because respondents can give up to three answers. MIP = Most important problem.

It is striking that in no country—not even the most severely affected—does AIDS top the list of the most salient problems. Indeed, across all twenty countries, only around 4 percent of respondents mention AIDS as one of the most important political problems. Given the social and economic impact of AIDS, this is a seemingly very low number, which is lower than the percentage of people mentioning transportation (6.1 percent) as an important problem. However, in South Africa, more than 20 percent consider AIDS an important problem, and in Botswana and Namibia, approximately 10 percent of respondents mention AIDS as one of the most important problems. This suggests that AIDS may be a salient issue only in countries with generalized epidemics; that is, where HIV/AIDS has spread into the wider population (Barnett and Whiteside 2006, 101; UNAIDS 2008, 100). However, in countries such as Lesotho, Zimbabwe, Zambia, Malawi, and Mozambique that suffer from generalized epidemics and double-digit prevalence rates, AIDS is a much less salient issue. In comparison, unemployment, poverty, and food shortage are considered important problems by much larger proportions of the population in most countries. Even in highly affected countries such as Botswana, Lesotho, Mozambique, South Africa, Zambia, and Zimbabwe, unemployment and poverty rank higher than AIDS on the list of important problems.

On the one hand, the numbers in Table 1 reflect the fact that the political agenda in Africa is crowded because of the many economic problems related to poverty and unemployment that continue to plague the region. On the other hand, this also raises the question of why HIV/AIDS is not a more salient political issue for voters in Africa.

It is widely accepted that the salience of policy issues matters for whether they appear on the public agenda (Hobolt and Klemmensen 2008; Soroka and Wlezien 2010). A related, quite fundamental, question is why the salience of different policy issues varies across individuals and national contexts. Intuitively, it would seem that issues and problems that deeply affect the everyday lives of large groups of people—sometimes referred to as “doorstep issues” (Converse 1975; Zaller and Feldman 1992)—would be prime candidates for high-salience issues on the public agenda. In this respect, the AIDS issue in Africa is an interesting case because the salience of the issue is surprisingly low, despite the fact that HIV/AIDS directly affects large parts of the population in many African countries.

The Economics of AIDS Salience

Recent estimates show that around 50 percent of the population in Africa live in poverty, defined as incomes below US$1.25 a day (United Nations [UN] 2013, 6–7). For these groups, poverty, low income, and poor material living conditions in general are not only constraints on their consumption opportunities; they also affect which political issues people are mainly concerned with.

The mechanism linking wealth and poverty to the salience of AIDS rests on the assumption that the marginal utility of consumption is higher for poor and low-income groups compared with wealthier groups (Chakraborty 2004; Przeworski and Limongi 1997). This is parallel to saying that for poor people, the marginal utility from current consumption is higher than the marginal utility from future consumption. Poor individuals, therefore, tend to develop short time horizons, precisely because poverty has severe short-term consequences that reduce the probability of surviving into the future (Azariadis 2006, 20–21; Chakraborty 2004, 120). By implication, poor people will tend to spend most of their economic resources on consuming basic goods such as food, shelter, and clean water to cover their short-term needs. In contrast, more well-to-do groups care relatively more about future consumption, and typically do not need to worry too much about their short-term material needs.

Variations in poverty and material living conditions are also likely to affect the importance people attach to particular policy issues. If poor people discount the value of policies that increase consumption in the future, they are also more likely to attach high salience to policy issues that yield tangible, short-term benefits. Poverty, therefore, imposes an uncomfortable intertemporal trade-off, given by the choice between giving priority to policies that produce immediate benefits versus policies that yield payoffs in some distant future. In the extreme, this means that poor and destitute people worry more about surviving until tomorrow, and less about issues affecting them years into the future (Azariadis 2006; Chakraborty 2004).

These considerations are important for understanding the status of AIDS on the public agenda in Africa. Recent research suggests that, in the absence of treatment, the median survival time after infection with HIV is between nine and eleven years (UNAIDS 2008, 32). Even without treatment with antiretroviral drugs, HIV-positive people can survive for several years without significant symptoms before developing and dying of AIDS (Dionne 2011; UNAIDS 2008). Consequently, while hunger and poverty-related disease can kill you today or in the near future, HIV/AIDS will kill you only in the longer term. The salience of AIDS is, therefore, likely to be low for poor groups and increase with increasing levels of wealth, while issues such as food shortage and unemployment will be much more salient among poor people.

Pushing AIDS off the Public Agenda

The implication of this argument is that AIDS as a political issue does not figure highly on the agenda in Africa because many people are poor and compelled to worry about pressing material concerns. Relative to other problems, AIDS, therefore, becomes a neglected issue (Bratton, Mattes, and Gyimah-Boadi 2005, 102; de Waal 2006; Patterson 2006; van de Walle 2001, 85–86; Whiteside et al. 2002). This leads to a situation where large groups of voters, because of the strains of poverty, do not consider AIDS an important issue, at least not relative to other issues. Instead, the public agenda will tend to be dominated by issues such as poverty and unemployment, because these are perceived to be more urgent problems. Of course, the salience of a particular issue does not necessarily determine whether governments respond to public opinion, particularly in African countries where democratic institutions are often weak (Hyden 2006; Jensen and Justesen 2014). Even so, if AIDS is not salient to voters—or at least less salient than other issues—it is unlikely to attract even some attention from politicians and governments, who, in turn, face fewer incentives to respond effectively to the problems caused by HIV/AIDS. Ultimately, this implies that neither the public nor their political representatives pay much attention to the issue, despite its fatal consequences. In this perspective, it is not surprising that HIV/AIDS has not been given much attention in recent elections in many African countries (de Waal 2006; Patterson 2006).

If the argument of this paper is valid, the widespread poverty that exists in most African countries is an important part of the explanation of the low salience of AIDS. This argument can be transformed into two empirically testable hypotheses:

Hypothesis 1: The salience of AIDS is lowest among poor groups and increases with decreasing levels of poverty.

Hypothesis 2: The salience of poverty, hunger, and unemployment is highest among poor groups and decreases with decreasing levels of poverty.

Both hypotheses are specific implications of a generic argument. Indeed, while Hypothesis 1 relates specifically to the case of AIDS, the point of Hypothesis 2 is precisely to test the broader implications of the theoretical argument on three of the most salient issues across all waves of the Afrobarometer (cf. Bratton, Mattes, and Gyimah-Boadi 2005, 89–103).

Data and Method⁵

To test the hypotheses in an African context, the paper uses data from four waves of the Afrobarometer. Specifically, I use data from the first, second, third, and fourth waves of the Afrobarometer, containing questions that are similarly phrased and comparable across countries and over time. The first wave data were collected in 1999–2001 and covers twelve countries; the second wave covers 2002–2003 for sixteen countries; the third wave was conducted in 2005–2006 for eighteen countries; and the fourth wave covers twenty countries in 2008–2009. The list of countries included in the four waves is available in the online appendix (http://prq.sagepub.com/supplemental/). Data from these four waves contain a total of sixty-six country surveys that are merged into one dataset, providing a set of repeated surveys within countries, with more than eighty thousand respondents surveyed in up to twenty countries during the period from 1999 to 2009. Therefore, the data contains both individual- and country-level variation over time and across national contexts.

Doing survey research in new democracies in the developing world is obviously much more challenging than doing surveys in Europe and North America (Mattes 2007). However, the Afrobarometer has gone through great lengths to ensure that the data quality is as high as possible. The procedures and rules for sampling and data collection are described in detail in a Survey Manual for each wave of the survey (e.g., Afrobarometer 2007). The sample design is based on a stratified multistage procedure that produces a randomly selected and broadly representative cross-section of adult individuals within each country (Bratton, Mattes, and Gyimah-Boadi 2005, 54). To achieve a random, nationally representative sample, the most recent census statistics are used as the sampling frame. At a minimum, information is obtained on the population size of each country, and how the population is distributed across subnational regions, and across urban and rural areas within regions (Afrobarometer 2007, 26). This is intended to minimize the chance that people from particular regions or with distinct ethnic, linguistic, or religious affiliations are not sampled. The standard sample size is 1,200 respondents for each country, but increases to 2,400 in a few highly fractionalized countries (e.g., South Africa, Uganda, and Nigeria). Interviews are done face-to-face in respondents’ local language. The questionnaire is standardized and contains the same questions in all countries (possibly plus some country-specific questions). However, it is important to note that the countries in the Afrobarometer are not a random sample from the sub-Saharan region. Therefore, the results should not be generalized to the region as a whole. A more detailed description of the Afrobarometer data is available in the online appendix.

Dependent Variable

The salience of AIDS is measured using the following survey question: “In your opinion, what are the most important problems facing this country that the government should address?” This question is widely used in the literature to evaluate the salience voters attach to different political issues; that is, the extent to which voters give priority to a given issue and consider it important (Hobolt and Klemmensen 2008; Soroka and Wlezien 2010). The question was posed to respondents as open-ended, meaning that respondents can answer the question using their own words, without a fixed and predefined set of response categories. Responses were subsequently grouped into categories such as HIV/AIDS, unemployment, food shortage, and so on (Afrobarometer 2006; Bratton, Mattes, and Gyimah-Boadi 2005, 98–102). An advantage of the wording of the “most important problem” (MIP) question is that interviewers were instructed to let respondents name only up to three problems, which forces people to prioritize and select those problems they consider the most important. However, we cannot know with confidence if the first answer is considered the most important problem, the second answer the second-most important problem, and so on. Nor do we have any way of knowing which of the problems respondents find most important in absolute terms, or how much more importance is attached to one problem relative to another (Eifert, Miguel, and Posner 2010, 498).⁶ Therefore, AIDS salience is coded as a dichotomous variable: if a respondent mentions “HIV/AIDS” as one of the three most important problems, the variable is coded as one (1), and zero (0) otherwise. This metric is used to measure the salience individuals attach to the HIV/AIDS issue.

However, the MIP question is not an entirely unproblematic indicator of salience (Soroka and Wlezien 2010; Wlezien 2005). For instance, a policy “issue” may be important to voters, even though it is not considered a “problem” (Soroka and Wlezien 2010, 101). Nevertheless, the MIP question is the standard way to measure the salience of policy issues on the public agenda in political science (Hobolt and Klemmensen 2008; Soroka and Wlezien 2010) and also seems to be the best available indicator of the priorities citizens give to different policy problems, including HIV/AIDS, in Africa.

To test Hypothesis 2, the salience of unemployment, poverty, and food shortage have also been coded into binary variables, following the same steps described above. These issues are useful for testing the reverse side of theoretical argument—that poor people care more about issues that affect them in the short term. Summary statistics for the dependent variables are shown in the online appendix.

Explanatory Variable: Poverty

The best way to assess the material living conditions of respondents in the Afrobarometer survey is to construct a direct measure of poverty (Bratton 2008; Bratton, Mattes, and Gyimah-Boadi 2005; Mattes, Bratton, and Davids 2003). The Afrobarometer contains six questions concerning poverty, construed as people’s experience with lack of basic necessities. Specifically, respondents are asked how often during the past year, they or anyone in their family have gone without (1) enough food to eat, (2) enough clean water for home use, (3) medicines or medical treatment, (4) enough fuel to cook food, or (5) a cash income.⁷ Answers are given using a five-point scale from never to always. Following the standard in the literature, I combine these six items into an index of poverty (Bratton, Mattes, and Gyimah-Boadi 2005; Jensen and Justesen 2014; Justesen and Bjørnskov 2014; Mattes, Bratton, and Davids 2003).⁸ The strength of this index is that it assesses poverty in terms of individuals’ experience with lacking basic material necessities such as food and water on a regular basis. High values on the index reflect that people are poor, whereas low reflect that people are well-off in material terms.

The macro implication of the theoretical argument is that people in wealthier countries are expected to consider AIDS a more salient issue. To test the impact of variations in national wealth, a measure of GDP per capita is included in the regressions. Data are from the Penn World Tables 7.1 (Heston, Summers, and Aten 2012). Summary statistics are available in the online appendix.

Control Variables

To guard against spurious correlations, the regressions include a number of individual- and country-level controls. At the individual level, up to twelve variables are included. Three variables aim at capturing cognitive and informational aspects of AIDS salience. The first is a measure of cognitive political engagement. This variable equals the sum of two items measuring interest in public affairs and the frequency with which respondents discuss politics with friends or family. If people express interest in public affairs and discuss political issues on a regular basis, they may be more likely to view AIDS as an important political issue. Including this variable also minimizes concerns that the poverty measure captures the effect of low levels of cognitive political engagement among poor people. Second, a measure of how often respondents receive news from the radio is included as a proxy for information. Controlling for information access is important because poor people may give lower priority to AIDS because of limited access to information rather than short time horizons. This variable is measured on a five-point scale, with high values indicating that respondents frequently listen to radio news. Third, a measure of individuals’ educational background is included with the expectation that higher levels of education may raise awareness to the AIDS issue and increase the capacity of respondents to follow politics. Education is measured using two dummy variables: one for respondents with a primary school background, and one for respondents with secondary schooling or higher. Respondents with no (formal) schooling are the reference group.

Open discussion on AIDS is considered taboo in some communities, and religion in particular seems to affect people’s inclination to associate AIDS with taboo and social stigmatization (de Waal 2006; Smith 2004). This may make people less inclined to mention HIV/AIDS as an important problem and can create a downward bias in the observed salience of the issue. To capture the effect of religiosity, a variable measuring whether people attend religious services (or are members of a religious group) is included, assuming that religious people attend religious services more often. Specific religious affiliations may also to be related to individuals’ attitudes and reactions to AIDS (de Waal 2006; Smith 2004). Therefore, three dummy variables are constructed for Catholics, Muslims, and people identifying with other religions (including Protestants). The reference group is respondents not adhering to any religion. Admittedly, including these variables is an imperfect way of dealing with religiously given attitudes and taboos related to HIV/AIDS, but it is the best that can be done given the data at hand. Data on religious affiliations are not available in the first wave of the survey and are, therefore, included only in one regression.

AIDS salience may also be affected by people’s personal experience (Dionne 2012). For instance, if respondents have known people who have died of AIDS, they are more likely to consider the disease an important problem requiring government action. Therefore, a dummy variable is included for respondents with close friends/relatives who have died of AIDS. However, because including this variable reduces the number of observations, it enters only in two regressions.⁹

Finally, a series of demographic control variables are included. A dummy variable indicating whether respondents live in urban (1) or rural (0) areas accounts for the possibility that HIV prevalence is higher in urban areas, and may therefore be more salient among urban dwellers. Moreover, controlling for urban versus rural residence is important to ensure that the poverty variable does not capture effects of differences in poverty levels between people in urban and rural areas. The regressions also control for gender (females = 1) and age (in years), as HIV prevalence is higher among women and young people in Africa (Barnett and Whiteside 2006, 10).

At the country level, the number of control variables is limited because the surveys contain data for at most twenty countries per wave.¹⁰ As four waves of survey data are used, the country-level variables vary both across and within countries. That is, the values of the country-level variables are matched to the year of the survey (e.g., 1999 for the first wave). Four country-level control variables are included. First, a variable measuring the nationwide prevalence of HIV is included because HIV/AIDS is likely to figure higher on the public agenda in countries with high levels of HIV prevalence. Data are from the World Development Indicators.

Second, the inflow of foreign aid for health purposes—including HIV/AIDS—to Africa has increased substantially in recent years. This may result in more widely available treatment and health facilities, but may also affect the way ordinary Africans respond to and perceive the issue (Swidler and Watkins 2009). Large inflows of health aid are often accompanied by an increase in donor and non-governmental organization (NGO) activity in the cities, towns, and villages of Africa. Using the case of Malawi as an example of donor influence, Swidler and Watkins (2009) report that large donations related to HIV/AIDS programs are often announced in newspapers and the radio—regular news sources for many Africans. Moreover, rural villagers also “. . . see signs of donor wealth in the 4 × 4’s with NGO logos zipping along the roads, or stories of a friend of a friend who got 825 kwacha/day at a three-day NGO workshop” (Swidler and Watkins 2009, 1185). This heavy donor and NGO involvement in HIV/AIDS-related activities may imply that many Africans do not think HIV/AIDS is an issue that national governments are responsible for—or that it is a problem that the national government even has the capacity to address. That is, because the large role played by foreign aid, external donors, and NGOs in combatting HIV/AIDS in Africa, ordinary Africans may think of international agencies as being responsible for dealing with HIV/AIDS, and be less inclined to attribute responsibility for HIV/AIDS policies to national governments. To address this possibility, a variable measuring foreign aid for health purposes is included. Foreign aid for health is measured per capita, and data are from the Institute for Health Metrics and Evaluation (2010).

In a related manner, citizens’ perceptions of the ability of governments to deliver solutions to problems such as HIV/AIDS may also matter. Therefore, a measure of government effectiveness from the Worldwide Governance Indicators is included (Kaufmann, Kraay, and Mastruzzi 2010). This variable measures (perceptions of) the quality of public services and the ability of governments to implement policies, with high values indicating effective government.

Finally, AIDS salience may also be shaped by policies and opinions expressed by political leaders. For instance, voters attitudes are likely to respond to the policies adopted by politicians in specific domains (Soroka and Wlezien 2010, 22) and may be affected by the framing of issues by political leaders (Druckman and Lupia 2000). Moreover, if government policy is already effectively addressing HIV/AIDS, respondents may worry less about the issue. I therefore include data from the AIDS Effort Program Index (API), developed and compiled by the United States Agency for International Development (USAID), Joint United Nations Programme on AIDS/Acquired Immune Deficiency Syndrome (UNAIDS), World Health Organization (WHO), and the POLICY project. The index combines information on policies and political responses to AIDS in ten areas (Patterson 2006, 25), including, for example, assessments of whether political leaders publicly address AIDS, whether government have national policy plans addressing HIV/AIDS, and the existence of prevention and treatment programs, for example, involving antiretroviral medicines (a full description of the index is available in the online appendix). The API scores, therefore, capture the combined effect of AIDS policies and political leadership on the salience voters attach to AIDS. Including a measure of AIDS policy obviously gives rise to endogeneity concerns, as it may both affect and be affected by salience. Moreover, API data are not available for Cape Verde and Liberia, and only for 2003. Therefore, the API is included only in one model. The index is measured on a scale from 0 to 100, with high values reflecting the most extensive AIDS policies. Further variable descriptions and summary statistics are available in the online appendix.

Estimation Method: The Multilevel Model

Given that individuals are surveyed in different national contexts, the paper uses a multilevel/hierarchical model to estimate the impact of poverty on AIDS salience. These models allow us to estimate the simultaneous impact of individual- and country-level variables while taking unobserved group-level heterogeneity into account (Gelman and Hill 2007; Rabe-Hesketh and Skrondal 2008; Steenbergen and Jones 2002). As the data consist of repeated surveys of individuals within different countries, we could in principle have a three-level model with individuals nested in the year of the survey, which again is nested in countries. However, with only four waves of the survey, we treat individuals as level 1 units, countries as the level 2 units, and then control for survey-wave (year) fixed effects.

Because the dependent variable is dichotomous, all models are estimated using logistic multilevel regressions. Applying the logit link function and indexing individuals with i (i = 1,…, N), countries with j (j = 1,…J), and survey-waves by t (t = 1,…,T), the individual-level (level 1) equation becomes

\begin{array}{l} \Pr (y_{i j t} = 1) = {logit}^{- 1} \\ (α_{j t} + β_{1} {Poverty}_{i j t} + β_{2} Z_{i j t}), \end{array}

where Pr(y_ijt = 1) is the probability that individual i in country j surveyed in wave t considers AIDS a salient issue, conditional on the explanatory variables. The main interest is in the coefficient, β₁, of individual-level poverty. Z_ijt denotes a vector of control variables with corresponding coefficients. Given the hierarchical structure of the data, country-level characteristics are allowed to affect country-level AIDS salience by modeling the level 1 constant in the level 2 equation:

α_{j t} = μ + γ_{1} {GDP}_{j t} + γ_{2} C_{j t} + η_{j} .

Here, the constant term, α_jt, is a function of country-level GDP per capita at the time of survey-wave t and a vector of level 2 controls, C_jt, that vary both between countries, j, and over survey waves, t. η_j is a country-specific (time-invariant) random intercept varying around an average, µ, common to all countries. The incorporation of η_j is important because it models unobserved country-specific heterogeneity that is unaccounted for by the explanatory variables in equation (2). Substituting equation (2) into equation (1), and adding survey-wave fixed effects, $γ_{t}$ , we end up with the multilevel model:

\Pr (y_{i j t} = 1) = {logit}^{- 1} (\begin{array}{l} β_{1} P o v e r t y_{i j t} + β_{2} Z_{i j t} \\ + γ_{1} G D P_{j t} + γ_{2} C_{j t} + \\ μ + η_{j} + γ_{t} \end{array}),

where η_j ~ N(0,σ²) is a random intercept drawn from a normal distribution. Equation (3) is basically a logistic random effects model (Gelman and Hill 2007; Rabe-Hesketh and Skrondal 2008), with a fixed part consisting of the regression coefficients and a random part summarized by its variance around the intercept. Moreover, having data from different waves (years) is controlled for, using survey-wave fixed effects, $γ_{t}$ , which pick up the effects of factors that are common to individuals in countries at the different waves of the survey.

Results

Table 2 shows results from logistic multilevel regressions for the probability that individuals consider AIDS a salient issue, corresponding to tests of Hypothesis 1. The three country-level variables—GDP per capita, foreign aid for health, and HIV prevalence—appear in models 3 to 6. Government effectiveness is included in models 3 to 5, but is substituted for the AIDS policy index in model 6. The country-level variance component is shown below the regression coefficients.

Table 2.

Multilevel Estimates of the Salience of AIDS in Africa.

Model	(1)	(2)	(3)	(4)	(5)	(6)
Individual-level regressors
Poverty	−0.63*** (8.96)	−0.56*** (7.93)	−0.30*** (3.92)	−0.29*** (3.69)	−0.38*** (4.64)	−0.37*** (4.49)
Cognitive engagement			0.01 (1.20)	0.02 (1.41)	−0.001 (0.06)	−0.002 (0.14)
Radio news			0.07*** (4.64)	0.06*** (3.67)	0.07*** (4.42)	0.07*** (4.60)
Primary schooling			0.13* (2.02)	0.12 (1.88)	0.07 (1.11)	0.07 (1.08)
Secondary schooling+			0.26*** (4.08)	0.22** (3.25)	0.20** (2.95)	0.19** (2.82)
Religiosity			−0.06 (1.57)	−0.11* (2.45)	−0.05 (1.27)	−0.06 (1.48)
Urban residence			0.11** (3.18)	0.09** (2.61)	0.07* (1.97)	0.07 (1.92)
Gender (female = 1)			0.10** (3.29)	0.10** (2.87)	0.11** (3.24)	0.12 (3.42)
Age			−0.01*** (6.26)	−0.01*** (6.72)	−0.01*** (6.21)	−0.01*** (5.97)
Catholic				0.27*** (4.55)
Muslim				0.07 (1.42)
Other religion				0.07 (1.40)
AIDS experience					0.43*** (11.57)	0.43*** (11.64)
Country-level regressors
GDP per capita			0.11 (1.68)	0.41*** (5.76)	0.11 (1.64)	0.07 (1.12)
HIV prevalence			0.08*** (6.30)	0.05** (2.72)	0.08*** (5.80)	0.07*** (5.34)
Foreign aid for health			−0.01*** (4.31)	−0.01*** (5.32)	−0.01*** (4.68)	−0.01*** (4.66)
Government effectiveness			−0.45* (2.53)	−0.85*** (3.72)	−0.31 (1.68)
AIDS Policy Index						0.01 (0.28)
Second wave		1.28*** (24.71)	1.20*** (18.75)		1.17*** (17.16)	1.21*** (18.07)
Third wave		0.95*** (18.39)	0.91*** (11.71)	−0.40*** (8.32)	0.87*** (10.74)	0.92*** (12.04)
Fourth wave		0.41*** (7.48)	0.40*** (3.81)	−1.08*** (13.40)	0.38*** (3.50)	0.48*** (4.84)
Intercept	−3.49*** (11.03)	−4.27*** (13.60)	−5.55*** (20.47)	−4.67*** (16.30)	−5.47*** (20.70)	−5.60 (4.31)
Random effects
Country-level variance component (σ_η)	1.30***	1.37***	0.91***	0.84***	0.83***	0.80***
Log likelihood	−17,046.9	−16,596.7	−15,652.7	−13,870.5	−13,777.474	−13,500.9
Pseudo R²	.10	.12	.17	.27	.27	.26
Individuals	83,579	83,579	79,443	67,478	72,556	68,508
Countries	20	20	20	20	20	18
Waves	4	4	4	3	4	4

Coefficients are log odds from multilevel logistic models (absolute value of z-statistics in parentheses), obtained using xtmelogit in Stata 12. The first wave is the reference group for the survey-wave fixed effects. In model 4, data on religious affiliations are not available for the first wave. In model 6, data for the AIDS Policy Index are not available for Cape Verde and Liberia. The pseudo R² is the improvement in the log-likelihood (LL) of the full model compared with the null model without explanatory variables (LL_null − LL_full) / LL_null.

p < .05. **p < .01. ***p < .001.

The results in Table 2 are quite clear and show a robust and significantly negative effect of poverty on the salience of AIDS. Model 1 shows the result from a simple regression with poverty as the only explanatory variable. This shows that poverty has a statistically significant effect (p < .001) on the probability that AIDS is considered a salient political issue. As hypothesized, high values on the poverty index—corresponding to a state of poverty with lack of basic material necessities on a regular basis—do seem to decrease the likelihood that individuals consider AIDS salient. Similarly, wealthier people (low values on the index) tend to find AIDS more salient. Model 2 adds survey-wave fixed effects to the regressions, but this does little to change to impact of the poverty variable. Model 3 includes eight individual-level and three country-level control variables, and model 4 adds controls for individuals’ religious affiliation, meaning that data from the first wave are excluded. While this decreases the impact and significance of poverty, its effect is still substantial and highly significant (p < .001). Finally, models 5 and 6 control for respondents’ personal experience with having friends or relatives who have died of AIDS. However, this does not change the result for the poverty variable. Nor does including the AIDS Policy Index in model 6. These results strongly support the hypothesis that poverty affects AIDS salience in emerging democracies in Africa.

However, when we turn to the impact of country-level wealth, the story is not quite as clear. As expected, a higher level of GDP has a positive effect on the likelihood that individuals attach importance to AIDS across (and within) countries. However, the significance of country-level GDP varies largely. In fact, GDP is statistically significant at conventional levels only in model 4, when data from the first wave are excluded. This suggests that differences in country-level wealth do not necessarily have a large impact on the importance individuals attach to the AIDS issue but rather that the material basis of AIDS salience is more firmly rooted in individual-level poverty.¹¹

While the individual-level poverty index remains highly significant in all models, the magnitude of the coefficient drops (in absolute terms) when controls are added. Based on Model 5, a one unit increase in the poverty index—corresponding to a change from the lowest to the highest value—decreases the estimated log odds of AIDS salience by −0.38, which is similar to a multiplicative effect on the odds ratio of the magnitude e^−0.38 = 0.68 (Gelman and Hill 2007, 82–83; Rabe-Hesketh and Skrondal 2008, 249). In comparison with wealthy people, being very poor therefore decreases the odds of finding AIDS salient by a factor of 0.68. Similarly, the odds of AIDS being salient for well-off people is around 1.5 (1/0.68) times higher than for poor people.

To see how the likelihood of AIDS being salient varies with poverty, Figure 2 plots the predicted probabilities of considering AIDS salient based on model 5 in Table 2.

Figure 2.

Predicted probabilities of HIV/AIDS being salient by poverty and national context.

Figure 2 shows the effect of poverty with all other individual-level variables held constant, assuming the respondent is female, religious, thirty-six years old, lives in an urban area, has attended secondary school, knows someone who has died of AIDS, and receives radio news and has a level of cognitive political engagement corresponding to the sample average.

The effect of variations in poverty for this individual is shown for two different national contexts at the third (2006/2006) and fourth (2008/2009) wave of the Afrobarometer, respectively, allowing for different country random effects. The top two lines represent cases where the country-level characteristics are set equal to the values of South Africa during the third (punctuated line) and fourth wave (bold line). The bottom two lines show predicted probabilities for an individual living in a country with country-level variables matching Lesotho’s during the third (punctuated line) and fourth (dotted line) wave of the survey. South Africa and Lesotho both have very high levels of HIV prevalence—18 and 24 percent of the adult population, respectively—but differ markedly in their level of GDP per capita, with South Africa being more than five times as wealthy as Lesotho. However, it is important to note that the two countries are also noticeably different on other country-level variables. For instance, in terms of government effectiveness, South Africa scores markedly higher than Lesotho, while Lesotho receives much more foreign aid for health than South Africa.¹² Therefore, the two lines in Figure 2 do not represent a comparison with all other country-level variables held constant.

Across national contexts, the plot shows that the probability of considering AIDS an important issue decreases as poverty increases. In the South African case, the predicted probability of AIDS being salient drops by 8 (third wave) and 7 (fourth wave) percentage points as the poverty index increases from the lowest (well-off) to the highest (very poor) value. However, it is also clear that the overall salience of AIDS has decreased substantially in the period from the third (2005/2006) to the fourth (2008/2009) wave of the Afrobarometer. Specifically, during the third wave, an individual who is poor and experiences a frequent lack of basic material necessities had a 26 percent probability of finding AIDS salient, while the corresponding probability is 34 percent for a similar individual who is well-off. During the fourth wave, these probabilities have dropped to 19 and 26 for poor and well-off individuals, respectively.

In the case of Lesotho, the salience of AIDS dropped from an already low level in 2005 to an even lower level in 2008. Moreover, the probabilities for poor people decrease by only 2 (fourth wave) and 3 (third wave) percentage points compared with well-off people. This illustrates that the likelihood that a person in Lesotho considers AIDS salient is much lower overall than in South Africa for all levels of poverty. This is also evident from the prediction lines, which clearly demonstrate the strong impact of the country-level variables on national levels of AIDS salience. Overall, the results in Table 2 clearly support Hypothesis 1. However, the results also show that it is mainly microlevel poverty that affects the salience of AIDS as a political issue. Variation in individual-level poverty, therefore, appears to be an important factor for explaining differences in the salience of AIDS in Africa.

The results for the other explanatory variables in Table 2 also shed light on some important reasons for variations in AIDS salience. For the individual-level regressors, cognitive political engagement is statistically insignificant, while receiving information from radio news significantly increases the probability that AIDS is salient. Secondary schooling also generally increases AIDS salience. In contrast, religiosity and religious affiliation seem to be rather weakly related to AIDS salience, with the possible exception of Catholics who tend to find AIDS a more important political issue. This suggests that material poverty is more important than religious beliefs for the salience of AIDS. Consistent with the findings of Dionne (2012), having personally known someone who has died of AIDS also leads to a significant increase in the likelihood that AIDS is salient. Finally, in terms of the demographic controls, urban residents, women, and younger people find the AIDS issue more important than people living in rural areas, men, and older people, respectively. These results fit nicely with the fact that HIV prevalence is higher in urban areas, and among women and younger people (Barnett and Whiteside 2006, 10).

Among the country-level variables, HIV prevalence has a positive and significant impact on national levels of AIDS salience, which shows that AIDS is generally a more salient issue in countries with high levels of HIV prevalence. While government effectiveness seems to decrease AIDS salience, it is perhaps more surprising that AIDS policy does not have systematic effects at the country level. This does not mean that political leaders cannot shape voter attitudes to HIV/AIDS. Rather, attempts at doing so seem to have differing effects that have not—so far—produced systematic cross-country differences in the salience voters attach to AIDS in Africa. Interestingly, foreign aid for health tends to decrease the national-level salience of AIDS. However, it is difficult to know precisely how to interpret this correlation. On the one hand, it is possible that foreign health aid contributes to effective government responses to HIV/AIDS—and that people for that reason tend to think of AIDS as being a less important problem that is already being addressed. On the other hand, foreign aid may also divert accountability for HIV/AIDS policies away from national governments and toward international organizations and NGOs, which may imply that people no longer perceive the government as being responsible for dealing with HIV/AIDS (cf. Swidler and Watkins 2009). In either case, it seems that foreign health aid contributes to push HIV/AIDS off the public agenda in Africa’s emerging democracies. Overall, while these results show that many factors are important and must be accounted for when we try to understand the salience of AIDS, they also suggest that poverty is an important part of the explanation of the status of AIDS as a political issue in Africa.

Testing Generic Implications: The Salience of Unemployment, Poverty, and Hunger

While the main topic of this paper is the relationship between poverty and AIDS salience, the theoretical argument has generic implications. Specifically, issues such as poverty, food shortage, and unemployment are likely to be important for poor groups because they compel people to worry about policies that increase their short-term consumption opportunities, as stated in Hypothesis 2.

To test this hypothesis, Table 3 shows results from multilevel logistic regressions, where the dependent variables are the probabilities that unemployment (models 1 to 2), poverty (models 3 to 4), and food shortage (models 5 to 6) are considered salient political issues. Table 3 employs the battery of regressors used above, except “AIDS experience” at the individual level and “AIDS policy” at the country level as it is unclear why these should affect the salience of unemployment, poverty, and food shortage. To save space, Table 3 only reports results for individual-level poverty and country-level GDP per capita. The full table with results for the controls is available in the online appendix.

Table 3.

Multilevel Estimates of the Salience of Unemployment, Poverty, and Food Shortage in Africa.

Model	(1)	(2)	(3)	(4)	(5)	(6)
Dependent variable	Unemployment	Unemployment	Poverty	Poverty	Food shortage	Food shortage
Individual-level regressors
Poverty	−0.15*** (3.93)	−0.19*** (4.61)	0.20*** (4.89)	0.22*** (5.20)	0.78*** (17.24)	0.78*** (16.68)
Individual-level controls?	Yes	Yes	Yes	Yes	Yes	Yes
Control for religious affiliation?	Yes	No	Yes	No	Yes	No
Country-level regressors
GDP per capita	−0.12** (2.97)	−0.21*** (3.84)	0.08* (2.24)	−0.11** (2.65)	−0.12* (2.05)	−0.32*** (4.91)
Country-level controls?	Yes	Yes	Yes	Yes	Yes	Yes
Survey-wave fixed effects?	Yes	Yes	Yes	Yes	Yes	Yes
Intercept	−0.61** (2.49)	0.51 (1.50)	−2.56*** (18.51)	−0.81*** (5.18)	−3.07*** (13.86)	−1.80*** (7.00)
Random effects
Country-level variance component (σ_η)	1.00***	1.37***	0.42***	0.46***	0.75***	0.87***
Observations	79,443	67,478	79,443	67,478	79,443	67,478
Countries	20	20	20	20	20	20
Waves	4	3	4	3	4	3

Unemployment, poverty, and food shortage are binary indicators of responses to the “most important problems” question (see text for details). Full table including results for the controls is available in the online appendix at http://prq.sagepub.com/supplemental/. Coefficients are log(odds) from multilevel logistic models (z-statistics in parentheses), obtained using xtmelogit in Stata 12. For other details, see note to Table 2.

p < .05. **p < .01. ***p < .001.

The results in Table 3 are generally consistent with Hypothesis 2, with one important exception. In models 1 and 2, poverty is negatively related to unemployment salience in the sense that people who experience a frequent lack of basic necessities are less likely to find unemployment salient compared with well-off people. Moreover, unemployment salience tends to decrease with GDP. This means that people in poorer countries, on average, tend to find unemployment more salient. However, within poorer countries, it is not the very poorest, but the relatively well-off who are most concerned with unemployment.

While this result is not as expected, part of the explanation may be found by inspecting the findings in models 3 to 6. Substituting unemployment salience with the salience of poverty and food shortage shows that individual-level poverty has negative and statistically significant effects (p < .001). That is, poor people are more likely to consider poverty and food shortage salient compared with people who are better off. Indeed, the large coefficient of the poverty index in models 5 and 6 suggests that food shortage is a particularly salient issue among poor groups.

The fact that the salience of poverty and food shortage is higher for poor groups may partly explain why unemployment is less salient for these groups. Precisely because the poor worry about destitution and food shortage, formal employment may not be an issue that—in relative terms—tops the list of important problems. This suggests that there is a “poverty threshold” below which unemployment is not a salient issue. This relationship may be strengthened in areas where formal labor markets are practically absent (cf. Swidler and Watkins 2009). For poor people operating outside the formal economy, hunger and poverty will always be more salient issues than formal employment, which arguably makes little difference to their subsistence. Because experiencing poverty mainly affects the likelihood that poverty and food shortage are salient political issues, Hypothesis 2 receives only conditional support. In terms of country-level wealth, GDP per capita has a strongly negative effect on the salience of food shortage in models 5 and 6, while its impact is unclear in models 3 to 4, which is arguably because economic inequality is still very high and poverty widespread even in countries such as Botswana and South Africa that have the highest average levels of GDP per capita among the Afrobarometer countries.

To illustrate the impact of poverty on the salience of food shortage, Figure 3 plots predicted probabilities based on model 5. The probabilities are shown for countries similar to South Africa (bold line) and Lesotho (punctuated line) on the level 2 variables during the fourth wave.¹³

Figure 3.

Predicted probabilities of food shortage being salient by poverty and national context.

Regardless of national context, the lines show that the likelihood that food shortage is salient increases with poverty. GDP per capita has a negative effect on the salience of food shortage, which is also evident in the comparison of Lesotho and South Africa. Specifically, a poor individual living in Lesotho has a 38 percent probability of considering food shortage salient. The probability for a similar individual who never experiences poverty is 22 percent, corresponding to a 16 percentage point difference caused by poverty alone. In other words, a poor individual in Lesotho is almost twice as likely to consider hunger salient compared with someone who is not poor. While the relationship is similar in South Africa, the probabilities are much lower across all values of poverty.

Conclusion

HIV/AIDS in Africa has transformed from mainly being a health issue to being a major issue of human and economic development with devastating effects on the lives of millions of Africans. Nonetheless, citizens in Africa’s emerging democracies do not generally rank HIV/AIDS among the most important problems requiring government action. This paper has tried to shed new light on why AIDS is a low-salience issue for many Africans, and how we can explain variations in AIDS salience across individuals and national contexts. The explanation has emphasized that the salience of different policy issues is to a large extent related to poverty. Using survey data from the Afrobarometer, the empirical evidence clearly shows that poverty has significant effects on AIDS salience, particularly at the microlevel. Poor groups are significantly less likely to consider AIDS an important political issue compared with wealthier groups. A consequence of the widespread poverty in the region therefore seems to be that people are relatively more concerned about deprivation and issues such as food shortage than about HIV/AIDS.

A number of implications follow in the wake of these findings. First, the fight against poverty and the fight against HIV/AIDS are related. Lifting people out of poverty not only improves their living conditions, it also means that AIDS is likely to become a more salient political issue and attain a more prominent position on the public agenda. Second, if AIDS is not salient to voters, it is less likely to attract attention from political representatives and governments. If so, government responses to HIV/AIDS may not be very effective. However, the HIV/AIDS issue has attracted some attention during election campaigns in a few of Africa’s poorer countries (Patterson 2006, 66–68). This suggests that even when circumstances are unfavorable, there are conditions under which HIV/AIDS figures more highly on the agenda during elections in Africa. While we know little about what these conditions are, it is a question that clearly deserves further attention. Third, the theory and findings in this paper have generic implications for the wider literature on policy salience. For instance, the salience of other issues such as education and health policies may also be rooted in variations in income and poverty. How voters prioritize such issues may therefore also be affected by the distribution of costs and benefits over time. Moreover, evidence from field experiments on voter education campaigns (Vicente 2014) and public deliberation programs (Fujiwara and Wantchekon 2013) in Africa show that such programs can be effective in reducing the role of vote-buying and clientelist policies during elections. This suggests that an avenue for future research on attitudes to AIDS would be to use field experiments to examine whether voter education and information programs raise awareness to the AIDS issue and contribute to offset the effects of poverty.

Footnotes

Acknowledgements

I am grateful for constructive comments from three anonymous reviewers and the editors of Political Research Quarterly (PRQ).

Author’s Note

Replication data for this article can be viewed at . An earlier version of this paper was presented at the 2011 meeting of the American Political Science Association. Any remaining errors are mine alone.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Afrobarometer. 2006. “The Public Agenda: Change and Stability in South Africans’ Rating of National Priorities.” Afrobaro-meter Briefing Paper No. 45. www.afrobarometer.org.

Afrobarometer. 2007. “Round 4 Survey Manual.” www.afro-barometer.org.

Azariadis

Costas

. 2006. “The Theory of Poverty Traps: What Have We Learned?” In Poverty Traps, edited by Bowles

Samuel

Durlauf

Steven N.

Hoff

Karla

, 17–40. Princeton: Princeton University Press.

Barnett

Tony

Whiteside

Alan

. 2006. AIDS in the Twenty-First Century: Disease and Globalization. Basingstoke: Palgrave Macmillan.

Bratton

Michael

. 2008. “Poor People and Democratic Citizenship in Africa.” In Poverty, Participation, and Democracy: A Global Perspective, edited by Krishna

Anirudh

, 28–64. Cambridge: Cambridge University Press.

Bratton

Michael

Mattes

Robert

Gyimah-Boadi

2005. Public Opinion, Democracy, and Market Reform in Africa. Cambridge: Cambridge University Press.

Chakraborty

Shankha

. 2004. “Endogenous Life Time and Economic Growth.” Journal of Economic Theory 116:119–37.

Converse

Philip E.

1975. “Public Opinion and Voting Behavior.” In Handbook of Political Science, edited by Greenstein

Fred I.

Polsby

Nelson W.

, 75–170. Reading: Addison-Wesley.

de Waal

Alex

. 2006. AIDS and Power: Why There Is No Political Crisis—Yet. London: Zed Books.

10.

Dionne

Kim Yi

. 2011. “The Role of Executive Time Horizons in State Responses to AIDS in Africa.” Comparative Political Studies 44 (1): 55–77.

11.

Dionne

Kim Yi

. 2012. “Local Demand for Global Intervention: Policy Priorities in the Time of AIDS.” World Development 40 (12): 2468–77.

12.

Druckman

James N.

Lupia

Arthur

. 2000. “Preference Formation.” Annual Review of Political Science 3:1–24.

13.

Eifert

Benn

Miguel

Edward

Posner

Daniel N.

2010. “Political Competition and Ethnic Identification in Africa.” American Journal of Political Science 54 (2): 494–510.

14.

Fujiwara

Thomas

Wantchekon

Leonard

. 2013. “Can Informed Public Deliberation Overcome Clientelism? Experimental Evidence from Benin.” American Economic Journal: Applied Economics 5 (4): 241–55.

15.

Gelman

Andrew

Hill

Jennifer

. 2007. Data Analysis Using Regression and Multilevel/Hierarchical Models. New York: Cambridge University Press.

16.

Heston

Alan

Summers

Robert

Aten

Bettina

. 2012. “Penn World Tables Version 7.1.” Center for International Comparisons of Production, Income and Prices, University of Pennsylvania.

17.

Hobolt

Sara

Klemmensen

Robert

. 2008. “Government Responsiveness and Political Competition in Comparative Perspective.” Comparative Political Studies 41 (3): 309–37.

18.

Hyden

Goran

. 2006. African Politics in Comparative Perspective. New York: Cambridge University Press.

19.

Institute for Health Metrics and Evaluation. 2010. Financing Global Health 2010. Seattle: University of Washington.

20.

Iqbal

Zaryab

Zorn

Christopher

. 2010. “Violent Conflict and the Spread of HIV/AIDS in Africa.” Journal of Politics 72 (1): 149–62.

21.

Jensen

Peter S.

Justesen

Mogens K.

2014. “Poverty and Vote Buying: Survey-Based Evidence from Africa.” Electoral Studies 33 (1): 220–32.

22.

Joint United Nations Programme on HIV/AIDS. 2008. Report on the Global AIDS Epidemic. Geneva: Joint United Nations Programme on HIV/AIDS.

23.

Joint United Nations Programme on HIV/AIDS. 2013. Report on the Global AIDS Epidemic. Geneva: Joint United Nations Programme on HIV/AIDS.

24.

Justesen

Mogens K.

2012. “Democracy, Dictatorship, and Disease: Political Regimes and HIV/AIDS.” European Journal of Political Economy 28:373–89.

25.

Justesen

Mogens K.

Bjørnskov

Christian

. 2014. “Exploiting the Poor: Corruption and Poverty in Africa.” World Development 58 (June): 106–15.

26.

Kaufmann

Daniel

Kraay

Aart

Mastruzzi

Massimo

. 2010. “The Worldwide Governance Indicators: A Summary of Methodology, Data and Analytical Issues.” World Bank Policy Research Working Paper No. 5430. The World Bank, Washington, DC.

27.

Lieberman

Evan. S.

2007. “Ethnic Politics, Risk, and Policy-Making: A Cross-National Statistical Analysis of Government Responses to HIV/AIDS.” Comparative Political Studies 40 (12): 1407–32.

28.

Mattes

Robert

. 2007. “Public Opinion Research in Emerging Democracies.” In The SAGE Handbook of Public Opinion Research, edited by Donsbach

Wolfgang

Traugott

Michael W.

, 113–20. London: SAGE.

29.

Mattes

Robert

Bratton

Michael

Davids

Yul Derek

. 2003. “Poverty, Survival, and Democracy in Southern Africa.” Afrobarometer Working Paper No. 23. www.afrobaro-meter.org.

30.

Patterson

Amy

. 2006. The Politics of AIDS in Africa. Boulder: Lynne Rienner.

31.

Przeworski

Adam

Limongi

Fernando

. 1997. “Modernization: Theories and Facts.” World Politics 49 (2): 155–83.

32.

Rabe-Hesketh

Sophia

Skrondal

Anders

. 2008. Multilevel and Longitudinal Modeling Using Stata. 2nd ed. College Station: Stata Press.

33.

Smith

Daniel J.

2004. “Youth, Sin and Sex in Nigeria: Christianity and HIV/AIDS-Related Beliefs and Behavior among Rural-Urban Migrants.” Culture, Health & Sexuality 6 (5): 425–37.

34.

Soroka

Stuart N.

Wlezien

Christopher

. 2010. Degrees of Democracy: Politics, Public Opinion, and Policy. Cambridge: Cambridge University Press.

35.

Steenbergen

Marco R.

Jones

Bradford S.

2002. “Modeling Multilevel Data Structures.” American Journal of Political Science 46 (1): 218–37.

36.

Swidler

Ann

Watkins

Susan Cotts

. 2009. “‘Teach a Man to Fish’: The Sustainability Doctrine and Its Social Consequences.” World Development 37 (7): 1182–96.

37.

United Nations. 2013. The Millennium Development Goals Report 2013. New York: United Nations.

38.

van de Walle

Nicholas

. 2001. African Economies and the Politics of Permanent Crisis, 1979–1999. Cambridge: Cambridge University Press.

39.

Vicente

Pedro

. 2014. “Is Vote-Buying Effective? Evidence from a Field Experiment in West Africa.” Economic Journal 124 (February): 356–87.

40.

Whiteside

Alan

Mattes

Robert

Willan

Samantha

Manning

Ryann

. 2002. “Examining HIV/AIDS in Southern Africa through the Eyes of Ordinary Southern Africans.” Afrobarometer Working Paper No. 21. www.afrobaro-meter.org.

41.

Wlezien

Christopher

. 2005. “On the Salience of Political Issues: The Problems with ‘the Most Important Problem.’” Electoral Studies 24 (4): 555–79.

42.

Youde

Jeremy

. 2012. “Public Opinion and Support for Government AIDS Policies in Sub-Saharan Africa.” Social Science & Medicine 74 (1): 52–57.

43.

Zaller

John

Feldman

Stanley

. 1992. “A Simple Theory of Survey Response: Answering Questions versus Revealing Preferences.” American Journal of Political Science 36 (3): 579–616.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

1.71 MB

0.51 MB

0.00 MB

Too Poor to Care? The Salience of AIDS in Africa

Abstract

Keywords

Linking Poverty to Policy Salience

The Economics of AIDS Salience

Pushing AIDS off the Public Agenda

Data and Method 5

Dependent Variable

Explanatory Variable: Poverty

Control Variables

Estimation Method: The Multilevel Model

Results

Testing Generic Implications: The Salience of Unemployment, Poverty, and Hunger

Conclusion

Footnotes

Acknowledgements

Author’s Note

Declaration of Conflicting Interests

Funding

Notes

References

Supplementary Material

Data and Method⁵